diff --git a/.claude/CLAUDE.md b/.claude/CLAUDE.md
deleted file mode 100644
index 5377b0b..0000000
--- a/.claude/CLAUDE.md
+++ /dev/null
@@ -1,83 +0,0 @@
-# GitHub Copilot Instructions
-
-## Project Philosophy
-
-Cratis builds tools for event-sourced systems with a focus on **ease of use**, **productivity**, and **maintainability**. Every rule in these instructions serves one or more of these core values:
-
-- **Lovable APIs** — APIs should be pleasant to use. Provide sane defaults, make them flexible, extensible, and overridable. If an API feels awkward, it is wrong.
-- **Easy to do things right, hard to do things wrong** — Convention over configuration. Artifact discovery by naming. Minimal boilerplate. The framework should guide developers into the pit of success.
-- **Events are facts** — Immutable records of things that happened. Never nullable, never ambiguous, never multipurpose. If you find yourself adding a nullable property to an event, you need a second event.
-- **High cohesion through vertical slices** — Everything for a behavior lives together: backend, frontend, specs. Navigate by feature, not by technical layer. A developer working on "author registration" should never need to jump between `Commands/`, `Handlers/`, and `Events/` folders.
-- **Full-stack type safety** — Shared models flow from C# through proxy generation to TypeScript. End-to-end typing without manual synchronization.
-- **Specialization over reuse** — Build focused, purpose-specific projections and read models rather than reusing one model across conflicting scenarios. Dedicated models are easier to maintain, perform better, and never break unrelated features.
-- **Consistency is king** — When in doubt, follow the established pattern. Consistency across the codebase trumps local optimization. A slightly less elegant solution that matches the rest of the codebase is better than a clever one that stands out.
-
-When these instructions don't explicitly cover a situation, apply these values to make a judgment call.
-
-## General
-
-- **Always use American English spelling** in all code, comments, documentation, and XML docs — no exceptions.
-  - `-ize` not `-ise`: initialize, serialize, customize, normalize, organize, authorize, specialize, centralize, utilize
-  - `-or` not `-our`: behavior, color, favor, honor, humor, neighbor, flavor
-  - `-ization` not `-isation`: initialization, serialization, customization, normalization, organization, authorization
-  - `-er` not `-re`: center, fiber, meter
-  - `-og` not `-ogue`: dialog, catalog, analog
-  - `-ling` not `-lling`: modeling, signaling, labeling, canceling
-  - `-ense` not `-ence`: license, defense, offense
-  - `-ment` not `-ement`: judgment, acknowledgment
-  - Other: gray (not grey), program (not programme), fulfill (not fulfil), enroll (not enrol)
-  - When in doubt, use the US spelling — check a US dictionary.
-- Write clear and concise comments for each function.
-- Make only high confidence suggestions when reviewing code changes.
-- Never change global.json unless explicitly asked to.
-- Never change package.json or package-lock.json files unless explicitly asked to.
-- Never change NuGet.config files unless explicitly asked to.
-- Always ensure that the code compiles without warnings.
-- Always treat warnings as errors and fix them before considering the work complete.
-- Always ensure that the code passes all tests.
-- Always ensure that the code adheres to the project's coding standards.
-- Always ensure that the code is maintainable.
-- For PR descriptions, use short release-note bullets that focus on **user-facing impact only** — new APIs, changed behavior, fixed bugs. Do not include internal implementation details (storage changes, converter updates, gRPC internals, spec additions). Append the **actual** issue number only when the PR is associated with a real GitHub issue (for example `(#351)`). If there is no associated issue, omit the reference entirely. Never use placeholder text like `(#issue)`, never leave the literal example `(#123)`, and never invent a random issue number. Never include Copilot "Original prompt" blocks. **Always verify the issue number using the `search_issues` or `list_issues` GitHub MCP tool — never guess or invent a number.**
-- Always reuse the active terminal for commands.
-- Do not create new terminals unless current one is busy or fails.
-- When asked to commit, push, create a PR, ship, or land changes, always use the **ship-changes** skill.
-
-## Development Workflow
-
-- After creating each new file, run `dotnet build` (C#) or `yarn compile` (TypeScript) immediately before proceeding to the next file. Fix all errors as they appear — never accumulate technical debt.
-- Before adding parameters to interfaces or function signatures, review all usages to ensure the new parameter is needed at every call site.
-- When modifying imports, audit all occurrences — verify additions are used and removals don't break other files.
-- Before concluding any task, run the relevant specs/tests for every affected project and do not stop until they pass.
-- At the end of every task, from repository root run `dotnet clean` and then `dotnet build -c Release`. The task is not complete until build output is zero warnings and zero errors.
-- **After pushing changes to a PR**, use the GitHub MCP tools (`pull_request_read` with `get_check_runs`, `get_job_logs`) to monitor CI check results. If any checks fail, investigate the logs, fix the failures, and push again. The task is not complete until all CI checks pass or the remaining failures are confirmed to be pre-existing flaky tests unrelated to the PR changes.
-
-## Detailed Guides
-
-These guides contain the full rules, examples, and rationale for each topic. The sections above are the global defaults; the guides go deeper into each area:
-   - [Code Quality](./instructions/code-quality.instructions.md)
-   - [Code Quality — C#](./instructions/code-quality.csharp.instructions.md)
-   - [Code Quality — TypeScript](./instructions/code-quality.typescript.instructions.md)
-   - [C# Conventions](./instructions/csharp.instructions.md)
-   - [How to Write Specs](./instructions/specs.instructions.md)
-   - [How to Write C# Specs](./instructions/specs.csharp.instructions.md)
-   - [How to Write TypeScript Specs](./instructions/specs.typescript.instructions.md)
-   - [Entity Framework Core](./instructions/efcore.instructions.md)
-   - [Entity Framework Core Specs](./instructions/efcore.specs.instructions.md)
-   - [Concepts (ConceptAs)](./instructions/concepts.instructions.md)
-   - [Documentation](./instructions/documentation.instructions.md)
-   - [Pull Requests](./instructions/pull-requests.instructions.md)
-   - [Vertical Slices](./instructions/vertical-slices.instructions.md)
-   - [TypeScript Conventions](./instructions/typescript.instructions.md)
-   - [React Components](./instructions/components.instructions.md)
-   - [Dialogs](./instructions/dialogs.instructions.md)
-   - [Reactors](./instructions/reactors.instructions.md)
-   - [Orleans](./instructions/orleans.instructions.md)
-
-## Header
-
-All files should start with the following header:
-
-```csharp
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-```
diff --git a/.claude/CLAUDE.md b/.claude/CLAUDE.md
new file mode 120000
index 0000000..c930a01
--- /dev/null
+++ b/.claude/CLAUDE.md
@@ -0,0 +1 @@
+../.ai/rules/general.md
\ No newline at end of file
diff --git a/.claude/agents b/.claude/agents
new file mode 120000
index 0000000..2f6c069
--- /dev/null
+++ b/.claude/agents
@@ -0,0 +1 @@
+../.ai/agents
\ No newline at end of file
diff --git a/.claude/hooks b/.claude/hooks
new file mode 120000
index 0000000..5e5c5a2
--- /dev/null
+++ b/.claude/hooks
@@ -0,0 +1 @@
+../.ai/hooks
\ No newline at end of file
diff --git a/.claude/prompts b/.claude/prompts
new file mode 120000
index 0000000..d24f01f
--- /dev/null
+++ b/.claude/prompts
@@ -0,0 +1 @@
+../.ai/prompts
\ No newline at end of file
diff --git a/.claude/rules/code-quality.csharp.md b/.claude/rules/code-quality.csharp.md
deleted file mode 100644
index a24b621..0000000
--- a/.claude/rules/code-quality.csharp.md
+++ /dev/null
@@ -1,89 +0,0 @@
----
-applyTo: "**/*.cs"
----
-
-# Code Quality — C#
-
-C#-specific applications of the general [Code Quality](./code-quality.instructions.md) principles.
-
-## Composition over Inheritance
-
-Use constructor injection to compose behavior. Primary constructors make this natural in modern C# — the type's dependencies are visible at a glance and can be substituted in tests.
-
-```csharp
-// ❌ Inheritance — child is tightly coupled to parent internals
-public class ReportExporter : BaseExporter
-{
-    public override void Export(Report report) { ... }
-}
-
-// ✅ Composition — behavior is injected and interchangeable
-public class ReportExporter(IExportStrategy strategy)
-{
-    public void Export(Report report) => strategy.Execute(report);
-}
-```
-
-**Rules:**
-- Never extend a concrete class to add or change behavior — inject a collaborator instead.
-- Use interfaces and `ConceptAs<T>` record wrappers rather than inheritance chains.
-- Inheritance is acceptable only for framework integration points with a well-defined extension mechanism (e.g. `Specification`, `Migration`, `AggregateRoot`).
-
-## Open/Closed Principle
-
-The framework's `IInstancesOf<T>` mechanism makes the open/closed pattern effortless — adding a new implementation is all it takes to extend behavior. Use it instead of growing `switch` statements.
-
-```csharp
-// ❌ Modified every time a new format is added
-public class ReportFormatter
-{
-    public string Format(Report report, string formatType)
-    {
-        if (formatType == "csv") return FormatAsCsv(report);
-        if (formatType == "json") return FormatAsJson(report);
-        throw new UnknownFormat(formatType);
-    }
-}
-
-// ✅ New formats added by implementing the interface — no existing code changes
-public interface IReportFormatter
-{
-    string Format(Report report);
-}
-
-public class CsvReportFormatter : IReportFormatter { ... }
-public class JsonReportFormatter : IReportFormatter { ... }
-```
-
-**Rules:**
-- Prefer strategy interfaces over `switch`/`if-else` chains that grow over time.
-- Use `IInstancesOf<T>` to discover all implementations by convention — no manual registration needed.
-- Design public APIs as contracts (interfaces/records) rather than concrete implementations.
-
-## Separation of Concerns
-
-The Chronicle + Arc stack has clear layer boundaries. Violating them creates coupling that is hard to undo.
-
-**Rules:**
-- Domain types must not reference EF Core, MongoDB, or HTTP concepts directly.
-- Command handlers express intent in domain terms — they delegate persistence and I/O to injected collaborators.
-- Projections build read models; they must not trigger commands or produce side effects.
-- Reactors handle side effects; they must not directly read or write the event log.
-
-## Low Coupling
-
-**Rules:**
-- Depend on abstractions (interfaces, records), not on concrete implementations.
-- Use constructor injection — it makes dependencies explicit and testable.
-- Avoid reaching through an object to call methods on its dependencies (`a.B.C.Do()` is a sign of tight coupling).
-- Limit constructor dependencies to four or five — more is a signal the type is doing too much.
-- Never reference types from unrelated features directly; go through a shared contract or event instead.
-
-## Cross-Cutting Concerns
-
-**Rules:**
-- Never write logging statements directly inside command handlers, projections, or domain types. Use the `[LoggerMessage]` pattern in a co-located `*Logging.cs` partial class.
-- Never perform authorization checks inside domain logic — express them as attributes or middleware applied at the boundary.
-- Never duplicate error-handling or retry logic across handlers — centralize it in a pipeline or middleware.
-- Use `ICommandPipeline`, middleware, and decorators to apply cross-cutting concerns at the infrastructure layer so that domain code remains unaware of them.
-- When you notice the same infrastructural pattern appearing in two or more places (logging a specific event, catching a specific exception, checking a specific condition), extract it into a shared cross-cutting mechanism rather than duplicating it.
diff --git a/.claude/rules/code-quality.csharp.md b/.claude/rules/code-quality.csharp.md
new file mode 120000
index 0000000..1b8150c
--- /dev/null
+++ b/.claude/rules/code-quality.csharp.md
@@ -0,0 +1 @@
+../../.ai/rules/code-quality.csharp.md
\ No newline at end of file
diff --git a/.claude/rules/code-quality.md b/.claude/rules/code-quality.md
deleted file mode 100644
index c429fe3..0000000
--- a/.claude/rules/code-quality.md
+++ /dev/null
@@ -1,82 +0,0 @@
----
-applyTo: "**/*"
----
-
-# Code Quality
-
-Good code is not just code that works — it is code that can be understood, changed, and extended safely. The principles below are the foundation for writing code that remains maintainable as the system grows. They are not abstract ideals; each one has a concrete, practical consequence for how you write and structure code in this project.
-
-When these principles don't explicitly cover a situation, apply these values to make a judgment call. See the language-specific guides for concrete rules and examples:
-- [Code Quality — C#](./code-quality.csharp.instructions.md)
-- [Code Quality — TypeScript](./code-quality.typescript.instructions.md)
-
-## Composition over Inheritance
-
-Prefer composing behavior from smaller, focused collaborators over building class hierarchies. Inheritance couples the child tightly to the parent's internal structure — a change to the parent can break every subclass. Composition keeps collaborators independent and replaceable.
-
-**Rules:**
-- Never extend a concrete class to add or change behavior — inject a collaborator instead.
-- Inheritance is acceptable only for framework integration points where a base class is part of a well-defined extension mechanism.
-
-## Single Responsibility Principle
-
-Every type and every method should have **one reason to change** — it should do one thing and do it well. A class that fetches data, transforms it, validates it, and sends an email has four reasons to change. When any of those concerns shifts, you have to touch — and risk breaking — all the others.
-
-**Rules:**
-- A class or method that requires a comment explaining what each section does is a sign it should be split.
-- Methods longer than ~20 lines are a signal they are doing too much — extract collaborators or helper methods.
-- If a type needs collaborators from two unrelated domains, question whether it has two responsibilities.
-- Follow the [File Size Guideline](#file-size--200-line-guideline) below.
-
-## Open/Closed Principle
-
-Types should be **open for extension, closed for modification**. Once a type is in use, changing its internals to support new behavior risks breaking existing callers. Instead, design extension points — interfaces, strategies, event hooks — that allow new behavior to be added without touching existing code.
-
-**Rules:**
-- Prefer strategy interfaces over `switch`/`if-else` chains that grow over time.
-- Design public APIs as contracts (interfaces/records) rather than concrete implementations.
-
-## Separation of Concerns
-
-Each layer and each module should own exactly one concern. Mixing concerns — for example, querying the database and formatting the HTTP response in the same method — creates entanglement that makes both concerns harder to change or test independently.
-
-**Rules:**
-- Keep domain logic out of infrastructure — domain types must not reference infrastructure or transport concepts directly.
-- Keep infrastructure out of domain logic — handlers and domain types express intent; they delegate to collaborators for persistence, messaging, and I/O.
-
-## Low Coupling
-
-Coupling is the degree to which one module depends on the internals of another. High coupling means a change in one place forces changes everywhere else. Low coupling means modules can evolve independently.
-
-**Rules:**
-- Depend on abstractions, not on concrete implementations.
-- Avoid reaching through an object to call methods on its dependencies — this is a sign of tight coupling.
-- Limit the number of dependencies a single type takes — more than four or five is a signal it is doing too much.
-- Never reference types from unrelated features directly; go through a shared contract or event instead.
-
-## High Cohesion
-
-Cohesion measures how closely related the responsibilities within a module are. A highly cohesive class has all its methods and properties working together toward a single goal. A low-cohesion class is a collection of unrelated utilities that happen to live in the same file.
-
-**Rules:**
-- Group code by feature, not by technical role — everything for a behavior belongs together.
-- If you find yourself writing methods in a type that use completely different sets of fields or dependencies, the type likely needs to be split.
-- Utilities and helpers are acceptable only when the operations they provide are genuinely shared across features; otherwise, keep logic in the feature that owns it.
-
-## File Size — 200-Line Guideline
-
-A file exceeding **200 lines** is a strong signal that it contains too many responsibilities. This is not a hard limit — some files are legitimately longer — but whenever you find yourself adding to a file that already approaches this size, stop and ask: can this be split?
-
-**Rules:**
-- When a file crosses 200 lines, look for natural split points: a sub-concept that could become its own type, a behavior that could move to a collaborator, or a section that belongs in a different layer.
-- Aim for files that can be understood in a single reading without scrolling.
-- Instruction and documentation files follow the same principle — a guide over 200 lines usually contains multiple distinct topics that deserve their own files.
-
-## Cross-Cutting Concerns
-
-Cross-cutting concerns — logging, validation, authorization, error handling, metrics, caching — affect many parts of the system but belong to none of them. Scattering them through business logic creates noise and duplication. Centralizing them in infrastructure keeps domain code clean.
-
-**Rules:**
-- Never write logging or authorization checks inside domain logic — apply them at the infrastructure boundary.
-- Never duplicate error-handling or retry logic — centralize it in a pipeline, middleware, or decorator.
-- When you notice the same infrastructural pattern appearing in two or more places, extract it into a shared cross-cutting mechanism rather than duplicating it.
diff --git a/.claude/rules/code-quality.md b/.claude/rules/code-quality.md
new file mode 120000
index 0000000..c76eaba
--- /dev/null
+++ b/.claude/rules/code-quality.md
@@ -0,0 +1 @@
+../../.ai/rules/code-quality.md
\ No newline at end of file
diff --git a/.claude/rules/code-quality.typescript.md b/.claude/rules/code-quality.typescript.md
deleted file mode 100644
index 715cb99..0000000
--- a/.claude/rules/code-quality.typescript.md
+++ /dev/null
@@ -1,86 +0,0 @@
----
-applyTo: "**/*.ts,**/*.tsx"
----
-
-# Code Quality — TypeScript
-
-TypeScript/React-specific applications of the general [Code Quality](./code-quality.instructions.md) principles.
-
-## Composition over Inheritance
-
-React is built on composition — components accept children, hooks compose other hooks, and higher-order utilities wrap behavior. Avoid class hierarchies entirely; the language and framework have moved on.
-
-```tsx
-// ❌ Inheritance — fragile, couples component to base class internals
-class SpecialButton extends BaseButton {
-    override render() { ... }
-}
-
-// ✅ Composition — wrap or delegate, keep each piece independent
-export const SpecialButton = ({ onClick, label }: SpecialButtonProps) => (
-    <Button onClick={onClick} className="special">
-        {label}
-    </Button>
-);
-```
-
-**Rules:**
-- Never use class inheritance for React components — compose with props, children, and hooks instead.
-- Extract repeated UI patterns into small, focused components rather than adding conditions to a shared parent.
-- Extract repeated logic into custom hooks — a hook that does two unrelated things should be two hooks.
-
-## Open/Closed Principle
-
-TypeScript discriminated unions and generic constraints let you add new variants without touching existing code. Prefer them over ever-growing `if-else` / `switch` chains.
-
-```ts
-// ❌ Grows every time a new shape is needed
-function area(shape: string, a: number, b?: number): number {
-    if (shape === 'circle') return Math.PI * a * a;
-    if (shape === 'rectangle') return a * (b ?? 0);
-    throw new Error('Unknown shape');
-}
-
-// ✅ New shapes extend the union — existing handler functions are untouched
-type Circle = { kind: 'circle'; radius: number };
-type Rectangle = { kind: 'rectangle'; width: number; height: number };
-type Shape = Circle | Rectangle;
-
-function area(shape: Shape): number {
-    switch (shape.kind) {
-        case 'circle': return Math.PI * shape.radius ** 2;
-        case 'rectangle': return shape.width * shape.height;
-    }
-}
-```
-
-**Rules:**
-- Model variation with discriminated unions rather than optional fields or string literals.
-- Design utility functions to accept an interface or generic constraint so new types can be handled by adding a new implementation, not by editing existing code.
-
-## Separation of Concerns
-
-React components have one job: render UI and delegate events. Keep data-fetching, business logic, and side effects in dedicated hooks or services — not inline in the component body.
-
-**Rules:**
-- Never write data-fetching or business logic directly in a component — extract it into a hook.
-- Component files (`.tsx`) must not import from infrastructure layers such as HTTP clients or storage utilities directly — go through an abstraction or a generated proxy.
-- Keep style concerns in co-located `.css` files; keep data concerns in hooks; keep rendering in the component.
-
-## Low Coupling
-
-Coupling in TypeScript is often hidden in deep import paths. Barrel files and path aliases make coupling explicit and keep refactoring safe.
-
-**Rules:**
-- Import from barrel `index.ts` files, not from deep internal paths — this limits the blast radius of refactoring.
-- Use the configured path aliases (e.g. `Strings`, `Components`) rather than relative `../../../` chains.
-- Never import from an unrelated feature's internal files — go through that feature's public barrel export.
-- Keep the number of imports in a single file reasonable — many imports from many different areas is a coupling smell.
-
-## Cross-Cutting Concerns
-
-**Rules:**
-- Use React Error Boundaries to centralize error display — never scatter `try/catch` blocks inside component render paths.
-- Use a single top-level provider or hook for global state (e.g. authentication, theming) — never drill context down through many component layers.
-- Centralize API error handling in a shared hook or service layer — do not duplicate toast/notification logic per component.
-- Apply logging, analytics, and monitoring at the infrastructure edge (e.g. router callbacks, global error handlers) so that feature components remain unaware of them.
diff --git a/.claude/rules/code-quality.typescript.md b/.claude/rules/code-quality.typescript.md
new file mode 120000
index 0000000..26b440a
--- /dev/null
+++ b/.claude/rules/code-quality.typescript.md
@@ -0,0 +1 @@
+../../.ai/rules/code-quality.typescript.md
\ No newline at end of file
diff --git a/.claude/rules/components.md b/.claude/rules/components.md
deleted file mode 100644
index d1ad855..0000000
--- a/.claude/rules/components.md
+++ /dev/null
@@ -1,121 +0,0 @@
----
-applyTo: "**/*.tsx"
----
-
-# Building React Components
-
-## Composition over Monoliths
-
-A well-built component tree is like a well-organized kitchen — every tool has a place, and you can find what you need without opening every drawer. Large components that do everything are hard to understand, hard to test, and hard to change without breaking something unrelated.
-
-- Split components into small, focused pieces and compose them together. Each component should have a single, clear responsibility.
-- Parent components own state and event handlers; children receive props. This makes data flow predictable and debuggable.
-- If you find yourself writing a block comment like `// Author list section` inside a component, that section should be its own component. The comment is a code smell — the component name should provide that context instead.
-
-## Folder Structure
-
-- Single-file component → place directly in the parent feature folder.
-- Multi-file component (sub-components, hooks, CSS) → create a folder named after the component:
-
-```
-PrototypeWindow/
-  PrototypeWindow.tsx      ← composition root
-  PrototypeWindow.css      ← styles for the composition
-  TitleBar.tsx             ← sub-component
-  CanvasArea.tsx           ← sub-component
-  ResizeHandle.tsx         ← sub-component
-  index.ts                 ← re-exports public API
-```
-
-Add an `index.ts` that re-exports the public surface so import paths stay stable.
-
-## Styling
-
-Consistent styling comes from discipline: static styles in CSS files, dynamic values inline, and colors always from PrimeReact's design tokens. This ensures theming works automatically and no component breaks the visual language.
-
-- Use **CSS classes in co-located `.css` files** for static styles.
-- Each component must have its own CSS file — never add sub-component styles to the parent's CSS. This keeps styles co-located with the component they belong to.
-- The composition root's CSS only contains layout/grid rules for positioning children — it should not style the children themselves.
-- Use inline `style` props **only** for runtime-dynamic values (pixel positions, computed sizes).
-- Use **PrimeReact CSS variables** for all colors, backgrounds, borders. This ensures the application respects theming and dark/light mode switches:
-  - `var(--surface-0)` through `var(--surface-900)`, `var(--surface-card)`, `var(--surface-border)`, `var(--surface-ground)`
-  - `var(--text-color)`, `var(--text-color-secondary)`, `var(--primary-color)`, `var(--primary-color-text)`, `var(--highlight-bg)`
-  - Never hard-code hex or `rgb()` for UI chrome — it will break when themes change. Only hard-code colors that are intentionally theme-independent (e.g. brand-specific accent dots, traffic-light indicators).
-- Name CSS classes with a BEM-like prefix matching the component name.
-
-## Props
-
-Props are a component's public API. They should be clear, minimal, and well-documented.
-
-- Each sub-component declares its own `*Props` interface with JSDoc on every prop.
-- Pass only needed props — avoid threading large prop bags through component trees.
-- Event handlers follow `on*` naming: `onPointerDown`, `onSelect`.
-
-## Dialogs
-
-See [dialogs.instructions.md](./dialogs.instructions.md) for the full dialog guide.
-
-**Summary:** Never import `Dialog` from `primereact/dialog`. Use `CommandDialog` from `@cratis/components/CommandDialog` for command-executing dialogs and `Dialog` from `@cratis/components/Dialogs` for data-collection dialogs. Do not render manual `<Button>` components for dialog actions — the dialog components handle footers.
-
-## Icons
-
-Follow these rules when working with SVG icons:
-
-- **Distinguish icons from status/interactive components.** A pure SVG icon is a simple presentational element. A component that wraps an icon with interactive behavior (e.g. a dropdown, tooltip, or complex state) is a _component_, not an icon — name it accordingly (e.g. `SliceStatus`, not `SliceStatusIcon`).
-- **Store each SVG as a separate `.svg` file** inside the icon's folder. Do not embed SVG markup directly in `.tsx` files.
-  - Import SVG files with the `?raw` suffix to get the raw SVG string: `import iconSvg from './Icon.svg?raw';`
-  - Render inline using `dangerouslySetInnerHTML={{ __html: iconSvg }}` so that CSS `currentColor` is honored.
-- **Use subfolders for grouping related icons or complex components.**
-  - A folder named `SliceStatus/` groups the four status SVG files together with the interactive `SliceStatus` component that uses them.
-  - Simple, standalone icons may live directly in the `icons/` root if they have no related siblings.
-- **Every icon folder must have an `index.ts`** that re-exports the public API, keeping import paths for consumers stable.
-- **Barrel-export all icons through `icons/index.ts`** so consumers import from the `icons` path alias, not from deep paths.
-
-**Example structure:**
-```
-icons/
-  SliceStatus/
-    NotStarted.svg         ← raw SVG file
-    InProgress.svg         ← raw SVG file
-    ReadyForReview.svg     ← raw SVG file
-    Done.svg               ← raw SVG file
-    SliceStatus.tsx        ← interactive component using the SVG files
-    SliceStatus.css
-    index.ts
-  CogWheelIcon/
-    CogWheel.svg           ← raw SVG file
-    CogWheelIcon.tsx       ← thin wrapper component
-    CogWheelIcon.css
-    index.ts
-  WireframeIcon.tsx        ← simple component with no SVG (stays at root)
-  WireframeIcon.css
-  index.ts                 ← re-exports everything
-```
-
-## Storybook
-
-- Storybook runs at **http://localhost:6006** — never restart it.
-- Use the `click` tool to interact with Storybook for visual verification.
-
-## Verification
-
-After every task, run both:
-1. `yarn lint`
-2. `npx tsc -b`
-
-## README.md for Complex Components
-
-Complex components accumulate knowledge that lives nowhere else — why a particular state structure was chosen, how sub-components divide responsibilities, what conventions the CSS follows. Without a README, the next developer (or AI) has to reverse-engineer all of this from the code.
-
-Every component folder with sub-components, hooks, or non-trivial architecture **must** have a `README.md`.
-
-**Before starting work:** Check for an existing README and read it first. It may contain context that changes your approach.
-
-**A README must cover:**
-- Component hierarchy — tree of components and what each owns
-- Architecture decisions — what was chosen and why
-- State management — where state lives, what each piece controls
-- CSS conventions — patterns used across children
-- How to extend — steps for common modifications
-
-**Keep READMEs current** — update in the same commit when changing architecture, layout, or state structure.
diff --git a/.claude/rules/components.md b/.claude/rules/components.md
new file mode 120000
index 0000000..d7e24c0
--- /dev/null
+++ b/.claude/rules/components.md
@@ -0,0 +1 @@
+../../.ai/rules/components.md
\ No newline at end of file
diff --git a/.claude/rules/concepts.md b/.claude/rules/concepts.md
deleted file mode 100644
index 119ff16..0000000
--- a/.claude/rules/concepts.md
+++ /dev/null
@@ -1,68 +0,0 @@
----
-applyTo: "**/*.cs"
----
-
-# Concepts — Strongly-Typed Domain Values
-
-## Why
-
-A `Guid` is a `Guid` is a `Guid` — the compiler can't tell you if you accidentally passed a `UserId` where an `AuthorId` was expected. Both are `Guid`, both compile, and the bug hides until production.
-
-`ConceptAs<T>` wraps every domain value in a named type. The compiler now enforces that an `AuthorId` is not a `UserId`, method signatures become self-documenting, and code review becomes about logic rather than "is this the right ID?"
-
-Never use raw primitives (`string`, `int`, `Guid`) in domain models, commands, events, or queries.
-
-## Rules
-
-- Inherit from `ConceptAs<T>` as a positional `record` — this gives you value equality and immutability for free.
-- `ConceptAs<T>` already provides an implicit conversion **from** `T` **to** the concept.
-- Always add an implicit conversion operator **from** the concept **to** `T` for easy extraction.
-- Add a `static readonly` sentinel value instead of using `null` — sentinels make "no value" explicit and avoid nullable reference type noise:
-  - `Guid` → `NotSet` backed by `Guid.Empty`
-  - `string` → `NotSet` backed by `string.Empty`
-  - `int` / `long` → `NotSet` backed by `0` / `0L`
-- For `Guid`-backed identity concepts, add a `static New()` factory method — it reads better than `new AuthorId(Guid.NewGuid())`.
-- If the concept is used as a Chronicle event source key, add an implicit conversion to `EventSourceId` — this lets Chronicle resolve the key automatically.
-- Place concepts in the **feature folder they belong to** — never create a standalone `Concepts/` folder. Concepts shared between slices go in the feature root; shared between features go in `Features/` root.
-- One concept per file, named after the concept (e.g. `AuthorId.cs`, `AuthorName.cs`).
-
-## Examples
-
-**Guid identity concept (full canonical pattern):**
-
-```csharp
-public record AuthorId(Guid Value) : ConceptAs<Guid>(Value)
-{
-    public static readonly AuthorId NotSet = new(Guid.Empty);
-
-    public static implicit operator Guid(AuthorId id) => id.Value;
-    public static implicit operator AuthorId(Guid value) => new(value);
-    public static implicit operator EventSourceId(AuthorId id) => new(id.Value.ToString());
-
-    public static AuthorId New() => new(Guid.NewGuid());
-}
-```
-
-**String value concept:**
-
-```csharp
-public record AuthorName(string Value) : ConceptAs<string>(Value)
-{
-    public static readonly AuthorName NotSet = new(string.Empty);
-
-    public static implicit operator string(AuthorName name) => name.Value;
-    public static implicit operator AuthorName(string value) => new(value);
-}
-```
-
-**Int value concept:**
-
-```csharp
-public record Age(int Value) : ConceptAs<int>(Value)
-{
-    public static readonly Age NotSet = new(0);
-
-    public static implicit operator int(Age age) => age.Value;
-    public static implicit operator Age(int value) => new(value);
-}
-```
diff --git a/.claude/rules/concepts.md b/.claude/rules/concepts.md
new file mode 120000
index 0000000..0742d14
--- /dev/null
+++ b/.claude/rules/concepts.md
@@ -0,0 +1 @@
+../../.ai/rules/concepts.md
\ No newline at end of file
diff --git a/.claude/rules/csharp.md b/.claude/rules/csharp.md
deleted file mode 100644
index 6660637..0000000
--- a/.claude/rules/csharp.md
+++ /dev/null
@@ -1,180 +0,0 @@
----
-applyTo: "**/*.cs"
-paths:
-  - "**/*.cs"
----
-
-
-# C# Conventions
-
-The goal is minimal ceremony, maximum clarity. Modern C# (13+) gives us records, primary constructors, pattern matching, and file-scoped namespaces — use them everywhere. The less boilerplate in a file, the faster a reader can understand what it *does*.
-
-## Building
-
-- Use `dotnet build` from the command line.
-- Use `dotnet format` to format code.
-- Use `dotnet test` to run tests.
-
-## Formatting
-
-These rules exist so that every file in the codebase reads the same way. When formatting is consistent, code review focuses on logic, not style.
-
-- Apply code-formatting style defined in `.editorconfig`.
-- Use file-scoped namespace declarations — one less level of indentation for the entire file.
-- Use single-line `using` directives, sorted alphabetically.
-- Never qualify a type that is already unambiguously in scope via a `using` directive. When two `using` directives introduce conflicting type names, qualify only the conflicting occurrences using the shortest unambiguous path (e.g. `Concepts.Events.Foo` or `Contracts.Events.Foo`) — do not add `using` aliases for every conflicting type.
-- Insert a blank line before the opening `{` of every code block (`if`, `for`, `foreach`, `try`, `using`, etc.).
-- Ensure the final `return` statement of a method is on its own line.
-- Use pattern matching and switch expressions wherever possible — they are more readable and the compiler verifies exhaustiveness.
-- Use `nameof` instead of string literals — it survives refactoring.
-- Place private class declarations at the bottom of the file — public API first, implementation details last.
-
-## Language — American English Only
-
-All identifiers, comments, XML docs, and string literals must use **American English** spelling. This applies to class names, method names, properties, parameters, variables, namespaces, and documentation. Common mistakes to avoid:
-
-| Wrong (UK) | Correct (US) |
-|---|---|
-| Initialise, Serialise, Normalise | Initialize, Serialize, Normalize |
-| Behaviour, Colour, Favour, Honour | Behavior, Color, Favor, Honor |
-| Organisation, Authorisation | Organization, Authorization |
-| Centre, Fibre | Center, Fiber |
-| Modelling, Signalling, Cancelling | Modeling, Signaling, Canceling |
-| Dialogue, Catalogue | Dialog, Catalog |
-| Licence, Defence | License, Defense |
-| Judgement, Acknowledgement | Judgment, Acknowledgment |
-| Grey | Gray |
-
-## Naming
-
-- PascalCase for type names, method names, and public members.
-- camelCase for private fields and local variables.
-- Prefix private fields with `_` (e.g. `_myField`).
-- Prefix interface names with `I` (e.g. `IMyService`).
-
-## Code Style
-
-Every rule here reduces noise. `var` avoids redundant type repetition. Expression bodies eliminate braces for trivial members. Primary constructors remove the constructor-plus-field ceremony.
-
-- Prefer `var` over explicit types — the right side of the assignment already tells you the type.
-- Use expression-bodied members for simple methods and properties.
-- Favor primary constructors for all types — they eliminate field declarations for injected dependencies.
-- Use string interpolation instead of `string.Format()` or concatenation.
-- Favor collection initializers and object initializers.
-- Use `IEnumerable<T>` for collections that are not modified; never return mutable collections from public APIs.
-- Don't use regions — they hide code instead of organizing it. If a file needs regions, it needs refactoring.
-- Never add postfixes like `Async`, `Impl`, `Service` to class names — they add noise without information.
-- For types with no implementation body, omit the braces (e.g. `public interface IMyInterface;`).
-- Prefer `record` types for immutable data structures (events, commands, read models, concepts) — they give you value equality, immutability, and concise syntax for free.
-
-## Nullable Reference Types
-
-Embrace the type system — it is the first line of defense against null-related bugs. When it says something cannot be null, trust it.
-
-- Use `is null` / `is not null` — never `== null` / `!= null`.
-- Trust the C# null annotations; don't add defensive null checks when the type system guarantees a value.
-- Add `!` operator where nullability warnings occur and you are certain the value is non-null.
-- Use `is not null` checks before dereferencing potentially null values.
-
-## XML Documentation
-
-XML doc comments are the public API's first impression. They must be multiline — never cram `<summary>` onto a single line. Every public type, method, property, and operator must have XML docs.
-
-- Always use **multiline** `<summary>` tags — opening and closing tags on their own lines:
-  ```csharp
-  /// <summary>
-  /// Represents the unique identifier of a project.
-  /// </summary>
-  ```
-- **Never** use single-line summaries:
-  ```csharp
-  // ❌ Wrong
-  /// <summary>Represents the unique identifier of a project.</summary>
-  ```
-- Every method or operator with parameters **must** include `<param name="...">` for each parameter.
-- Every method or operator that returns a value (non-void) **must** include `<returns>`.
-- Every method that throws must document the exception with `<exception cref="...">` tags.
-- Use `<see cref="..."/>` and `<paramref name="..."/>` to cross-reference types and parameters.
-- Keep summaries concise and purposeful — only document when it adds understanding beyond the name itself.
-
-Example:
-
-```csharp
-/// <summary>
-/// Represents an instance of <see cref="ICommandFilters"/>.
-/// </summary>
-/// <param name="filters">The collection of <see cref="ICommandFilter"/> to use for filtering commands.</param>
-[Singleton]
-public class CommandFilters(IInstancesOf<ICommandFilter> filters) : ICommandFilters
-{
-    /// <summary>
-    /// Filters the command execution through all registered command filters.
-    /// </summary>
-    /// <param name="context">The <see cref="CommandContext"/> to filter.</param>
-    /// <returns>A <see cref="CommandResult"/> representing the aggregated filter outcome.</returns>
-    public async Task<CommandResult> OnExecution(CommandContext context)
-    {
-        // ...
-    }
-}
-```
-
-## Exceptions
-
-Every exception type in the codebase should communicate *what went wrong* in domain terms. Built-in types like `InvalidOperationException` tell you nothing about the problem — a custom `AuthorAlreadyRegistered` tells you everything.
-
-- Use exceptions for exceptional situations only — never for control flow.
-- Always create a custom exception type that derives from `Exception`.
-- Never use built-in exception types (`InvalidOperationException`, `ArgumentException`, etc.).
-- Never suffix exception class names with `Exception` — `AuthorNotFound` reads better than `AuthorNotFoundException`.
-- Always provide a meaningful message when throwing.
-- Add XML doc on the exception type starting with "The exception that is thrown when ...".
-
-## Dependency Injection
-
-The framework discovers and wires dependencies by convention. Explicit registration is the exception, not the rule.
-
-- Prefer constructor injection; avoid `IServiceProvider` directly (service locator anti-pattern).
-- For singletons, use the `[Singleton]` attribute — no explicit registration needed.
-- Systems with a convention of `IFoo → Foo` do not need to be registered explicitly.
-- Command/query `Handle()` method parameters are automatically resolved from DI by type.
-
-## Logging
-
-- Use structured logging with named parameters.
-- Use `ILogger<T>` where `T` is the class name.
-- Keep log messages in a separate `<ClassName>Logging.cs` partial static internal class.
-- Use `[LoggerMessage]` attribute (without `eventId`).
-
-## Async
-
-- Use `async`/`await` for asynchronous programming.
-- Use `Task` and `Task<T>` for asynchronous methods.
-
-## Chronicle & Arc — Key API Types
-
-These are the building blocks. Each type has a specific role in the vertical slice architecture — using the right type in the right place means the framework handles discovery, wiring, and proxy generation automatically.
-
-| Type | Purpose |
-|---|---|
-| `ConceptAs<T>` | Strongly-typed domain value wrapper (see concepts instructions) |
-| `[EventType]` | Marks a record as a Chronicle event — **never** pass arguments to the attribute |
-| `[Command]` | Marks a record as a model-bound command — define `Handle()` directly on the record |
-| `[ReadModel]` | Marks a record as a model-bound query — define static query methods on the record |
-| `CommandValidator<T>` | FluentValidation validator for commands |
-| `IProjectionFor<T>` | Fluent projection definition — AutoMap is on by default |
-| `IReducerFor<T>` | Imperative reducer — receives current state, returns new state |
-| `IReactor` | Marker interface for side-effect observers — method dispatch by event type parameter |
-| `IConstraint` | Constraint definition — enforced server-side by Chronicle at append time |
-| `AggregateRoot` | Chronicle aggregate root with `Apply()` and `Commit()` |
-| `ICommandPipeline` | Programmatic command execution from reactors or other code |
-| `EventSourceId` | Primary identity binding — all events for an entity share the same event source ID |
-| `EventContext` | Event metadata: `Occurred`, `SequenceNumber`, `CorrelationId`, `EventSourceId`, etc. |
-| `ISubject<T>` | Observable query return type — enables real-time WebSocket push |
-| `IMongoCollection<T>` | MongoDB collection — use `.Observe()` for reactive queries |
-
-**Key conventions:**
-- Prefer `ConceptAs<T>` over raw primitives in all domain models, commands, events, and queries — prevents accidental mix-ups and makes APIs self-documenting. See [concepts.instructions.md](./concepts.instructions.md) for details.
-- Projections join **events**, never read models — projections rebuild state from the event stream, not from other projections.
-- For fluent projections, AutoMap is on by default — just call `.From<EventType>()` without manually mapping every property.
-- Use model-bound projection attributes (`[FromEvent<T>]`, `[SetFrom<T>]`, etc.) when possible; fall back to `IProjectionFor<T>` for complex cases.
diff --git a/.claude/rules/csharp.md b/.claude/rules/csharp.md
new file mode 120000
index 0000000..6d08578
--- /dev/null
+++ b/.claude/rules/csharp.md
@@ -0,0 +1 @@
+../../.ai/rules/csharp.md
\ No newline at end of file
diff --git a/.claude/rules/dialogs.md b/.claude/rules/dialogs.md
deleted file mode 100644
index 2ee69f7..0000000
--- a/.claude/rules/dialogs.md
+++ /dev/null
@@ -1,200 +0,0 @@
-````instructions
----
-applyTo: "**/*.tsx"
----
-
-# Using Dialogs
-
-The Cratis dialog wrappers handle command execution, validation timing, loading states, and footer buttons consistently. Using PrimeReact's raw `Dialog` bypasses all of this and leads to inconsistent UX.
-
-## Choose the Correct Dialog Type
-
-- If confirm executes a command, use `CommandDialog` from `@cratis/components/CommandDialog`.
-- If no command is executed on confirm, use `Dialog` from `@cratis/components/Dialogs`.
-- **Never** import `Dialog` from `primereact/dialog` directly.
-
-## When Using `CommandDialog`
-
-- Pass the command constructor to `command={}`. `CommandDialog` handles instantiation, execution, and confirm/cancel buttons.
-- Use command form fields (`InputTextField`, `TextAreaField`, etc. from `@cratis/components/CommandForm`) for user-input values.
-- `CommandDialog` automatically disables confirm while the command executes.
-- Any value that must be present for the form to be considered valid (i.e. passes `validateRequiredProperties`) must be supplied via `initialValues`, **not** via `onBeforeExecute`.
-  - `onBeforeExecute` fires only at execution time — the command is already validated before it runs, so values set there never influence `isValid` and the OK/Submit button will remain permanently disabled.
-  - Use `initialValues` for injected context values (e.g. a parent entity id passed as a prop).
-  - Use `onBeforeExecute` only for transformations that should not affect form validity (e.g. generated IDs).
-
-```tsx
-import { DialogProps, DialogResult } from '@cratis/arc.react/dialogs';
-import { CommandDialog } from '@cratis/components/CommandDialog';
-import { InputTextField } from '@cratis/components/CommandForm';
-import { RegisterProject } from './Registration';
-import { Guid } from '@cratis/fundamentals';
-
-export const AddProject = ({ closeDialog }: DialogProps) => {
-    return (
-        <CommandDialog<RegisterProject>
-            command={RegisterProject}
-            title="Add Project"
-            okLabel="Add"
-            cancelLabel="Cancel"
-            onBeforeExecute={(values) => {
-                values.projectId = Guid.create();  // generated, not user input
-                return values;
-            }}>
-            <InputTextField<RegisterProject>
-                value={instance => instance.name}
-                title="Project name"
-                placeholder="My Project"
-            />
-        </CommandDialog>
-    );
-};
-```
-
-To await the result from the parent:
-
-```tsx
-const [AddProjectDialog, showAddProjectDialog] = useDialog(AddProject);
-
-const [result] = await showAddProjectDialog();
-if (result === DialogResult.Ok) {
-    // Dialog confirmed and command executed successfully
-}
-```
-
-## When Using `Dialog`
-
-Use this for dialogs that collect data and return it without executing a command (e.g. confirmation prompts, pure data-entry dialogs). `Dialog` defaults to OK + Cancel buttons. Use `isValid` to control confirm button state, `okLabel`/`cancelLabel` to customise button text.
-
-```tsx
-import { useState } from 'react';
-import { DialogProps, DialogResult } from '@cratis/arc.react/dialogs';
-import { Dialog } from '@cratis/components/Dialogs';
-import { InputText } from 'primereact/inputtext';
-
-export const AddProject = ({ closeDialog }: DialogProps<{ name: string }>) => {
-    const [name, setName] = useState('');
-    const isValid = name.trim().length > 0;
-
-    return (
-        <Dialog
-            title="Add Project"
-            width='32rem'
-            isValid={isValid}
-            onConfirm={() => closeDialog(DialogResult.Ok, { name })}
-            onCancel={() => closeDialog(DialogResult.Cancelled)}
-        >
-            <InputText
-                value={name}
-                onChange={event => setName(event.target.value)}
-                autoFocus
-            />
-        </Dialog>
-    );
-};
-```
-
-## Prefer `DialogButtons` over Custom Button JSX
-
-Use the built-in `DialogButtons` enum instead of rendering manual `<Button>` elements in the `buttons` prop:
-
-```tsx
-import { DialogButtons, DialogResult, useDialogContext } from '@cratis/arc.react/dialogs';
-```
-
-### Available Button Sets
-
-| `buttons` value | Shows |
-|---|---|
-| `DialogButtons.OkCancel` | Ok + Cancel |
-| `DialogButtons.YesNo` | Yes + No |
-| `DialogButtons.YesNoCancel` | Yes + No + Cancel |
-| `DialogButtons.Ok` | Ok only |
-| `null` | No buttons (content-only dialog) |
-
-## Customising Built-in Buttons
-
-Use `okLabel`/`cancelLabel` to rename the buttons, and `isValid` to disable the confirm button:
-
-```tsx
-<Dialog
-    title="Import Orders"
-    visible={true}
-    buttons={DialogButtons.OkCancel}
-    okLabel="Upload"
-    isValid={!!file && !isUploading}
-    onConfirm={handleUpload}
-    onCancel={() => closeDialog(DialogResult.Cancelled)}
->
-```
-
-## Validation Guard — Keep Dialog Open on Failure
-
-When `onConfirm` needs to keep the dialog open (e.g. a command fails), annotate the handler as `Promise<boolean>` and return `false` to block the close. Return `true` to let the Dialog close itself:
-
-```tsx
-const handleConfirm = async (): Promise<boolean> => {
-    const result = await myCommand.execute();
-    if (!result.isSuccess) return false; // dialog stays open
-    return true; // dialog closes
-};
-```
-
-> **TypeScript note:** Always annotate the function as `Promise<boolean>`. Without it TypeScript infers `Promise<false | void>` which does not satisfy the `ConfirmCallback` type.
-
-## Passing Result Data on Confirm
-
-When the dialog must return data to its caller (e.g. a postal code lookup result), use `onClose` and call `closeDialog` manually, returning `false` to prevent the Dialog from calling it a second time:
-
-```tsx
-<Dialog
-    title="Confirm Location"
-    visible={true}
-    buttons={DialogButtons.OkCancel}
-    isValid={isValid}
-    onClose={(result) => {
-        if (result === DialogResult.Ok) {
-            closeDialog(DialogResult.Ok, { postalCode, city, latitude, longitude } as MyResult);
-            return false; // prevent Dialog from calling closeDialog(Ok) again
-        }
-        // Cancelled: return undefined so the Dialog calls closeDialog(Cancelled)
-    }}
->
-```
-
-## Content-only Dialogs (No Action Buttons)
-
-Use `buttons={null}` for dialogs that contain their own internal actions (e.g. a menu + data table) and don't need a confirm/cancel footer:
-
-```tsx
-<Dialog
-    title="Hubs"
-    visible={true}
-    width="50vw"
-    buttons={null}
-    onCancel={() => closeDialog(DialogResult.Cancelled)}
->
-    <Menubar model={menuItems} />
-    <Listing configurationId={configurationId} />
-</Dialog>
-```
-
-## Props Reference
-
-| Prop | Type | Notes |
-|---|---|---|
-| `title` | `string` | Header text (replaces PrimeReact `header`) |
-| `visible` | `boolean` | Controls visibility |
-| `buttons` | `DialogButtons \| ReactNode \| null` | Prefer `DialogButtons` enum; `null` for no footer |
-| `isValid` | `boolean` | Disables the confirm button when `false` |
-| `okLabel` | `string` | Override the Ok/Confirm button label |
-| `cancelLabel` | `string` | Override the Cancel button label |
-| `onConfirm` | `() => boolean \| void \| Promise<boolean> \| Promise<void>` | Called when Ok is clicked; return `false` to keep dialog open, `true` to close |
-| `onCancel` | `() => void \| Promise<void>` | Called when Cancel is clicked |
-| `onClose` | `(result: DialogResult) => boolean \| void \| Promise<...>` | Combined handler for both Ok and Cancel |
-| `width` | `string` | Dialog width (e.g. `'50vw'`) — replaces PrimeReact `style={{ width }}` |
-| `resizable` | `boolean` | Default `false` |
-
-PrimeReact-specific props (`style`, `contentStyle`, `modal`, `dismissableMask`, `draggable`, `footer`, `onHide`) are **not** available — do not use them.
-
-````
diff --git a/.claude/rules/dialogs.md b/.claude/rules/dialogs.md
new file mode 120000
index 0000000..db237f3
--- /dev/null
+++ b/.claude/rules/dialogs.md
@@ -0,0 +1 @@
+../../.ai/rules/dialogs.md
\ No newline at end of file
diff --git a/.claude/rules/documentation.md b/.claude/rules/documentation.md
deleted file mode 100644
index 9f97bcb..0000000
--- a/.claude/rules/documentation.md
+++ /dev/null
@@ -1,49 +0,0 @@
----
-applyTo: "Documentation/**/*.md"
----
-
-# How to Write Documentation
-
-Documentation exists for one audience: **developers who need to use the framework** — not the team that built it. Write from the reader's perspective. They want to know *what this does*, *why they should care*, and *how to use it* — in that order.
-
-Every page should answer: "If I were a developer encountering this concept for the first time, what would I need to understand to use it correctly?"
-
-## General
-
-- All documentation lives in the `Documentation/` folder.
-- Site is built using [DocFX](https://dotnet.github.io/docfx/).
-- Use [Markdown](https://www.markdownguide.org/) with [GitHub Flavored Markdown](https://github.github.com/gfm/).
-- Use [Mermaid](https://mermaid-js.github.io/mermaid/#/) for diagrams — architecture flows, state transitions, and sequence diagrams are far clearer as visuals than as prose.
-- Follow [DocFX Markdown](https://dotnet.github.io/docfx/markdown/) guidelines.
-
-## Structure
-
-- Every folder must have its own `toc.yml` for navigation.
-- Every folder must have an `index.md` as a landing page with links to subtopics.
-- When linking to a folder in `toc.yml`, link to the folder's `toc.yml`, not to `index.md`.
-- Use relative links for internal references.
-
-## Writing Style
-
-The project's voice is **direct, practical, and opinionated**. Write like an experienced colleague explaining something to a capable developer — confident but never condescending.
-
-- **Active voice, present tense.** "Chronicle appends the event" not "The event is appended by Chronicle."
-- **Emphasize *why* before *how*.** The reason behind a design choice is more valuable than the steps to implement it. A developer who understands the *why* can handle edge cases the docs don't cover.
-- **Don't document the obvious.** If the API is self-explanatory, a code example is enough. Save prose for concepts that are genuinely non-obvious or where the reasoning is important.
-- Use headings, lists, and code blocks to organize content — dense paragraphs lose readers.
-- Use consistent terminology throughout. If it's called an "event source" in one place, don't call it an "event stream" elsewhere.
-- Focus on public APIs and features — not internal implementation.
-- Do not document third-party libraries.
-
-## Code Examples
-
-- Prefer `record` types for data structures (events, commands, read models) — this matches the actual codebase.
-- When specifying `[EventType]`, never add an explicit name argument — just `[EventType]`.
-- Never include verbatim code from the repository — APIs may change. Write purpose-built examples.
-- Every code example must be complete and correct — no pseudo-code, no `// ...` elisions that leave the reader guessing.
-
-## File Rules
-
-- Always end generated markdown with a single trailing newline in the file content itself.
-- Never use shell commands to modify files after writing them.
-- Run `verify-markdown.sh` in the Documentation folder after writing to validate links and formatting.
diff --git a/.claude/rules/documentation.md b/.claude/rules/documentation.md
new file mode 120000
index 0000000..3ee688d
--- /dev/null
+++ b/.claude/rules/documentation.md
@@ -0,0 +1 @@
+../../.ai/rules/documentation.md
\ No newline at end of file
diff --git a/.claude/rules/efcore.md b/.claude/rules/efcore.md
deleted file mode 100644
index db82457..0000000
--- a/.claude/rules/efcore.md
+++ /dev/null
@@ -1,232 +0,0 @@
----
-applyTo: "**/*.cs"
----
-
-# Entity Framework Core Instructions
-
-> **⚠️ APPLIES ONLY TO PROJECTS USING ENTITY FRAMEWORK CORE**
-> If your project does not reference `Microsoft.EntityFrameworkCore` or any EF Core packages, **ignore this entire file**. These rules are irrelevant outside of EF Core contexts.
-
-## Project Structure
-
-Responsibilities are split across three projects:
-
-| Project | Responsibility |
-|---------|----------------|
-| `Database` | Migrations only – no entities, no DbContexts |
-| `Core` | Entities and feature DbContexts (co-located with features) |
-| `Infrastructure` | DbContext registration, migration runner, cross-cutting EF setup |
-
-**Critical dependency rule**: `Database` must NEVER import from `Core`. Migrations reference only `WellKnownTables` constants (strings) and EF migration types. The dependency chain is:
-
-```
-Core → Infrastructure → Database
-```
-
-## DbContext Base Types
-
-Always use the Cratis Arc base types — never inherit directly from `DbContext`:
-
-- **`ReadOnlyDbContext`** — for all read model / projection contexts (the vast majority)
-- **`BaseDbContext`** — only for writable contexts that own state (e.g. device state, infrastructure state)
-
-```csharp
-// ✅ Read model context
-public class StartupPhaseDbContext(DbContextOptions<StartupPhaseDbContext> options)
-    : ReadOnlyDbContext(options)
-{
-    public DbSet<StartupPhase> StartupPhases => Set<StartupPhase>();
-    public DbSet<PersonnelAssignment> PersonnelAssignments => Set<PersonnelAssignment>();
-}
-
-// ✅ Writable (state-owning) context
-public class DeviceStateDbContext(DbContextOptions<DeviceStateDbContext> options)
-    : BaseDbContext(options)
-{
-    public DbSet<DeviceState> DeviceStates => Set<DeviceState>();
-}
-```
-
-Use the primary constructor pattern. Expose `DbSet<T>` as expression-bodied properties using `Set<T>()`.
-
-## Feature Contexts — Not God Contexts
-
-Create one focused DbContext per feature or tightly-related feature group. Never aggregate unrelated entities into a single context.
-
-```csharp
-// ❌ God context
-public class AppDbContext : DbContext
-{
-    public DbSet<Mission> Missions { get; set; }
-    public DbSet<User> Users { get; set; }
-    public DbSet<Station> Stations { get; set; }
-    // ... many more
-}
-
-// ✅ Focused feature context
-public class StartupPhaseDbContext(DbContextOptions<StartupPhaseDbContext> options)
-    : ReadOnlyDbContext(options)
-{
-    public DbSet<StartupPhase> StartupPhases => Set<StartupPhase>();
-    public DbSet<PersonnelAssignment> PersonnelAssignments => Set<PersonnelAssignment>();
-}
-```
-
-Co-locate the DbContext file with its feature:
-
-```
-Missions/Ongoing/StartupPhase/
-├── StartupPhase.cs
-├── StartupPhaseDbContext.cs
-└── ...
-```
-
-## State Mutation — The Golden Rule
-
-> **Never mutate state directly through a DbContext.**
-
-All state changes must flow through events and Chronicle projections. Direct writes bypass the audit trail and event log.
-
-```csharp
-// ❌ Direct mutation — forbidden
-dbContext.StartupPhases.Add(new StartupPhase(...));
-await dbContext.SaveChangesAsync();
-
-// ✅ Correct: emit an event, let the projection handle writes
-[Command]
-public record UpdateStartupPhase(MissionId MissionId, ...) { ... }
-```
-
-Only infrastructure projection code, the Chronicle event engine, and reference data sync may write through DbContexts.
-
-## Registration
-
-Use the Cratis Arc `Cratis.Arc.EntityFrameworkCore` extension methods:
-
-```csharp
-// Register a single writable DbContext
-services.AddDbContextWithConnectionString<DeviceStateDbContext>(connectionString, optionalConfigure);
-
-// Auto-discover and register ALL ReadOnlyDbContext subtypes from given assemblies
-services.AddReadModelDbContextsWithConnectionStringFromAssemblies(
-    connectionString,
-    configureOptions,
-    [Assembly.GetExecutingAssembly()]);
-```
-
-Configure the database provider using `UseDatabaseFromConnectionString`, which auto-detects PostgreSQL vs SQLite from the connection string:
-
-```csharp
-options.UseDatabaseFromConnectionString(connectionString);
-```
-
-Centralise all DbContext setup in a single `AddApplicationDbContexts` extension method per layer.
-
-## Multiple Database Support
-
-The application supports both PostgreSQL (ASP.NET mode) and SQLite (MAUI mode) from the same code. The provider is selected at runtime via the connection string — `UseDatabaseFromConnectionString` handles the detection.
-
-Never hardcode a provider (e.g. `UseSqlite` or `UseNpgsql`) in application code. Always use `UseDatabaseFromConnectionString`.
-
-## Migrations
-
-Migrations live exclusively in the **`Database`** project, never in `Core` or `Infrastructure`.
-
-### Organisation
-
-Each entity category has its own folder with versioned migration files:
-
-```
-Database/
-├── Missions/
-│   ├── v1_0_0.cs
-│   └── v1_1_0.cs
-├── Users/
-│   └── v1_0_0.cs
-└── WellKnownTables.cs
-```
-
-### Naming
-
-Version files using the pattern `v{major}_{minor}_{patch}.cs` and place them inside a namespace matching their folder:
-
-```csharp
-namespace Database.Missions;
-
-public class v1_0_0 : Migration
-{
-    protected override void Up(MigrationBuilder migrationBuilder) { ... }
-    protected override void Down(MigrationBuilder migrationBuilder) { ... }
-}
-```
-
-The migration ID is composed as `{folder}_{ClassName}` (e.g. `Missions_v1_0_0`).
-
-### Cross-Database Column Helpers
-
-Always use the Cratis Arc `MigrationBuilder` extension helpers to define columns. These abstract over PostgreSQL and SQLite type differences:
-
-| Helper | Use for |
-|--------|---------|
-| `table.StringColumn(migrationBuilder)` | Text / varchar columns |
-| `table.GuidColumn(migrationBuilder)` | UUID / GUID columns |
-| `table.NumberColumn<T>(migrationBuilder)` | Integer, long, or numeric columns |
-| `table.DateTimeOffsetColumn(migrationBuilder)` | Timestamps with timezone |
-
-Never use raw EF `table.Column<string>()` etc. — the helpers ensure cross-database compatibility.
-
-```csharp
-// ✅ Cross-database migration
-migrationBuilder.CreateTable(
-    name: WellKnownTables.Missions,
-    columns: table => new
-    {
-        Id = table.StringColumn(migrationBuilder, nullable: false),
-        Title = table.StringColumn(migrationBuilder, maxLength: 200, nullable: false),
-        ResourceId = table.NumberColumn<int>(migrationBuilder, nullable: true),
-        DispatchTime = table.DateTimeOffsetColumn(migrationBuilder),
-        UrgencyId = table.GuidColumn(migrationBuilder)
-    },
-    constraints: table =>
-    {
-        table.PrimaryKey("PK_Missions", x => x.Id);
-        table.ForeignKey("FK_Missions_Urgency", x => x.UrgencyId,
-            WellKnownTables.MissionUrgencies, "Id", onDelete: ReferentialAction.SetNull);
-    });
-```
-
-### Table Names — WellKnownTables
-
-Always reference table names from `WellKnownTables` constants. Never use magic strings directly in migrations.
-
-```csharp
-// ❌ Magic string
-migrationBuilder.CreateTable(name: "Missions", ...);
-
-// ✅ Constant
-migrationBuilder.CreateTable(name: WellKnownTables.Missions, ...);
-```
-
-Add new table names to `Database/WellKnownTables.cs` before writing the migration.
-
-### Applying Migrations
-
-Migrations are applied via a custom runner from the `Database` project (not `dotnet ef database update`):
-
-```csharp
-// ASP.NET mode
-await app.ApplyAllMigrations(connectionString);
-
-// MAUI / IServiceProvider mode
-await services.ApplyAllMigrations(connectionString);
-```
-
-The runner discovers all `Migration` subclasses from the `Database` assembly, checks the EF history table, and applies pending migrations in version order within a transaction. Both PostgreSQL and SQLite are supported through the same runner.
-
-## Auto-Discovery
-
-The Cratis Arc `IImplementationsOf<T>` mechanism discovers types at runtime:
-
-- `IImplementationsOf<BaseDbContext>` — all DbContext subtypes across all loaded assemblies
-- `.NotReadonly()` extension — filters out `ReadOnlyDbContext` subtypes to isolate writable contexts
-- `IReadModelDbContexts.GetAll()` — returns all registered `ReadOnlyDbContext` instances
diff --git a/.claude/rules/efcore.md b/.claude/rules/efcore.md
new file mode 120000
index 0000000..052fa02
--- /dev/null
+++ b/.claude/rules/efcore.md
@@ -0,0 +1 @@
+../../.ai/rules/efcore.md
\ No newline at end of file
diff --git a/.claude/rules/efcore.specs.md b/.claude/rules/efcore.specs.md
deleted file mode 100644
index d8866af..0000000
--- a/.claude/rules/efcore.specs.md
+++ /dev/null
@@ -1,40 +0,0 @@
----
-applyTo: "**/for_*/**/*.cs, **/when_*/**/*.cs"
----
-
-# Entity Framework Core Specs
-
-> **⚠️ APPLIES ONLY TO PROJECTS USING ENTITY FRAMEWORK CORE**
-> If your project does not reference `Microsoft.EntityFrameworkCore` or any EF Core packages, **ignore this entire file**. These rules are irrelevant outside of EF Core contexts.
-
-EF Core specs verify database interaction logic — migrations, queries, and error handling. They use SQLite in-memory databases, which are fast, isolated, and disposable. This keeps specs independent of any real database server.
-
-These specs are for code that interacts directly with `DbContext`. For event-sourcing integration specs (testing commands against Chronicle), use the Chronicle integration spec pattern in [specs.csharp.instructions.md](./specs.csharp.instructions.md) instead.
-
-## Database Setup
-
-- Use SQLite in-memory database for all `DbContext` specs — it's fast and each spec gets a clean database.
-- Configure and dispose the in-memory database properly to avoid state leakage between tests. Each spec run should start with a fresh schema.
-
-## Mocking DbContext
-
-- `SaveChanges()` and `SaveChangesAsync()` are virtual — mock with NSubstitute.
-- `DbSet<T>` methods are virtual — mock as needed.
-- Pass options when substituting: `Substitute.For<YourDbContext>(options)`.
-
-```csharp
-var options = new DbContextOptionsBuilder<YourDbContext>()
-    .UseSqlite("DataSource=:memory:")
-    .Options;
-
-var context = Substitute.For<YourDbContext>(options);
-```
-
-## Simulating Failures
-
-Mock `SaveChanges` / `SaveChangesAsync` to throw exceptions for testing error handling and recovery:
-
-```csharp
-context.SaveChangesAsync(Arg.Any<CancellationToken>())
-    .Throws(new DbUpdateException("Simulated failure"));
-```
diff --git a/.claude/rules/efcore.specs.md b/.claude/rules/efcore.specs.md
new file mode 120000
index 0000000..da11c11
--- /dev/null
+++ b/.claude/rules/efcore.specs.md
@@ -0,0 +1 @@
+../../.ai/rules/efcore.specs.md
\ No newline at end of file
diff --git a/.claude/rules/git-commits.md b/.claude/rules/git-commits.md
deleted file mode 100644
index 62322d9..0000000
--- a/.claude/rules/git-commits.md
+++ /dev/null
@@ -1,94 +0,0 @@
----
-applyTo: "**/*"
----
-
-# How to Write Git Commits
-
-Commits are the permanent record of how the codebase evolved. Each commit should tell a clear story: *what* changed and *why*. A reviewer reading `git log --oneline` should understand the arc of the work without opening any diffs.
-
-## Logical Grouping
-
-Every commit must be a **single logical unit of work**. Group related changes together; separate unrelated changes into distinct commits.
-
-### What belongs in one commit
-
-- A bug fix and the spec that proves it.
-- A new file and the changes to existing files needed to integrate it (imports, registrations, wiring).
-- A refactor that moves or renames code — only the mechanical transformation, nothing else.
-- An interface change together with all implementation updates required to compile.
-
-### What does NOT belong in one commit
-
-- A bug fix mixed with an unrelated feature.
-- Source code changes mixed with unrelated spec additions for a different area.
-- Formatting or style cleanups bundled with behavioral changes.
-- Multiple independent fixes or features squashed into a single commit.
-
-### Deciding where to split
-
-Ask: "If I needed to revert this commit, would I lose exactly one coherent change?" If reverting would undo two unrelated things, it should be two commits.
-
-Common split points:
-1. **Infrastructure / plumbing first** — interface additions, new types, or schema changes that later commits build on.
-2. **Core logic second** — the behavioral change that uses the new infrastructure.
-3. **Specs / tests third** — the specs that prove the behavioral change works. Specs may also be combined with the core logic commit when they are tightly coupled (e.g., a TDD red-green cycle or a bug fix with its regression test).
-4. **Integration or wiring last** — connecting the new behavior to the rest of the system (DI registration, routing, UI hookup).
-
-When a task produces both source fixes and new integration specs, prefer separate commits for the source changes and the specs — unless the specs are inseparable from the fix (e.g., a single bug fix + its regression test).
-
-## Commit Messages
-
-### Format
-
-```
-<imperative summary of what this commit does>
-
-<optional body — why the change was made, context, trade-offs>
-```
-
-- **Subject line**: imperative mood, present tense. Start with a verb: `Add`, `Fix`, `Remove`, `Rename`, `Extract`, `Update`, `Support`.
-- **No period** at the end of the subject line.
-- **72-character limit** on the subject line. If you cannot describe the change in 72 characters, the commit is probably too large — split it.
-- **Body**: separated from the subject by a blank line. Explain *why*, not *what* (the diff shows the what). Use bullet points for multi-part changes.
-
-### Good examples
-
-```
-Fix duplicate key crash in IdentityStorage.Populate
-
-The upsert used InsertOne which threw on existing identities.
-Replace with ReplaceOne using upsert: true.
-```
-
-```
-Add type-safe event migration API with expression-based builders
-
-Introduce EventTypeMigration<TUpgrade, TPrevious> base class with
-typed property builders for Split, Join, Rename, and DefaultValue
-operations. Migrators are discovered automatically by convention.
-```
-
-```
-Add integration specs for observer replay on redaction
-```
-
-### Bad examples
-
-- `Fix stuff` — meaningless.
-- `WIP` — never commit work-in-progress; stage and commit when the unit is complete.
-- `Add files` — says nothing about what or why.
-- `Fix bug and add new feature and update docs` — three unrelated things.
-- `Changed Observer.Handling.cs` — describes a file, not a behavior.
-
-## When to Commit
-
-- **After each logical unit passes the build** — `dotnet build` with zero errors and zero warnings, or `yarn compile` with zero errors.
-- **Before starting a different kind of work** — about to switch from fixing a bug to adding a feature? Commit the bug fix first.
-- **After completing specs for a change** — if the specs are a separate commit from the source change.
-- **Never commit code that does not compile.** Every commit must be a buildable, working state of the codebase.
-
-## Staging Discipline
-
-- Use `git add <specific files>` rather than `git add .` or `git add -A`. Only stage files that belong to the current logical unit.
-- Review `git diff --cached` before committing to verify nothing unrelated was staged.
-- If you realize mid-commit that unrelated changes are mixed in, unstage them and commit only the related subset.
diff --git a/.claude/rules/git-commits.md b/.claude/rules/git-commits.md
new file mode 120000
index 0000000..cf627c8
--- /dev/null
+++ b/.claude/rules/git-commits.md
@@ -0,0 +1 @@
+../../.ai/rules/git-commits.md
\ No newline at end of file
diff --git a/.claude/rules/managing-ai-rules.md b/.claude/rules/managing-ai-rules.md
deleted file mode 100644
index 0acd2a5..0000000
--- a/.claude/rules/managing-ai-rules.md
+++ /dev/null
@@ -1,111 +0,0 @@
----
-applyTo: "**/*"
----
-
-# Managing AI Rules and Instructions
-
-`.ai/rules/` is the **single source of truth** for all AI assistant rules in this repository. Rules are written once and surfaced to each AI tool through symlinks. When asked to add, rename, or update a rule, always work in `.ai/rules/` — never edit the symlinks directly.
-
-## Folder structure
-
-```
-.ai/
-├── rules/          ← canonical rule files (real files, not symlinks)
-├── workflows/      ← shared GitHub Actions workflow files
-├── prompts/        ← reusable prompt templates
-└── agents/         ← agent definitions
-
-.github/
-├── copilot-instructions.md   ← symlink → ../.ai/rules/general.md
-└── instructions/
-    └── <name>.instructions.md  ← symlinks → ../../.ai/rules/<name>.md
-
-.claude/
-├── CLAUDE.md                 ← symlink → ../.ai/rules/general.md
-└── rules/
-    └── <name>.md             ← symlinks → ../../.ai/rules/<name>.md
-```
-
-## Rule file format
-
-Every rule file in `.ai/rules/` must start with a YAML frontmatter block containing at minimum an `applyTo` field (for GitHub Copilot). Add a `paths` field when the rule should also be scoped for Claude Code.
-
-```markdown
----
-applyTo: "**/*.cs"
-paths:
-  - "**/*.cs"
----
-
-# Rule Title
-
-Rule content here.
-```
-
-Use `applyTo: "**/*"` (and omit `paths`) for rules that apply to all files.
-
-## Adding a new rule
-
-1. **Create the canonical file** in `.ai/rules/<name>.md` with the appropriate frontmatter and content.
-
-2. **Create the Copilot symlink** in `.github/instructions/`:
-
-   ```bash
-   cd .github/instructions
-   ln -s ../../.ai/rules/<name>.md <name>.instructions.md
-   ```
-
-3. **Create the Claude symlink** in `.claude/rules/`:
-
-   ```bash
-   cd .claude/rules
-   ln -s ../../.ai/rules/<name>.md <name>.md
-   ```
-
-4. If the rule applies to all files globally (like `general.md`), update the top-level symlinks:
-   - `.github/copilot-instructions.md` → `../.ai/rules/general.md`
-   - `.claude/CLAUDE.md` → `../.ai/rules/general.md`
-
-## Updating an existing rule
-
-Edit the canonical file in `.ai/rules/<name>.md`. The symlinks in `.github/instructions/` and `.claude/rules/` automatically reflect the change — nothing else needs to be touched.
-
-## Renaming a rule
-
-1. Rename the file in `.ai/rules/`.
-2. Remove the old symlinks and recreate them pointing to the new filename:
-
-   ```bash
-   # In .github/instructions/
-   rm <old-name>.instructions.md
-   ln -s ../../.ai/rules/<new-name>.md <new-name>.instructions.md
-
-   # In .claude/rules/
-   rm <old-name>.md
-   ln -s ../../.ai/rules/<new-name>.md <new-name>.md
-   ```
-
-3. Update any cross-references within other rule files that link to the renamed file by path.
-
-## Symlink path conventions
-
-Symlink targets use **relative paths** from the symlink's location to the canonical file:
-
-| Symlink location | Target prefix |
-|---|---|
-| `.github/instructions/` | `../../.ai/rules/` |
-| `.claude/rules/` | `../../.ai/rules/` |
-| `.github/copilot-instructions.md` | `../.ai/rules/` |
-| `.claude/CLAUDE.md` | `../.ai/rules/` |
-
-## Propagation and symlinks
-
-The cross-repository propagation workflow reads `.github/instructions/` and `.github/copilot-instructions.md` via the GitHub API. Because the GitHub API returns symlink blob content verbatim (the raw target path string), a naïve script would push path strings instead of actual rule content to target repositories.
-
-The propagation script in this repository handles this correctly: when it encounters a symlink (Git mode `120000`) in the source tree, it resolves the target path and substitutes the real file's SHA before propagating. This means **symlinks work as expected** — target repositories receive the actual instruction content, not path strings.
-
-Both `.claude/` and `.github/instructions/` therefore use symlinks consistently. There is no need to maintain real file copies anywhere.
-
-## Shared workflows
-
-Workflow files intended to be synced to other repositories live in `.ai/workflows/`. They follow the same symlink pattern — the propagate workflow copies `.ai/workflows/` content to target repositories.
diff --git a/.claude/rules/managing-ai-rules.md b/.claude/rules/managing-ai-rules.md
new file mode 120000
index 0000000..7618121
--- /dev/null
+++ b/.claude/rules/managing-ai-rules.md
@@ -0,0 +1 @@
+../../.ai/rules/managing-ai-rules.md
\ No newline at end of file
diff --git a/.claude/rules/orleans.md b/.claude/rules/orleans.md
deleted file mode 100644
index 7efeee3..0000000
--- a/.claude/rules/orleans.md
+++ /dev/null
@@ -1,50 +0,0 @@
-````instructions
----
-applyTo: "**/*.cs"
----
-
-# Orleans Conventions
-
-> **⚠️ APPLIES ONLY TO PROJECTS USING MICROSOFT ORLEANS**
-> If your project does not reference `Microsoft.Orleans` or any Orleans packages, **ignore this entire file**. These rules are irrelevant outside of Orleans contexts.
-
-These conventions apply to projects that use Microsoft Orleans for distributed grain-based actors.
-
-## General
-
-- Do **not** split Orleans grain interfaces into a separate project. Keep all grain types in the same project as their implementations.
-- Use localhost clustering for local development (`UseLocalhostClustering()`).
-- Do **not** add `[Alias]` attributes to grain interfaces or methods — rely on the default alias generated by Orleans.
-
-## Storage Providers
-
-- Centralise all storage provider names in a single `WellKnownStorageProviders` static class at the root of the project.
-- Reference provider names via the constants in `WellKnownStorageProviders` — never use magic strings.
-- Use `WellKnownStorageProviders.Default` as the name for the primary MongoDB-backed storage provider.
-
-```csharp
-public static class WellKnownStorageProviders
-{
-    public const string Default = "MongoDbGrainStorage";
-}
-```
-
-- Annotate grains with `[StorageProvider(ProviderName = WellKnownStorageProviders.Default)]` when they need persistent state.
-
-## MongoDB Grain Storage
-
-- The `MongoDbGrainStorage<TState>` implementation accepts `IMongoCollection<TState>` as its only dependency — no additional plumbing is required.
-- Register it as a keyed singleton in the Orleans silo builder:
-
-```csharp
-builder.Host.UseOrleans(siloBuilder =>
-{
-    siloBuilder.UseLocalhostClustering();
-    siloBuilder.Services.AddKeyedSingleton<IGrainStorage>(
-        WellKnownStorageProviders.Default,
-        (sp, _) => new MongoDbGrainStorage<MyState>(
-            sp.GetRequiredService<IMongoCollection<MyState>>()));
-});
-```
-
-````
diff --git a/.claude/rules/orleans.md b/.claude/rules/orleans.md
new file mode 120000
index 0000000..6df7e49
--- /dev/null
+++ b/.claude/rules/orleans.md
@@ -0,0 +1 @@
+../../.ai/rules/orleans.md
\ No newline at end of file
diff --git a/.claude/rules/pull-requests.md b/.claude/rules/pull-requests.md
deleted file mode 100644
index 734934a..0000000
--- a/.claude/rules/pull-requests.md
+++ /dev/null
@@ -1,43 +0,0 @@
----
-applyTo: "**/*"
----
-
-# How to Do Pull Requests
-
-PR descriptions serve two purposes: they help reviewers understand the change *now*, and they become the release notes that users read *later*. Write them with both audiences in mind.
-
-## Description
-
-- Follow the [pull request template](../pull_request_template.md).
-- Focus on the **Added**, **Changed**, **Fixed**, **Removed**, **Security**, and **Deprecated** sections. Remove sections that are empty — don't leave blank headings.
-- Each bullet should be short, self-contained, and release-note ready.
-- **Write for users of the framework, not for internal developers.** Only include changes that have an impact on anyone using what we build — new APIs, changed behavior, fixed bugs, removed features. Do not list internal implementation details like storage changes, converter updates, gRPC contract internals, or spec additions. If a change is purely internal plumbing, it does not belong in the PR description.
-- Add the associated issue reference at the end of a bullet when there is a real GitHub issue for the change (e.g. `(#351)`). If there is no associated issue, omit the reference entirely. Never use a placeholder like `(#issue)` or leave the example number `(#123)` literally, and never invent a random issue number. **Always verify the issue number using the `search_issues` or `list_issues` GitHub MCP tool — never guess or invent a number.**
-- Include a summary only if there is a cohesive theme across the changes. If you find yourself restating individual bullets in slightly different words, the summary adds no value — remove it.
-- Never include Copilot prompt content in the PR description. Remove any "Original prompt" / coding agent transcript blocks before publishing.
-
-## Commits
-
-See the full [Git Commits guide](./git-commits.instructions.md) for rules on logical grouping, message format, and staging discipline.
-
-Quick reminders:
-- Imperative mood: "Add author registration" not "Added author registration".
-- Each commit = one logical unit of work. No WIP commits in the final PR.
-- Never mix unrelated changes in a single commit.
-
-## Labels
-
-- Label the PR according to semantic versioning impact:
-  - **major** — breaking changes to public APIs
-  - **minor** — new features, new slices, non-breaking additions
-  - **patch** — bug fixes, docs, refactoring with identical behavior
-
-## Quality Gates
-
-Before marking a PR ready for review:
-- `dotnet build` — zero errors, zero warnings
-- `dotnet test` — all specs pass
-- `yarn lint` — zero errors
-- `npx tsc -b` — zero TypeScript errors
-- Code follows all project coding standards and conventions
-- **CI checks pass** — after pushing, use GitHub MCP tools (`pull_request_read` with `get_check_runs`, `get_job_logs`) to monitor CI results. If any checks fail, investigate the logs, fix the failures, and push again. The task is not complete until all CI checks pass or the remaining failures are confirmed to be pre-existing flaky tests unrelated to the PR changes.
diff --git a/.claude/rules/pull-requests.md b/.claude/rules/pull-requests.md
new file mode 120000
index 0000000..c8ea2a6
--- /dev/null
+++ b/.claude/rules/pull-requests.md
@@ -0,0 +1 @@
+../../.ai/rules/pull-requests.md
\ No newline at end of file
diff --git a/.claude/rules/reactors.md b/.claude/rules/reactors.md
deleted file mode 100644
index 36a523f..0000000
--- a/.claude/rules/reactors.md
+++ /dev/null
@@ -1,103 +0,0 @@
----
-applyTo: "**/*.cs"
----
-
-# Reactor Instructions
-
-Reactors are the "if this then that" of event sourcing — they observe events and produce side effects. Unlike projections (which build state), reactors *do things*: send emails, trigger commands in other slices, call external APIs.
-
-## IReactor — Marker Interface
-
-`IReactor` is a **marker interface** with no methods to implement. Method dispatch is entirely by convention: the first parameter type of each public method determines which event it handles.
-
-```csharp
-public class ProjectRegisteredNotifier(INotificationService notifications) : IReactor
-{
-    /// <summary>
-    /// Reacts to <see cref="ProjectRegistered"/> events by sending a notification.
-    /// </summary>
-    /// <param name="event">The event.</param>
-    /// <param name="context">The event context.</param>
-    public async Task ProjectRegistered(ProjectRegistered @event, EventContext context) =>
-        await notifications.Notify($"Project '{@event.Name}' was registered.");
-}
-```
-
-## Method Signature
-
-```csharp
-public Task MethodName(TEvent @event, EventContext context)
-```
-
-- **First parameter** — the event type. This determines which events the method subscribes to.
-- **Second parameter** — `EventContext` (optional). Omit if event metadata is not needed.
-- **Return type** — `Task` (always async).
-- **Method name** — can be anything descriptive. The name is for readability, not dispatch.
-
-## Critical Rules
-
-1. **Idempotent** — Reactors may be called more than once for the same event (e.g. during replay or recovery). Design accordingly.
-2. **Use event data directly** — Never query the read model back inside a reactor. The event contains all the information you need.
-3. **Trigger commands for further writes** — If the reactor needs to produce new events, inject `ICommandPipeline` and execute a command. Never use `IEventLog` directly from a reactor.
-4. **Single responsibility** — Each reactor class should have a focused purpose. Multiple handler methods in one reactor are fine if they serve the same automation concern.
-5. **Failure behavior** — If a reactor throws, the failing event source partition pauses until the issue is resolved. Design for resilience.
-6. **No state** — Reactors should be stateless. Inject dependencies via primary constructor, but do not store mutable state on the class.
-
-## Slice Types That Use Reactors
-
-| Slice type | Pattern |
-|------------|---------|
-| **Automation** | Reacts to events, makes decisions, triggers side effects |
-| **Translation** | Adapts events from one slice/system by triggering commands in another |
-
-### Automation Example
-
-```csharp
-public class ProjectRegisteredNotifier(INotificationService notifications) : IReactor
-{
-    /// <summary>
-    /// Sends a notification when a project is registered.
-    /// </summary>
-    /// <param name="event">The event.</param>
-    /// <param name="context">The event context.</param>
-    public async Task ProjectRegistered(ProjectRegistered @event, EventContext context) =>
-        await notifications.Notify($"Project '{@event.Name}' was registered.");
-}
-```
-
-### Translation Example
-
-```csharp
-public class StockKeeping(IStockKeeper stockKeeper, ICommandPipeline commandPipeline) : IReactor
-{
-    /// <summary>
-    /// Reacts to a book reservation by decreasing stock.
-    /// </summary>
-    /// <param name="event">The event.</param>
-    /// <param name="context">The event context.</param>
-    public async Task BookReserved(BookReserved @event, EventContext context) =>
-        await commandPipeline.Execute(new DecreaseStock(@event.Isbn, await stockKeeper.GetStock(@event.Isbn)));
-}
-```
-
-## Testing Reactors
-
-Reactor specs follow the same BDD-style pattern as other specs. Use `IReactorInvoker` and `ReactorHandler` from the test infrastructure:
-
-```csharp
-// given/all_dependencies.cs
-public class all_dependencies : Specification
-{
-    protected IEventStore _eventStore;
-    protected IReactorInvoker _reactorInvoker;
-}
-
-// given/a_reactor_handler.cs
-public class a_reactor_handler : all_dependencies
-{
-    protected ReactorHandler _handler;
-    void Establish() => _handler = new(_eventStore, _reactorInvoker);
-}
-```
-
-See `specs.csharp.instructions.md` for full spec patterns.
diff --git a/.claude/rules/reactors.md b/.claude/rules/reactors.md
new file mode 120000
index 0000000..c7d78cb
--- /dev/null
+++ b/.claude/rules/reactors.md
@@ -0,0 +1 @@
+../../.ai/rules/reactors.md
\ No newline at end of file
diff --git a/.claude/rules/specs.csharp.md b/.claude/rules/specs.csharp.md
deleted file mode 100644
index 3449757..0000000
--- a/.claude/rules/specs.csharp.md
+++ /dev/null
@@ -1,249 +0,0 @@
----
-applyTo: "**/for_*/**/*.cs, **/when_*/**/*.cs"
----
-
-# How to Write C# Specs
-
-Extends the base [Specs Instructions](./specs.instructions.md) with C#-specific conventions.
-
-The `Cratis.Specifications` library was built to maintain the approach, structure, and syntax of Machine.Specifications (MSpec) — a BDD framework that makes specs read like a human-language specification. The `Establish → Because → should_` flow maps directly to "Given → When → Then" and keeps specs focused on *one action, one setup, one set of assertions*.
-
-## Frameworks
-
-- [xUnit](https://xunit.net/) for test execution.
-- [NSubstitute](https://nsubstitute.github.io/) for mocking — chosen for its clean API that reads naturally.
-- [Cratis.Specifications](https://github.com/Cratis/Specifications) for BDD-style specification by example.
-- Spec projects are named `<Source>.Specs` (e.g. `Infrastructure.Specs`).
-
-## Base Class
-
-`Specification` from `Cratis.Specifications` must be at the root of every spec's inheritance chain.
-It discovers `Establish`, `Because`, and `Destroy` methods by convention — no attributes needed.
-
-## BDD Pattern
-
-The three-phase pattern makes every spec self-explanatory: `Establish` sets up the world, `Because` performs the single action under test, and `should_*` facts verify individual outcomes. No test framework attributes are needed on `Establish` or `Because` — `Cratis.Specifications` discovers them by convention.
-
-```csharp
-public class when_combining_parts : Specification
-{
-    object[] _parts;
-    string _result;
-
-    void Establish() => _parts = ["First", "Second", "Third"];
-
-    void Because() => _result = KeyHelper.Combine(_parts);
-
-    [Fact] void should_combine_all_parts() => _result.ShouldEqual("First+Second+Third");
-}
-```
-
-| Method | Purpose | Notes |
-|---|---|---|
-| `void Establish()` | Setup — called before `Because()` | Each class in the chain gets its own, called base-first |
-| `void Because()` | The single action under test | Only in concrete spec files, never in reusable contexts |
-| `[Fact] void should_*()` | One assertion per fact | Use `ShouldXxx()` extension methods |
-| `void Destroy()` | Teardown after each test | |
-
-All methods can be `async Task` when needed.
-
-## Reusable Context Pattern
-
-Layered contexts build on each other like building blocks. The base layer mocks all dependencies, the next layer creates the system under test, and each spec only adds what's unique to its scenario. This eliminates setup duplication while keeping every spec self-contained.
-
-```csharp
-// for_Changeset/given/a_changeset.cs
-namespace MyApp.Changes.for_Changeset.given;
-
-public class a_changeset : Specification
-{
-    protected IObjectComparer _comparer;
-    protected Changeset<object, ExpandoObject> _changeset;
-
-    void Establish()
-    {
-        _comparer = Substitute.For<IObjectComparer>();
-        _changeset = new(_comparer, new object(), new ExpandoObject());
-    }
-}
-```
-
-```csharp
-// for_Changeset/when_adding_changes/and_there_are_differences.cs
-namespace MyApp.Changes.for_Changeset.when_adding_changes;
-
-public class and_there_are_differences : given.a_changeset
-{
-    bool _result;
-
-    void Establish() =>
-        _comparer.Compare(Arg.Any<object>(), Arg.Any<ExpandoObject>(), out Arg.Any<IEnumerable<PropertyDifference>>())
-            .Returns(true);
-
-    void Because() => _result = _changeset.HasChanges;
-
-    [Fact] void should_have_changes() => _result.ShouldBeTrue();
-}
-```
-
-**Layered givens (chaining contexts):**
-
-```csharp
-// given/all_dependencies.cs — mocks all deps
-public class all_dependencies : Specification
-{
-    protected IEventStore _eventStore;
-    protected IReactorInvoker _reactorInvoker;
-    void Establish() { _eventStore = Substitute.For<IEventStore>(); /* ... */ }
-}
-
-// given/a_reactor_handler.cs — builds the system under test
-public class a_reactor_handler : all_dependencies
-{
-    protected ReactorHandler _handler;
-    void Establish() => _handler = new(_eventStore, _reactorInvoker);
-}
-```
-
-## NSubstitute Patterns
-
-```csharp
-// Create substitutes
-_service = Substitute.For<IMyService>();
-
-// Setup returns
-_service.GetValue(Arg.Any<string>()).Returns("result");
-_service.ProcessAsync(Arg.Any<int>()).Returns(Task.FromResult(42));
-
-// Argument matchers
-Arg.Is<Request>(r => r.Id == expectedId && r.Name == expectedName)
-
-// Verify received calls
-_service.Received(1).DoSomething(Arg.Any<string>());
-_service.DidNotReceive().DoSomethingElse(Arg.Any<int>());
-
-// Capture arguments
-_service.When(s => s.Add(Arg.Any<IDictionary<string, string>>()))
-    .Do(info => _captured = info.Arg<IDictionary<string, string>>());
-
-// Throw exceptions
-_handler.Handle(Arg.Any<CommandContext>()).Throws(new Exception("fail"));
-```
-
-## Assertion Extension Methods
-
-From `Cratis.Specifications`:
-
-| Method | Usage |
-|---|---|
-| `.ShouldEqual(expected)` | `_result.ShouldEqual(42)` |
-| `.ShouldBeTrue()` | `_flag.ShouldBeTrue()` |
-| `.ShouldBeFalse()` | `_flag.ShouldBeFalse()` |
-| `.ShouldBeNull()` | `_error.ShouldBeNull()` |
-| `.ShouldNotBeNull()` | `_value.ShouldNotBeNull()` |
-| `.ShouldBeEmpty()` | `_list.ShouldBeEmpty()` |
-| `.ShouldNotBeEmpty()` | `_list.ShouldNotBeEmpty()` |
-| `.ShouldContain(item)` | `_list.ShouldContain(expected)` |
-| `.ShouldNotContain(item)` | `_list.ShouldNotContain(excluded)` |
-| `.ShouldContainOnly(items)` | `_list.ShouldContainOnly(expectedItems)` |
-| `.ShouldBeOfExactType<T>()` | `_obj.ShouldBeOfExactType<PropertiesChanged<ExpandoObject>>()` |
-| `.ShouldBeGreaterThan(n)` | `_count.ShouldBeGreaterThan(0)` |
-| `.ShouldBeLessThan(n)` | `_count.ShouldBeLessThan(100)` |
-
-**Catching exceptions:**
-
-```csharp
-async Task Because() => _error = await Catch.Exception(_sut.DoSomething);
-
-[Fact] void should_not_fail() => _error.ShouldBeNull();
-```
-
-## Using Statements
-
-- Common usings are provided globally in `GlobalUsings.Specs.cs` (`Xunit`, `NSubstitute`, `Cratis.Specifications`, etc.) — don't duplicate them.
-- Don't add a using statement for the namespace of the system under test.
-
-## Properties — What NOT to Spec
-
-Simple properties are compiler-verified — the type system already guarantees they work. Writing specs for them adds maintenance cost without catching real bugs. Save spec effort for code where errors actually hide: business logic, coordination between dependencies, and complex transformations.
-
-```csharp
-// ❌ Do NOT write specs for these
-public string TableName => tableName;                         // Returns constructor parameter
-public Key Key => key;                                        // Returns field
-public IEnumerable<Property> Properties => mapper.Properties; // Simple delegation
-
-// ✅ Only write specs for complex business logic in properties
-public decimal TotalCost => Items.Sum(i => i.Cost * i.Quantity * (1 + i.TaxRate));
-```
-
----
-
-## Chronicle Integration Specs
-
-Integration specs are the highest-value specs in the project. They test a complete vertical slice — from HTTP request through command handling, event appending, constraint checking, and projection — against a real Chronicle event store. If one of these passes, the entire stack works.
-
-They live under `when_<behavior>/` folders directly inside the slice folder — **no `for_` folder**, because there's no isolated unit under test. The "unit" is the entire slice.
-
-### Structure
-
-```csharp
-using context = MyApp.Authors.Registration.when_registering.and_name_already_exists.context;
-
-namespace MyApp.Authors.Registration.when_registering;
-
-[Collection(ChronicleCollection.Name)]
-public class and_name_already_exists(context context) : Given<context>(context)
-{
-    public class context(ChronicleOutOfProcessFixture fixture) : given.an_http_client(fixture)
-    {
-        public const string AuthorName = "John Doe";
-        public CommandResult<object>? Result;
-
-        async Task Establish() =>
-            await EventStore.EventLog.Append(AuthorId.New(), new AuthorRegistered(AuthorName));
-
-        async Task Because()
-        {
-            Result = await Client.ExecuteCommand<RegisterAuthor>(
-                "/api/authors/register",
-                new RegisterAuthor(AuthorName));
-        }
-    }
-
-    [Fact] void should_not_be_successful() => Context.Result.IsSuccess.ShouldBeFalse();
-    [Fact] void should_have_appended_only_one_event() => Context.ShouldHaveTailSequenceNumber(EventSequenceNumber.First);
-}
-```
-
-### Key Rules
-
-- `context` is an inner public class inheriting from `given.an_http_client(fixture)`.
-- Use `async Task Establish()` to seed the event store with preconditions.
-- Use `async Task Because()` to execute the command under test.
-- `ExecuteCommand<TCommand>(url, cmd)` returns `CommandResult<object>?` (single type parameter).
-- `ExecuteCommand<TCommand, TResult>(url, cmd)` returns `CommandResult<TResult>?` (two type parameters for typed response).
-- Declare `Result` initialized with `null!` to satisfy nullable analysis.
-- Add a `using context = ...` alias at the top of each spec file.
-
-### Integration Assertion Helpers
-
-| Assertion | Purpose |
-|---|---|
-| `Context.Result.IsSuccess.ShouldBeTrue()` | Command succeeded |
-| `Context.Result.IsSuccess.ShouldBeFalse()` | Command failed (validation, constraint, etc.) |
-| `Context.ShouldHaveTailSequenceNumber(n)` | Verify event log tail (First = sequence 0) |
-| `Context.ShouldHaveAppendedEvent<TEvent>(seq, eventSourceId, validator)` | Verify specific event was appended |
-
-## Formatting
-
-- Don't break long `should_` method lines — prefer one-line lambda assertions.
-- Don't add blank lines between multiple `should_` methods.
-
-## Entity Framework Core Specs
-
-- Use SQLite in-memory database for specs involving `DbContext`.
-- `SaveChanges` / `SaveChangesAsync` are virtual and can be mocked with NSubstitute.
-- `DbSet` methods are virtual — mock as needed.
-- Pass options when substituting: `Substitute.For<YourDbContext>(options)`.
-- Simulate failures by mocking `SaveChanges` to throw exceptions.
diff --git a/.claude/rules/specs.csharp.md b/.claude/rules/specs.csharp.md
new file mode 120000
index 0000000..51b6467
--- /dev/null
+++ b/.claude/rules/specs.csharp.md
@@ -0,0 +1 @@
+../../.ai/rules/specs.csharp.md
\ No newline at end of file
diff --git a/.claude/rules/specs.md b/.claude/rules/specs.md
deleted file mode 100644
index 3dac23c..0000000
--- a/.claude/rules/specs.md
+++ /dev/null
@@ -1,120 +0,0 @@
----
-applyTo: "**/for_*/**/*.*, **/when_*/**/*.*"
----
-
-# How to Write Specs
-
-We call automated tests **specs** (specifications), not tests. This is deliberate — specs are executable documentation that describe what the system *does*, written in the language of the domain. A new developer should be able to read the spec folder structure like a table of contents and understand the system's behavior without opening a single source file.
-
-This philosophy comes from Specification by Example and BDD (Behavior Driven Development). The goal is human-readable, navigable specifications that double as a living contract.
-
-## Core Philosophy
-
-- **Specify behaviors, not implementations.** A spec should verify what a method *promises from its signature* — its contract with callers. If the implementation changes but the contract holds, specs should still pass. When they don't, the spec was testing the wrong thing.
-- **One behavior, one spec.** Every public method that performs an action gets its own `when_` folder or file. Never bundle multiple behaviors into one spec — it obscures what broke and why.
-- **Specs are documentation first.** The folder tree, class names, and `should_` assertions form a specification anyone can read. Optimize for readability over DRY. A little repetition in setup is fine if it makes the spec self-contained and clear.
-- **Do not test logging** — it is too fragile and provides no value. Don't test simple delegation or trivial getters either. The cost of maintaining these specs exceeds the value they provide.
-
-## Folder & File Structure
-
-Specs mirror the source structure and read like a sentence when you trace the path: `for_Changeset / when_adding_changes / and_there_are_differences`. This is not accidental — the folder hierarchy *is* the specification. Every level adds context:
-
-```
-for_<TypeUnderTest>/
-├── given/
-│   ├── all_dependencies              ← common DI/mock setup
-│   └── a_<descriptive_name>          ← reusable context
-├── when_<behavior>/                  ← folder for behaviors with multiple outcomes
-│   ├── given/                        ← behavior-specific context (optional)
-│   │   └── a_<specific_setup>
-│   ├── and_<condition>.cs            ← spec file for one outcome
-│   ├── and_<condition>/              ← OR a sub-folder when that condition itself has multiple outcomes
-│   │   ├── with_<data_state>.cs
-│   │   └── without_<requirement>.cs
-│   ├── with_<data_state>.cs
-│   └── without_<requirement>.cs
-└── when_<simple_behavior>.cs         ← single file for single-outcome behaviors
-```
-
-The `and_`, `with_`, `without_`, `having_`, `given_` prepositions work **both** as file names and as folder names. Use a folder when there are multiple outcomes under that condition; use a file when there is only one.
-
-**Naming conventions — read them as English sentences:**
-| Element | Pattern | Reads as... |
-|---|---|---|
-| Unit folder | `for_<ClassName>` | "For the Changeset..." |
-| Behavior folder | `when_<action>` | "...when adding changes..." |
-| Condition folder or spec file | Descriptive preposition | "...and there are differences" |
-| Assertion | `should <expected result>` | "...it should return true" |
-
-**Allowed prepositions for spec file/class names (and sub-folder names):**
-- `and_*` — additional conditions or compound scenarios
-- `with_*` / `without_*` — specific data or state present/absent
-- `having_*` — possession or state-based conditions
-- `given_*` — precondition scenarios
-
-**Critical naming rule — never embed `when` in a spec file or folder name:**
-`when` belongs **only** in `when_<behavior>` folder names. A spec file, spec class, or non-`when_` folder must **never** contain the word `when` anywhere in it. If the name starts with a preposition (`with_`, `and_`, etc.) but also contains `_when_` somewhere in the middle (e.g. `with_a_registered_migration_when_appending_a_generation_1_event`), you have two "whens" in the sentence — which is always wrong.
-
-The fix is to fold the context into the `when_` folder name itself, then use preposition files/folders for the outcomes:
-
-```
-# ❌ Wrong — double when in the sentence path
-when_appending_event_with_migrations/
-└── with_a_registered_migration_when_appending_a_generation_1_event.cs
-
-# ❌ Still wrong — unnecessary extra level when a single flat file suffices
-when_appending_event_with_migrations/
-└── and_event_is_generation_1/
-    └── with_a_registered_migration.cs
-
-# ✅ Correct — context baked into the when_ folder; outcomes are flat files
-when_appending_event_with_registered_migration/
-├── and_event_is_generation_1.cs
-├── and_event_is_generation_2.cs
-└── and_event_has_default_value.cs
-```
-
-A sub-folder under `when_` is only needed when that condition has its **own** multiple outcomes that warrant further breakdown. If there is only one outcome per condition, use a flat file. This applies to **all languages** (C#, TypeScript, etc.).
-
-## What to Specify
-
-The goal is to cover *decisions and transformations* — code where bugs hide. Simple plumbing that the compiler already validates is noise.
-
-**Write specs for:**
-- Public methods that perform actions (behaviors)
-- Methods with branching logic or business rules
-- Methods that coordinate between dependencies
-
-**Do NOT write specs for:**
-- Simple auto-properties (`public string Name { get; set; }`)
-- Properties returning constructor parameters (`public Key Key => key;`)
-- Simple delegation (`public IEnumerable<Property> Properties => mapper.Properties;`)
-- Trivial null checks or basic validation without complex logic
-- Getters returning injected dependencies
-
-**Avoid file names starting with:** `when_getting_*`, `when_returning_*` — if a spec name starts with "getting" or "returning", it's probably testing a simple getter, which is not worth specifying.
-
-## Multiple Outcomes
-
-Each distinct outcome deserves its own spec file. This keeps specs small, focused, and independently verifiable. When a spec fails, you immediately know *which* outcome broke — no debugging through a multi-assertion file.
-
-- When a behavior has multiple outcomes, create a `when_<behavior>/` folder with separate files for each outcome.
-- For simple behaviors with a single outcome, use a single file: `when_<behavior>`.
-- Never write a single file that tests an entire class.
-
-## Reusable Context
-
-Contexts capture the "given" part of a specification — the world as it exists before the action under test. They prevent duplicating setup across specs while keeping each spec readable.
-
-- Place in `given/` folder within the unit folder.
-- Name with `a_` or `an_` prefix (e.g. `an_observer`, `a_command_pipeline`). This reads naturally: "given an observer, when handling..."
-- More specific contexts can use descriptive names (e.g. `two_queries`, `existing_query`).
-- Context properties must be accessible to the specs that use them — see language-specific instructions for the exact access modifiers and naming conventions.
-- Use the setup phase for context initialization — **never** put the action under test in a reusable context. The action under test belongs only in the concrete spec.
-- Contexts can build on each other in layers: `all_dependencies → a_reactor_handler → when_handling`.
-- Consider creating `all_dependencies` as a root context that mocks all common dependencies. This avoids duplicating mock creation across unrelated specs.
-
-## Formatting
-
-- Keep assertions concise — prefer single-line assertions where the logic is readable.
-- Don't add blank lines between related assertions for the same behavior.
diff --git a/.claude/rules/specs.md b/.claude/rules/specs.md
new file mode 120000
index 0000000..cbfa6a9
--- /dev/null
+++ b/.claude/rules/specs.md
@@ -0,0 +1 @@
+../../.ai/rules/specs.md
\ No newline at end of file
diff --git a/.claude/rules/specs.typescript.md b/.claude/rules/specs.typescript.md
deleted file mode 100644
index b62a1bd..0000000
--- a/.claude/rules/specs.typescript.md
+++ /dev/null
@@ -1,134 +0,0 @@
----
-applyTo: "**/for_*/**/*.ts, **/when_*/**/*.ts"
----
-
-# How to Write TypeScript Specs
-
-Extends the base [Specs Instructions](./specs.instructions.md) with TypeScript-specific conventions.
-
-TypeScript specs follow the same BDD philosophy as C# specs — they describe behaviors, not implementations. The `given()` helper and context classes mirror the Cratis.Specifications pattern on the C# side: setup is separated from the action, and each `it()` assertion verifies a single outcome.
-
-## Frameworks
-
-- [Vitest](https://vitest.dev/) for running tests.
-- [Mocha](https://mochajs.org) for test structure (`describe`, `it`, `beforeEach`).
-- [SinonJS](https://sinonjs.org) for mocking/stubbing.
-- [Chai](https://www.chaijs.com) for assertions — **always use the `.should` fluent interface**, never `expect()`. The fluent style reads as a natural sentence: `result.should.equal(expected)`.
-- Run tests with `yarn test` from each package.
-
-## File Structure
-
-Tests live alongside source code in `for_`, `when_`, or `given_` folders:
-
-```
-for_EventsCommandResponseValueHandler/
-├── given/
-│   └── an_events_command_response_value_handler.ts
-├── when_checking_can_handle/
-│   ├── with_valid_events_collection.ts
-│   ├── with_null_value.ts
-│   └── without_event_source_id.ts
-└── when_handling/
-    ├── empty_events_collection.ts
-    └── multiple_events_collection.ts
-```
-
-## BDD Pattern with `given()` Helper
-
-The `given()` function is the TypeScript equivalent of the C# `Specification` base class. It instantiates a context class (the "given"), passes it to the test suite, and ensures setup runs before assertions. This keeps the Establish/Because/should pattern consistent across both stacks.
-
-```typescript
-import { an_events_command_response_value_handler } from '../given/an_events_command_response_value_handler';
-import { given } from '../../../given';
-
-describe('when checking can handle with valid events collection', given(an_events_command_response_value_handler, context => {
-    let result: boolean;
-
-    beforeEach(() => {
-        result = context.handler.canHandle(context.commandContext, [new TestEvent('Test')]);
-    });
-
-    it('should return true', () => {
-        result.should.be.true;
-    });
-}));
-```
-
-For behaviors with multiple outcomes, include "when \<behavior\>" as a prefix in the `describe` text.
-
-## Reusable Context Classes
-
-Context classes play the same role as `given/` classes in C# — they capture preconditions that multiple specs share. Unlike C# (where fields are `protected`), TypeScript context properties are public because tests access them directly through the `context` parameter.
-
-```typescript
-export class an_events_command_response_value_handler {
-    handler: EventsCommandResponseValueHandler;
-    commandContext: CommandContext;
-
-    constructor() {
-        this.commandContext = /* setup */;
-        this.handler = new EventsCommandResponseValueHandler(/* deps */);
-    }
-}
-```
-
-- Properties are public (not protected) — tests access them via `context.propertyName`.
-- Import `given` from the package root: `import { given } from '../../given';`.
-- Simple tests without shared setup don't need a reusable context.
-
-## Simple Test Pattern (without context)
-
-```typescript
-describe('when replacing route parameters', () => {
-    let result: { route: string; unusedParameters: object };
-
-    beforeEach(() => {
-        result = UrlHelpers.replaceRouteParameters('/api/items/{id}', { id: '123' });
-    });
-
-    it('should replace the route parameter', () => {
-        result.route.should.equal('/api/items/123');
-    });
-});
-```
-
-## Naming Conventions
-
-TypeScript specs use spaces in `it()` descriptions (unlike C# which uses underscores) because they appear in test runner output as human-readable sentences.
-
-- Use **spaces** (not underscores) in `it()` descriptions:
-  - ✅ `it('should return invalid result', ...)`
-  - ❌ `it('should_return_invalid_result', ...)`
-- Start `it()` descriptions with "should".
-- `describe()` text describes the scenario in natural language.
-
-## Assertions — Chai Fluent Interface
-
-**Always use the `.should` fluent interface. Never use `expect()`.** The `.should` style reads as a natural English sentence — `value.should.equal(expected)` vs `expect(value).to.equal(expected)` — and matches the project's preference for code that reads like prose.
-
-```typescript
-value.should.equal(expected);
-value.should.be.true;
-value.should.be.false;
-value.should.be.null;
-value.should.not.be.null;
-value.should.deep.equal(expected);
-value.should.be.instanceOf(Type);
-array.should.contain(item);
-array.should.have.lengthOf(3);
-(() => throwingFn()).should.throw(ErrorType);
-```
-
-## Mocking with Sinon
-
-```typescript
-import sinon from 'sinon';
-
-const stub = sinon.createStubInstance(ConcreteClass);
-const fetchStub = sinon.stub(globalThis, 'fetch');
-fetchStub.resolves({ ok: true, json: async () => ({ /* data */ }) });
-```
-
-## Async
-
-`beforeEach`, `afterEach`, and `it` callbacks can all be `async` when needed.
diff --git a/.claude/rules/specs.typescript.md b/.claude/rules/specs.typescript.md
new file mode 120000
index 0000000..aacfe3c
--- /dev/null
+++ b/.claude/rules/specs.typescript.md
@@ -0,0 +1 @@
+../../.ai/rules/specs.typescript.md
\ No newline at end of file
diff --git a/.claude/rules/terminal-commands.md b/.claude/rules/terminal-commands.md
deleted file mode 100644
index 69bfd7a..0000000
--- a/.claude/rules/terminal-commands.md
+++ /dev/null
@@ -1,17 +0,0 @@
----
----
-
-# RTK (Rust Token Killer) - Token-Optimized Commands
-
-## Golden Rule
-
-**Always prefix commands with `rtk`**. If RTK has a dedicated filter, it uses it. If not, it passes through unchanged. This means RTK is always safe to use.
-
-**Important**: Even in command chains with `&&`, use `rtk`:
-```bash
-# ❌ Wrong
-git add . && git commit -m "msg" && git push
-
-# ✅ Correct
-rtk git add . && rtk git commit -m "msg" && rtk git push
-```
diff --git a/.claude/rules/terminal-commands.md b/.claude/rules/terminal-commands.md
new file mode 120000
index 0000000..850b224
--- /dev/null
+++ b/.claude/rules/terminal-commands.md
@@ -0,0 +1 @@
+../../.ai/rules/terminal-commands.md
\ No newline at end of file
diff --git a/.claude/rules/typescript.md b/.claude/rules/typescript.md
deleted file mode 100644
index f0533c8..0000000
--- a/.claude/rules/typescript.md
+++ /dev/null
@@ -1,208 +0,0 @@
----
-applyTo: "**/*.ts,**/*.tsx"
-paths:
-  - "**/*.ts"
-  - "**/*.tsx"
----
-
-
-# TypeScript Conventions
-
-TypeScript's type system is the primary tool for catching bugs before they reach production. Every rule here pushes toward maximum compiler coverage and self-documenting code. If the types are right, the code almost writes itself.
-
-## Enums over Magic Strings
-
-String literal unions look concise but provide no refactoring support, no namespace, and no discoverability. Enums give you all three — plus `switch` exhaustiveness checking.
-
-```ts
-// ✅ Correct — refactorable, discoverable, exhaustive
-export enum SliceType {
-    StateChange = 'stateChange',
-    StateView   = 'stateView',
-    Automation  = 'automation',
-    Translator  = 'translator',
-}
-
-// ❌ Wrong — no refactoring support, invisible to tooling
-export type SliceType = 'stateChange' | 'stateView' | 'automation' | 'translator';
-```
-
-- Use enum members everywhere — `switch` cases, comparisons, defaults.
-- Do **not** import enums as `type`; they are values.
-- Export enums from `index.ts` without the `type` keyword.
-
-## One Type or Enum per File
-
-Each type gets its own file because it makes the codebase navigable — finding `SliceType` means opening `SliceType.ts`, not hunting through `types.ts`. It also keeps diffs clean and makes imports explicit.
-
-- Every interface, type alias, and enum lives in **its own file**, named after the type (e.g. `SliceType.ts`).
-- **Never create** `types.ts`, `models.ts`, `interfaces.ts` grab-bag files — they become dumping grounds that grow without limit.
-- Exception: component props interfaces (`*Props`) may live alongside their component `.tsx` file since they are tightly coupled to that component.
-- Aggregate exports through the folder's `index.ts`.
-
-## Type Safety
-
-`any` disables the compiler — the one tool that catches bugs for free. Every `any` is a hole in the safety net. Use `unknown` and narrow with type guards instead.
-
-- Never use `any` — use `unknown`, `Record<string, unknown>`, or proper generic constraints.
-- Prefer `value as unknown as TargetType` over `value as any`.
-- Use `unknown` as default generic parameter instead of `any`.
-- React synthetic events (`React.MouseEvent<Element, MouseEvent>`) and DOM events (`MouseEvent`) are different types — don't mix them.
-
-## Localised Strings
-
-All user-visible text **must** come from translation files — never hard-code UI strings directly in component code. This is not optional even for English-only deployments — it centralizes copy, enables future localization, and makes it possible to audit all user-facing text in one place.
-
-### Structure
-
-```
-Source/Core/
-  Strings.ts                    ← re-exports the default (English) JSON
-  Locales/
-    en/
-      translation.json          ← all English strings, organized by feature/component
-```
-
-`translation.json` is a nested JSON object whose top-level keys group strings by feature or component (e.g. `projects`, `eventModeling`, `chat`). Add new keys under the appropriate existing group, or add a new top-level group if the feature is new.
-
-### Importing
-
-Import the strings object using the `Strings` path alias (configured in `tsconfig.json`):
-
-```ts
-import strings from 'Strings';
-```
-
-Do **not** use relative paths such as `'../../Strings'` or `'../Strings'`.
-
-### Usage
-
-Access strings directly through the nested object — TypeScript infers the full type from the JSON file:
-
-```tsx
-import strings from 'Strings';
-
-export const MyComponent = () => (
-    <Button label={strings.projects.addProject} />
-);
-```
-
-For attribute strings (e.g. `title`, `placeholder`, `aria-label`):
-
-```tsx
-<div title={strings.eventModeling.grid.detailPanel.event}>
-    ...
-</div>
-```
-
-### Adding New Strings
-
-1. Add the key to `Source/Core/Locales/en/translation.json` under the appropriate group.
-2. TypeScript picks up the new key automatically from `Strings.ts` (no regeneration step).
-3. Use the key via `strings.<group>.<key>` in the component.
-
-### Rules
-
-- **Never** use plain string literals for user-visible text in JSX or attribute props. This includes `label`, `header`, `placeholder`, `title`, `aria-label`, `emptyMessage`, and any visible text nodes.
-- Only constant, non-localised values are allowed as raw strings (CSS class names, `key` props, internal identifiers).
-
-## Arc Frontend Patterns
-
-Arc's proxy generator bridges C# and TypeScript automatically — every `[Command]` and `[ReadModel]` becomes a TypeScript class with `.use()` hooks, `.execute()` methods, and change tracking. This is the foundation of full-stack type safety: change a C# record and the TypeScript proxy updates on the next `dotnet build`.
-
-### Commands
-
-Auto-generated from C# `[Command]` records. The `.use()` hook returns a tuple: the command instance (with change tracking) and a setter for property values.
-
-```tsx
-const [command, setValues] = OpenAccount.use({ name: '', owner: '' });
-await command.execute();       // Sends command to backend, returns CommandResult
-await command.validate();      // Pre-flight validation only, no side effects
-command.hasChanges;            // True when any property differs from initial values
-command.revertChanges();       // Reset all properties to initial values
-```
-
-### Queries
-
-Auto-generated from C# `[ReadModel]` static query methods. Observable queries (returning `ISubject<T>` on the backend) auto-subscribe via WebSocket — the component re-renders when data changes on the server.
-
-```tsx
-const [result, perform] = AllProjects.use();
-// result.data — the query result
-// result.isPerforming — true while loading
-// result.hasData — true when data has arrived
-// result.isSuccess — true when query completed without errors
-```
-
-Paginated queries:
-```tsx
-const [result, , setPage] = AllProjects.useWithPaging(10);
-```
-
-### CommandScope
-
-Wraps multiple command-using components, aggregating their `hasChanges` state and enabling bulk `execute()` and `revertChanges()`. Useful for forms that span multiple components.
-
-### CommandForm
-
-Declarative form component with built-in field types, validation timing (`validateOn: blur|change|both`), and automatic server-side validation feedback.
-
-## Language — American English Only
-
-All identifiers, comments, JSDoc, and string literals must use **American English** spelling. This applies to variable names, function names, type names, enum members, and documentation. Common mistakes to avoid:
-
-| Wrong (UK) | Correct (US) |
-|---|---|
-| initialise, serialise, normalise | initialize, serialize, normalize |
-| behaviour, colour, favour, honour | behavior, color, favor, honor |
-| organisation, authorisation | organization, authorization |
-| centre, fibre | center, fiber |
-| modelling, signalling, cancelling | modeling, signaling, canceling |
-| dialogue, catalogue | dialog, catalog |
-| licence, defence | license, defense |
-| judgement, acknowledgement | judgment, acknowledgment |
-| grey | gray |
-
-## Variables and Naming
-
-- Prefer `const` over `let` over `var` when declaring variables.
-- Never use shortened or abbreviated names for variables, parameters, or properties.
-  - Use full descriptive names: `deltaX` not `dx`, `index` not `idx`, `event` not `e`, `previous` not `prev`, `direction` not `dir`, `position` not `pos`, `contextMenu` not `ctx`/`ctxMenu`.
-  - The only acceptable short names are well-established domain terms (e.g. `id`, `url`, `min`, `max`).
-
-## Imports and Compilation
-
-- Never leave unused import statements in the code.
-- Always ensure that the code compiles without warnings — use `yarn compile` to verify (successful runs produce no output).
-- Review each file for lint compliance before finalizing.
-- Never use placeholder or temporary types — use proper types from the start.
-- **Never modify any file inside `node_modules/` or any build cache (e.g. `.vite/deps/`).** These are managed by the package manager and will be overwritten on the next install. If something appears broken in a library, look harder at the application code — especially when other usages of the same library work fine. Fixes belong in application code or upstream in the library's own repo.
-
-## Folder Structure
-
-- Do not prefix a file, component, type, or symbol with the name of its containing folder or the concept it belongs to. Instead, use folder structure to provide that context.
-- Favor functional folder structure over technical folder structure.
-  - Group files by the feature or concept they belong to, not by their technical role.
-  - Avoid folders like `components/`, `hooks/`, `utils/`, `types/` at the feature level.
-
-## Advanced Type Safety
-
-Additional patterns for common tricky scenarios:
-
-**Storybook:**
-- Use `React.ComponentType<Record<string, never>>` for components with no props.
-- Always use `as unknown as` when converting component imports to avoid type mismatch errors.
-- Properly type story args — never use `any`.
-
-**External libraries with strict generic constraints:**
-- Import necessary types (e.g. `Command` from `@cratis/arc/commands`) rather than asserting to `any`.
-- Use type assertions through `unknown`: `props.command as unknown as Constructor<Command<...>>`.
-- Extract tuple results explicitly rather than destructuring when type assertions are needed.
-- Use proper library types when available; use specific property types (e.g. `{ canvas?: HTMLCanvasElement }`) over `any`.
-
-**Dynamic and generic types:**
-- Add type guards for unknown function parameters: `if (typeof accessor !== 'function') return ''`.
-- Type parameters with fallbacks: `function<T = unknown>(accessor: ((obj: T) => unknown) | unknown)`.
-- Cast arrays from `unknown` explicitly: `((obj as Record<string, unknown>).items || []) as string[]`.
-- Use `String(value)` for string conversions in generic contexts.
-- Use explicit Date parameter types: `new Date(value as string | number | Date)`.
diff --git a/.claude/rules/typescript.md b/.claude/rules/typescript.md
new file mode 120000
index 0000000..a8aad8f
--- /dev/null
+++ b/.claude/rules/typescript.md
@@ -0,0 +1 @@
+../../.ai/rules/typescript.md
\ No newline at end of file
diff --git a/.claude/rules/vertical-slices.md b/.claude/rules/vertical-slices.md
deleted file mode 100644
index 8cb4ef2..0000000
--- a/.claude/rules/vertical-slices.md
+++ /dev/null
@@ -1,177 +0,0 @@
----
-applyTo: "**/Features/**/*.*"
----
-
-# Vertical Slice Architecture
-
-A vertical slice contains *everything* for a single behavior: the command or query, the events it produces, the projections that build read models, the React component that renders the UI, and the specs that verify it all works. Everything lives together in one folder because everything changes together.
-
-## Technical Stack
-
-- .NET with C# 13 (ASP.NET Core) — Cratis Arc for CQRS, Cratis Chronicle for event sourcing, MongoDB for read models
-- React + TypeScript (Vite) — PrimeReact UI, Vitest + Mocha/Chai/Sinon for specs
-- xUnit + Cratis.Specifications + NSubstitute for C# specs
-
-## Core Rules
-
-These are non-negotiable because the frameworks rely on them for convention-based discovery and proxy generation:
-
-- **Each vertical slice has its own folder with a single `.cs` file containing ALL backend artifacts.**
-- **Commands define `Handle()` directly on the record — never create separate handler classes.**
-- **`[EventType]` must have NO arguments — the type name is used automatically.**
-- Complete one slice end-to-end before starting the next.
-- Drop the `.Features` segment from namespaces (e.g. `MyApp.Projects.Registration` not `MyApp.Features.Projects.Registration`).
-
----
-
-## Proxy Generation — Build Dependency
-
-`dotnet build` generates TypeScript proxies. Until the backend compiles, **no proxy files exist** and frontend code cannot reference them.
-
-**Sequencing constraint:** Backend → `dotnet build` → Frontend. Backend and frontend for the same slice **cannot** run in parallel.
-
----
-
-## Slice Types
-
-| Type | Purpose | What It Contains |
-|---|---|---|
-| **State Change** | Mutates system state | Command + events + validators/constraints |
-| **State View** | Projects events into queryable read models | Read model + projection + queries |
-| **Automation** | Reacts to events, makes decisions | Reactor + local read models |
-| **Translation** | Adapts events across slices/systems | Reactor → triggers commands in own slice |
-
----
-
-## Folder Structure
-
-```
-Features/
-├── <Feature>/
-│   ├── <Feature>.tsx              ← composition page (layout + menu)
-│   ├── <Concept>.cs               ← shared concepts for this feature
-│   ├── <SliceName>/
-│   │   ├── <SliceName>.cs         ← ALL backend artifacts in ONE file
-│   │   ├── <Component>.tsx        ← React component(s) for this slice
-│   │   └── when_<behavior>/       ← integration specs (state-change slices)
-│   │       ├── and_<scenario>.cs
-│   │       └── ...
-│   └── ...
-```
-
-**✅ CORRECT:**
-```
-Features/Authors/
-├── Authors.tsx
-├── AuthorId.cs
-├── Registration/
-│   ├── Registration.cs            ← command + event + constraint
-│   ├── AddAuthor.tsx
-│   └── when_registering/
-│       └── and_name_already_exists.cs
-└── Listing/
-    ├── Listing.cs                 ← read model + projection + query
-    └── Listing.tsx
-```
-
-**❌ WRONG — Never split by artifact type:**
-```
-Features/Authors/
-├── Commands/RegisterAuthor.cs
-├── Handlers/RegisterAuthorHandler.cs
-└── Events/AuthorRegistered.cs
-```
-
----
-
-## What Goes in a Single Slice File
-
-A single `<SliceName>.cs` contains ALL of: `[Command]` records with `Handle()`, validators, constraints, `[EventType]` records, `[ReadModel]` records with static query methods, projections/reducers, reactors, and slice-specific concepts.
-
----
-
-## Events — Rules
-
-- `[EventType]` takes **no arguments** — the type name is the identifier.
-- Past tense naming: `AuthorRegistered`, `BookReserved`, `AddressChanged`.
-- Never nullable properties — if something is optional, you need a second event.
-- One purpose per event — never multipurpose events with many nullable fields.
-
-```csharp
-[EventType]
-public record AuthorRegistered(AuthorName Name);
-```
-
----
-
-## Commands — Rules
-
-- `[Command]` record from `Cratis.Arc.Commands.ModelBound`.
-- `Handle()` defined directly on the record — no separate handler class.
-- `Handle()` returns: single event, tuple `(event, result)`, `Result<TSuccess, TError>`, or `void`.
-- Event source resolved from: `[Key]` parameter → `EventSourceId`-convertible type → `ICanProvideEventSourceId`.
-- Business rules via DCB: accept a read model parameter in `Handle()` — the framework injects current state.
-
-→ For step-by-step command creation, invoke the **`cratis-command`** skill.
-
----
-
-## Read Models & Projections — Rules
-
-- Prefer model-bound attributes (`[ReadModel]`, `[FromEvent<T>]`, `[Key]`, etc.) over fluent `IProjectionFor<T>`.
-- AutoMap is **on by default** — call `.From<EventType>()` directly, no `.AutoMap()` call needed.
-- Projections join **events**, never read models.
-- Query methods are **static** methods on the `[ReadModel]` record.
-- Favor reactive queries (`ISubject<T>`) for real-time updates.
-
-→ For step-by-step read model creation, invoke the **`cratis-readmodel`** skill.
-→ For adding a projection to an existing model, invoke the **`add-projection`** skill.
-
----
-
-## Concepts — Rules
-
-- Prefer `ConceptAs<T>` over raw primitives everywhere in domain models, commands, events, and queries.
-- Concepts shared between slices → feature folder. Shared between features → `Features/` root. One file per concept.
-
-→ See [concepts.instructions.md](./concepts.instructions.md) for full rules.
-→ For step-by-step concept creation, invoke the **`add-concept`** skill.
-
----
-
-## Reactors — Rules
-
-- `IReactor` is a marker interface — method dispatch is by first-parameter event type.
-- Reactors observe events and produce side effects — never use `IEventLog` directly from a reactor.
-- If a reactor needs to write new events, execute a command via `ICommandPipeline`.
-- Design for idempotency — reactors may be called more than once.
-
-→ See [reactors.instructions.md](./reactors.instructions.md) for full rules.
-→ For step-by-step reactor creation, invoke the **`add-reactor`** skill.
-
----
-
-## Development Workflow
-
-Work **slice-by-slice** in this exact order:
-
-1. **Backend** — implement the C# slice file
-2. **Specs** — write integration specs for state-change slices
-3. **Build** — run `dotnet build` to generate TypeScript proxies
-4. **Frontend** — implement React component(s) using the generated proxies
-5. **Composition** — register in the feature's composition page
-6. **Routes** — add/update routing if needed
-
-→ For end-to-end slice implementation, invoke the **`new-vertical-slice`** skill.
-→ For creating a new feature folder, invoke the **`scaffold-feature`** skill.
-
----
-
-## Dialogs — Rules
-
-- **Never** import `Dialog` from `primereact/dialog` — use Cratis wrappers.
-- `CommandDialog` from `@cratis/components/CommandDialog` — for dialogs that execute commands.
-- `Dialog` from `@cratis/components/Dialogs` — for data collection without commands.
-
-→ See [dialogs.instructions.md](./dialogs.instructions.md) for full dialog patterns.
-
diff --git a/.claude/rules/vertical-slices.md b/.claude/rules/vertical-slices.md
new file mode 120000
index 0000000..a4d7515
--- /dev/null
+++ b/.claude/rules/vertical-slices.md
@@ -0,0 +1 @@
+../../.ai/rules/vertical-slices.md
\ No newline at end of file
diff --git a/.claude/rules/web-fetching.md b/.claude/rules/web-fetching.md
deleted file mode 100644
index 2691739..0000000
--- a/.claude/rules/web-fetching.md
+++ /dev/null
@@ -1,10 +0,0 @@
----
-description: "Use when fetching data from the web, CI logs, artifact URLs, signed URLs, Azure Blob URLs, or simple API/text responses. Prefer curl in the terminal over webpage fetch tools for raw data retrieval."
----
-
-# Web Fetching
-
-- Prefer `curl` in the terminal for raw remote content such as CI logs, artifact downloads, signed URLs, plain-text endpoints, and JSON APIs.
-- Use webpage/content fetch tools only when the goal is to summarize or inspect rendered page content rather than retrieve the exact response body.
-- For expiring, authenticated, or redirecting URLs, default to `curl -L -s` and then pipe to `head`, `grep`, `sed`, or `jq` as needed.
-- When debugging remote responses, keep the raw output in the terminal path and filter locally instead of depending on fetch tools.
diff --git a/.claude/rules/web-fetching.md b/.claude/rules/web-fetching.md
new file mode 120000
index 0000000..905bff9
--- /dev/null
+++ b/.claude/rules/web-fetching.md
@@ -0,0 +1 @@
+../../.ai/rules/web-fetching.md
\ No newline at end of file
diff --git a/.claude/skills b/.claude/skills
new file mode 120000
index 0000000..6838a11
--- /dev/null
+++ b/.claude/skills
@@ -0,0 +1 @@
+../.ai/skills
\ No newline at end of file
diff --git a/.github/agents b/.github/agents
new file mode 120000
index 0000000..2f6c069
--- /dev/null
+++ b/.github/agents
@@ -0,0 +1 @@
+../.ai/agents
\ No newline at end of file
diff --git a/.github/agents/backend-developer.md b/.github/agents/backend-developer.md
deleted file mode 100644
index 56095f0..0000000
--- a/.github/agents/backend-developer.md
+++ /dev/null
@@ -1,125 +0,0 @@
----
-name: Backend Developer
-description: >
-  Specialist for C# backend code within a vertical slice.
-  Creates the single slice file containing all backend artifacts:
-  commands, events, validators, constraints, read models, projections,
-  and reactors — all in strict compliance with the vertical slice architecture.
-model: claude-sonnet-4-5
-tools:
-  - githubRepo
-  - codeSearch
-  - usages
-  - rename
-  - terminalLastCommand
----
-
-# Backend Developer
-
-You are the **Backend Developer** for Cratis-based projects.
-Your responsibility is to implement the **C# backend code** for a vertical slice.
-
-Always read and follow:
-- `.github/instructions/vertical-slices.instructions.md`
-- `.github/instructions/csharp.instructions.md`
-- `.github/instructions/concepts.instructions.md`
-- `.github/instructions/efcore.instructions.md`
-- `.github/copilot-instructions.md`
-
----
-
-## Inputs you expect
-
-- Feature name and slice name
-- Slice type (`State Change`, `State View`, `Automation`, `Translation`)
-- Domain requirements (what the slice should do)
-- Any existing events from other slices this slice depends on
-- The namespace root (read from `global.json` or existing source files, e.g. `Studio`, `Library`)
-
----
-
-## Process
-
-1. **Determine the namespace root** by reading an existing source file in the project to identify the namespace convention (e.g. `Studio`, `Library`, `MyApp`).
-2. **Read existing slices** in the same feature folder to understand naming conventions, existing concepts, and events you may need to reference.
-3. **Create a single `.cs` file** at `Features/<Feature>/<Slice>/<Slice>.cs`.
-4. **Validate** by running `dotnet build` from the repository root.
-5. Fix all compiler errors and warnings before handing back.
-
----
-
-## File structure rules (mandatory)
-
-- **One file per slice** — all artifacts in `<Slice>.cs`.
-- File header:
-  ```csharp
-  // Copyright (c) Cratis. All rights reserved.
-  // Licensed under the MIT license. See LICENSE file in the project root for full license information.
-  ```
-- Namespace: `<NamespaceRoot>.<Feature>.<Slice>` (no `.Features.` in the namespace).
-- Order of declarations in the file:
-  1. Concepts (if slice-specific)
-  2. Commands (with `Handle()` inline)
-  3. Validators
-  4. Business rules
-  5. Constraints
-  6. Events
-  7. Read models + query methods
-  8. Projections
-  9. Reactors
-
----
-
-## Commands — critical rules
-
-- Record decorated with `[Command]` from `Cratis.Arc.Commands.ModelBound`.
-- **`Handle()` defined directly on the record** — no separate handler class.
-- Return from `Handle()`: single event, `IEnumerable<event>`, tuple `(event, result)`, or `Result<,>`.
-- Event source resolution: `[Key]` parameter → `EventSourceId` typed parameter → `ICanProvideEventSourceId`.
-
-```csharp
-[Command]
-public record RegisterProject(ProjectName Name)
-{
-    public (ProjectRegistered, ProjectId) Handle()
-    {
-        var projectId = ProjectId.New();
-        return (new ProjectRegistered(Name), projectId);
-    }
-}
-```
-
----
-
-## Events — critical rules
-
-- Record decorated with `[EventType]` from `Cratis.Events`.
-- **`[EventType]` has NO arguments** — the type name is the identifier.
-
-```csharp
-[EventType]
-public record ProjectRegistered(ProjectName Name);
-```
-
----
-
-## Read models & projections — critical rules
-
-- Record decorated with `[ReadModel]` from `Cratis.Arc.Queries.ModelBound`.
-- Query methods are **static** methods on the record.
-- Always call `.AutoMap()` before any `.From<>()`.
-- Projections join **events**, never read models.
-
----
-
-## Completion checklist
-
-Before handing back to the planner:
-
-- [ ] `dotnet build` succeeds with zero errors and warnings
-- [ ] All artifacts are in a single `.cs` file
-- [ ] Namespace follows `<NamespaceRoot>.<Feature>.<Slice>` (no `.Features.`)
-- [ ] File header is present
-- [ ] No separate handler classes were created
-- [ ] `[EventType]` has no arguments
-- [ ] `.AutoMap()` is used before `.From<>()` in all projections
diff --git a/.github/agents/code-reviewer.md b/.github/agents/code-reviewer.md
deleted file mode 100644
index e06cdba..0000000
--- a/.github/agents/code-reviewer.md
+++ /dev/null
@@ -1,164 +0,0 @@
----
-name: Code Reviewer
-description: >
-  Quality gate agent for Cratis-based projects. Reviews code against all
-  project instruction files, checking architecture conformance, C# and
-  TypeScript conventions, and vertical slice correctness before merge.
-model: claude-sonnet-4-5
-tools:
-  - githubRepo
-  - codeSearch
-  - usages
-  - rename
-  - terminalLastCommand
----
-
-# Code Reviewer
-
-You are the **Code Reviewer** for Cratis-based projects.
-Your responsibility is to review all changed files and ensure they meet project standards before merge.
-
-Always check against:
-- `.github/copilot-instructions.md`
-- `.github/instructions/vertical-slices.instructions.md`
-- `.github/instructions/csharp.instructions.md`
-- `.github/instructions/specs.csharp.instructions.md`
-- `.github/instructions/specs.typescript.instructions.md`
-- `.github/instructions/typescript.instructions.md`
-- `.github/instructions/components.instructions.md`
-- `.github/instructions/dialogs.instructions.md`
-- `.github/instructions/concepts.instructions.md`
-- `.github/instructions/efcore.specs.instructions.md`
-
----
-
-## Review approach
-
-Review every changed file. For each issue found:
-- State the **file and line number**
-- Quote the **problematic code**
-- Explain **why it violates the standard**
-- Provide the **corrected code**
-
-When checking for unused code, missing references, or naming consistency, prefer the **`usages`** tool over grep — it uses LSP for precise, language-aware results. Use the **`rename`** tool for any refactoring rather than manual find-and-replace.
-
----
-
-## C# Architecture checklist
-
-- [ ] Each slice lives in its own file under `Features/<Feature>/<Slice>.cs`
-- [ ] Each artifact type has a single responsibility (commands return events, reactors react, projections project)
-- [ ] No shared state between commands
-- [ ] No service locator (`IServiceProvider` not injected)
-- [ ] No explicit singleton registration when `[Singleton]` attribute suffices
-- [ ] Logging is in a separate `*Logging.cs` partial file with `[LoggerMessage]`
-
-## C# Commands checklist
-
-- [ ] `record` type, not `class`
-- [ ] No properties with setters (immutable)
-- [ ] `Handle()` method is the single entry point
-- [ ] `Handle()` **returns** the event(s) — never calls `IEventLog` directly
-- [ ] Namespace matches folder path: `<NamespaceRoot>.<Feature>.<Slice>`
-
-## C# Read Models & Projections checklist
-
-- [ ] Read model is a `record` type with all required props
-- [ ] Preferred: projection uses model-bound attributes (`[FromEvent<T>]`, `[Key]`, etc.) directly on the read model — no separate projection class needed
-- [ ] If using fluent `IProjectionFor<T>`: `.AutoMap()` MUST appear before any `.From<>()` call
-- [ ] Projection does NOT join on the read model — joins are on Chronicle events only
-- [ ] No `ToList()`, `ToArray()`, or mutation of public-API collection returns
-
-## C# Concepts checklist
-
-- [ ] Strongly typed IDs use `ConceptAs<T>` pattern (see `concepts.instructions.md`)
-- [ ] No raw `Guid`, `string`, etc. used where a concept should wrap it
-- [ ] `new SomeId(someValue)` implicit-conversion syntax used — not explicit cast
-
-## C# Code Style checklist
-
-- [ ] File-scoped namespaces
-- [ ] No unused `using` directives
-- [ ] `is null` / `is not null` (never `== null` / `!= null`)
-- [ ] `var` preferred over explicit type declarations
-- [ ] No postfixes: `Async`, `Impl`, `Service` on class names
-- [ ] No regions
-- [ ] Copyright header present on every file
-- [ ] All public types, methods, and properties have multiline XML doc comments
-- [ ] `<summary>` tags are always multiline — never `/// <summary>Text</summary>` on one line
-- [ ] Methods with parameters have `<param name="...">` for each parameter
-- [ ] Non-void methods have `<returns>` documentation
-- [ ] Custom exception types only (no `InvalidOperationException`, `ArgumentException`, etc.)
-- [ ] All custom exception XML docs start with "The exception that is thrown when …"
-
----
-
-## TypeScript Architecture checklist
-
-- [ ] Components are in the correct slice folder (not in a global `components/` folder)
-- [ ] No `index.ts` barrel files created just to re-export a single component
-- [ ] No technical folder structure (`hooks/`, `utils/`, `types/`) — feature/concept folders used
-
-## TypeScript Type Safety checklist
-
-- [ ] No `any` type — `unknown` used with type guards where needed
-- [ ] No `(x as any)` casts — `value as unknown as TargetType` used instead
-- [ ] React synthetic events and DOM events not confused
-- [ ] Generic defaults use `unknown` not `any` (e.g. `<T = unknown>`)
-
-## TypeScript Styling checklist
-
-- [ ] No hard-coded hex/rgb values — PrimeReact CSS variables used
-- [ ] CSS co-located with component (`.css` file in same folder)
-- [ ] No `!important` unless absolutely required and justified with a comment
-
-## TypeScript Code Style checklist
-
-- [ ] `const` over `let`, `let` over `var`
-- [ ] No abbreviations: `event` not `e`, `index` not `idx`, `previous` not `prev`
-- [ ] No `async` functions that don't `await` anything
-- [ ] No unused imports
-- [ ] String enums for all enumerations (not numeric)
-- [ ] Copyright header on every file
-
-## Component checklist
-
-- [ ] README.md exists for complex component folders
-- [ ] `CommandDialog` from `@cratis/components/CommandDialog` used for command-based dialogs
-- [ ] `Dialog` from `@cratis/components/Dialogs` used for data-only dialogs
-- [ ] Never imports `Dialog` directly from `primereact/dialog`
-- [ ] No monolithic components — decomposed into smaller, focused sub-components
-
----
-
-## Specs checklist
-
-- [ ] Every state-change command has specs
-- [ ] Happy path covered
-- [ ] All validation rules covered
-- [ ] All constraint violations covered
-- [ ] No specs for simple property getters or constructor pass-throughs
-- [ ] Chai fluent interface used in TypeScript specs (not `expect()`)
-
----
-
-## Output format
-
-Start with a **summary**:
-> **Review result: ✅ Approved / ⚠️ Approved with comments / ❌ Changes requested**
-
-Then list issues grouped by file:
-
-```
-### <file path>
-
-**[BLOCKING]** … or **[SUGGESTION]** …
-> Line N: `problematic code`
-> Because: explanation
-> Fix:
-> ```
-> corrected code
-> ```
-```
-
-End with a checklist of passed / failed items so the developer knows what was verified.
diff --git a/.github/agents/coordinator.md b/.github/agents/coordinator.md
deleted file mode 100644
index 334a82d..0000000
--- a/.github/agents/coordinator.md
+++ /dev/null
@@ -1,148 +0,0 @@
----
-name: Coordinator
-description: >
-  General-purpose coordinator agent for Cratis-based projects.
-  Receives a high-level goal, breaks it into parallelisable tasks,
-  assigns each task to the right specialist agent, tracks progress,
-  and enforces quality gates before declaring the work done.
-  Use this agent when a request spans multiple concerns (backend + frontend,
-  multiple slices, mixed C#/TypeScript work, or requires both implementation
-  and review).
-model: claude-sonnet-4-5
-tools:
-  - githubRepo
-  - codeSearch
-  - usages
-  - terminalLastCommand
----
-
-# Coordinator
-
-You are the **Coordinator** for Cratis-based projects.
-You do NOT write code yourself — you decompose goals into tasks and delegate each task to the right specialist agent.
-
-Always read and follow:
-- `.github/copilot-instructions.md`
-- `.github/instructions/vertical-slices.instructions.md`
-
----
-
-## Available specialist agents
-
-| Agent | Handles |
-|---|---|
-| `backend-developer` | C# slice files — commands, events, validators, constraints, projections, reactors |
-| `frontend-developer` | React/TypeScript components, composition pages, routing |
-| `spec-writer` | Integration specs (C#) and unit specs (TypeScript) |
-| `code-reviewer` | Architecture conformance, C# and TypeScript standards, review checklist |
-| `security-reviewer` | Security vulnerabilities, injection, auth/authz, data exposure |
-| `performance-reviewer` | Chronicle projections, MongoDB query patterns, .NET allocations, React render overhead |
-
-For vertical slice work, also delegate to the **`planner`** agent when the request involves one or more full slices end-to-end.
-
----
-
-## Decomposition process
-
-When you receive a goal:
-
-1. **Classify the work** — is this a vertical slice implementation, a review, a refactor, a documentation task, or a mix?
-2. **Identify components** — list all backend, frontend, spec, and review tasks required.
-3. **Identify dependencies** — which tasks block which? (e.g. backend must finish before frontend).
-4. **Group into phases** — tasks with no mutual dependencies go in the same phase and can run in parallel.
-5. **Assign agents** — pick the right specialist for each task.
-6. **Output a plan** — always as a markdown checklist with agent assignments.
-
----
-
-## Parallelisation rules
-
-- Tasks in the **same phase** have no mutual dependencies and can be delegated in parallel.
-- **Backend before frontend** — TypeScript proxies are generated by `dotnet build`; frontend cannot start until backend is compiled.
-- **Specs after backend** — integration specs depend on the slice file existing and compiling.
-- **Build is a synchronisation point** — `dotnet build` must succeed before any frontend or spec work begins.
-- **Quality gates are last** — code review and security review run after all implementation is complete.
-- **Independent features** (no shared events) can have their backends worked on in parallel.
-
----
-
-## Plan template
-
-```markdown
-## Coordinator Plan: <goal summary>
-
-### Phase 1 — <description> [can run in parallel]
-- [ ] [<agent>] <task description>
-- [ ] [<agent>] <task description>
-
-### Phase 2 — <description> (depends on Phase 1)
-- [ ] [<agent>] <task description>
-
-### Phase 3 — Build
-- [ ] Run `dotnet build` — must succeed before any Phase 4 work
-
-### Phase 4 — <description> [can run in parallel]
-- [ ] [<agent>] <task description>
-
-### Phase 5 — Quality Gates
-- [ ] [code-reviewer] Review all changed files
-- [ ] [security-reviewer] Security review of all changed files
-```
-
----
-
-## Delegation instructions
-
-When handing off to a specialist agent:
-
-1. State **exactly which files** need to be created or modified.
-2. Provide **all context** the agent needs — feature name, slice name, slice type, existing events, namespace root.
-3. State **acceptance criteria** — what "done" looks like for this task.
-4. Tell the specialist **which agent to hand back to** when finished.
-5. Quote the **relevant instruction file** section that governs the work.
-
----
-
-## Quality gate criteria
-
-The work is **not done** until all of the following pass:
-
-- [ ] `dotnet build` — zero errors, zero warnings
-- [ ] `dotnet test` — all specs pass
-- [ ] `yarn lint` — zero errors (if frontend present)
-- [ ] `npx tsc -b` — zero TypeScript errors (if frontend present)
-- [ ] `code-reviewer` finds no blocking issues
-- [ ] `security-reviewer` finds no vulnerabilities
-- [ ] PR description follows the pull request template
-
----
-
-## When to delegate to the planner instead
-
-If the entire goal is one or more vertical slices (full backend-to-frontend), delegate directly to the **`planner`** agent rather than coordinating slice phases yourself. The planner is optimised for slice sequencing. Use the Coordinator for cross-cutting work that involves concerns beyond a single slice pipeline (e.g. shared infrastructure changes + slice implementation, documentation updates + implementation, multi-feature refactors).
-
----
-
-## Output format
-
-Always output a plan before starting any delegation:
-
-```markdown
-## Coordinator Plan: <goal>
-
-### Phase 1 — Backend [parallel]
-- [ ] [backend-developer] <task>
-
-### Phase 2 — Build
-- [ ] `dotnet build`
-
-### Phase 3 — Frontend + Specs [parallel]
-- [ ] [frontend-developer] <task>
-- [ ] [spec-writer] <task>
-
-### Phase 4 — Quality Gates
-- [ ] [code-reviewer] Review all changed files
-- [ ] [security-reviewer] Security review
-```
-
-Then delegate each task in order, respecting phase boundaries.
diff --git a/.github/agents/frontend-developer.md b/.github/agents/frontend-developer.md
deleted file mode 100644
index d805ea1..0000000
--- a/.github/agents/frontend-developer.md
+++ /dev/null
@@ -1,242 +0,0 @@
----
-name: Frontend Developer
-description: >
-  Specialist for TypeScript/React frontend code within a vertical slice.
-  Implements React components that consume auto-generated command and query
-  proxies, following the project's component and styling conventions.
-model: claude-sonnet-4-5
-tools:
-  - githubRepo
-  - codeSearch
-  - usages
-  - rename
-  - terminalLastCommand
----
-
-# Frontend Developer
-
-You are the **Frontend Developer** for Cratis-based projects.
-Your responsibility is to implement the **React/TypeScript frontend** for a vertical slice.
-
-Always read and follow:
-- `.github/instructions/vertical-slices.instructions.md`
-- `.github/instructions/components.instructions.md`
-- `.github/instructions/dialogs.instructions.md`
-- `.github/instructions/typescript.instructions.md`
-- `.github/copilot-instructions.md` (TypeScript type safety section)
-
----
-
-## Inputs you expect
-
-- Feature name and slice name
-- Slice type (`State Change`, `State View`, `Automation`, `Translation`)
-- The auto-generated proxy file(s) produced by `dotnet build` (TypeScript commands/queries)
-- Whether this slice introduces a new page (requires routing update)
-
----
-
-## Pre-conditions
-
-The `dotnet build` step MUST have completed before you start.
-Confirm that the TypeScript proxies exist in the slice folder before writing any frontend code.
-
----
-
-## Process
-
-1. **Read the existing feature composition page** (`Features/<Feature>/<Feature>.tsx`) to understand the current layout and imports.
-2. **Create component file(s)** in the slice folder (`Features/<Feature>/<Slice>/`).
-3. **Update the composition page** to import and use the new component.
-4. **Update routing** if the slice introduces a new page.
-5. **Validate** with `yarn lint` and `npx tsc -b`.
-
----
-
-## Component rules (mandatory)
-
-- Place `.tsx` files in the **same folder** as the corresponding `.cs` file.
-- Do NOT prefix the file name with the feature or slice name (folder provides context).
-- Each component has its own `.css` file for static styles.
-- Use PrimeReact CSS variables for all colours, backgrounds, and borders — never hard-code hex values.
-- Use `const` over `let`.
-- Use full descriptive names (never abbreviations like `e`, `idx`, `prev`).
-
----
-
-## Command usage pattern
-
-```tsx
-const [registerProject] = RegisterProject.use();
-
-const handleSubmit = async () => {
-    registerProject.name = name;
-    const result = await registerProject.execute();
-    if (result.isSuccess) {
-        closeDialog(DialogResult.Ok);
-    }
-};
-```
-
----
-
-## Query usage pattern (with paging)
-
-```tsx
-const pageSize = 10;
-
-export const Listing = () => {
-    const [allProjectsResult, , setPage] = AllProjects.useWithPaging(pageSize);
-
-    return (
-        <DataTable
-            lazy paginator
-            value={allProjectsResult.data}
-            rows={pageSize}
-            totalRecords={allProjectsResult.paging.totalItems}
-            alwaysShowPaginator={false}
-            first={allProjectsResult.paging.page * pageSize}
-            onPage={event => setPage(event.page ?? 0)}
-            scrollable scrollHeight="flex"
-            emptyMessage="No items found.">
-            <Column field="name" header="Name" />
-        </DataTable>
-    );
-};
-```
-
----
-
-## Dialog patterns
-
-Use this whenever the dialog executes a Cratis Arc command on confirm. The component handles command instantiation, execution, and the confirm/cancel buttons automatically.
-
-### Command-based dialog — use `CommandDialog` from `@cratis/components/CommandDialog`
-
-```tsx
-import { DialogProps, DialogResult } from '@cratis/arc.react/dialogs';
-import { CommandDialog } from '@cratis/components/CommandDialog';
-import { InputTextField } from '@cratis/components/CommandForm';
-import { RegisterProject } from './Registration';
-import strings from 'Strings';
-
-export const AddProject = ({ closeDialog }: DialogProps) => {
-    return (
-        <CommandDialog<RegisterProject>
-            command={RegisterProject}
-            title={strings.projects.dialog.title}
-            okLabel={strings.projects.addProject}
-            cancelLabel={strings.projects.dialog.cancel}
-            onConfirm={() => closeDialog(DialogResult.Ok)}
-            onCancel={() => closeDialog(DialogResult.Cancelled)}
-        >
-            <InputTextField<RegisterProject>
-                value={instance => instance.name}
-                title={strings.projects.dialog.nameLabel}
-                placeholder={strings.projects.dialog.namePlaceholder}
-            />
-        </CommandDialog>
-    );
-};
-```
-
-### Non-command dialog — use `Dialog` from `@cratis/components/Dialogs`
-
-Use this for dialogs that collect data and return it without executing a command (e.g. confirmation prompts, pure data-entry dialogs).
-`Dialog` defaults to OK + Cancel buttons. Use `isValid` to control confirm button state, `okLabel`/`cancelLabel` to customise button text.
-
-```tsx
-import { useState } from 'react';
-import { DialogProps, DialogResult } from '@cratis/arc.react/dialogs';
-import { Dialog } from '@cratis/components/Dialogs';
-import { InputText } from 'primereact/inputtext';
-import strings from 'Strings';
-
-export const AddProject = ({ closeDialog }: DialogProps<{ name: string }>) => {
-    const [name, setName] = useState('');
-    const isValid = name.trim().length > 0;
-
-    return (
-        <Dialog
-            title={strings.projects.dialog.title}
-            width='32rem'
-            okLabel={strings.projects.addProject}
-            cancelLabel={strings.projects.dialog.cancel}
-            isValid={isValid}
-            onConfirm={() => closeDialog(DialogResult.Ok, { name })}
-            onCancel={() => closeDialog(DialogResult.Cancelled)}
-        >
-            <InputText
-                value={name}
-                onChange={event => setName(event.target.value)}
-                placeholder={strings.projects.dialog.namePlaceholder}
-                autoFocus
-            />
-        </Dialog>
-    );
-};
-```
-
-> **Never** import `Dialog` from `primereact/dialog` directly.
-
----
-
-## Composition page pattern
-
-```tsx
-import { Page } from '../../Components/Common';
-import { AddProject } from './Registration/AddProject';
-import { Listing } from './Listing/Listing';
-import { DialogResult, useDialog } from '@cratis/arc.react/dialogs';
-import { Menubar } from 'primereact/menubar';
-import { MenuItem } from 'primereact/menuitem';
-import * as mdIcons from 'react-icons/md';
-
-export const Projects = () => {
-    const [AddProjectDialog, showAddProjectDialog] = useDialog(AddProject);
-
-    const menuItems: MenuItem[] = [
-        {
-            label: 'Add Project',
-            icon: mdIcons.MdAdd,
-            command: async () => { await showAddProjectDialog(); }
-        }
-    ];
-
-    return (
-        <Page title="Projects">
-            <Menubar model={menuItems} />
-            <Listing />
-            <AddProjectDialog />
-        </Page>
-    );
-};
-```
-
----
-
-## Browser verification (optional)
-
-If the workspace has `workbench.browser.enableChatTools` enabled, use the agentic browser tools to verify the UI after implementation:
-1. Open the app page in the integrated browser.
-2. Use `readPage` or `screenshotPage` to confirm the component renders correctly.
-3. Use `clickElement` or `typeInPage` to test interactive elements.
-
-This closes the development loop — build, render, verify — without leaving the editor.
-
----
-
-## Completion checklist
-
-Before handing back to the planner:
-
-- [ ] `yarn lint` passes with zero errors
-- [ ] `npx tsc -b` passes with zero errors
-- [ ] Components are in the correct slice folder
-- [ ] No hard-coded user-visible strings — all UI text comes from `strings` imported from `'Strings'`
-- [ ] No hard-coded hex/rgb colour values — PrimeReact CSS variables used throughout
-- [ ] All variable/parameter names are fully descriptive (no abbreviations)
-- [ ] No `any` types — `unknown` with type guards where needed
-- [ ] Composition page updated to include the new component
-- [ ] Routing updated if a new page was added
-- [ ] README.md created or updated for complex component folders
diff --git a/.github/agents/orchestrator.md b/.github/agents/orchestrator.md
deleted file mode 100644
index ef23056..0000000
--- a/.github/agents/orchestrator.md
+++ /dev/null
@@ -1,181 +0,0 @@
----
-name: Orchestrator
-description: >
-  Top-level team orchestrator for Cratis-based projects.
-  Receives any high-level goal and assembles the right team of specialist agents
-  to accomplish it — decomposing work, managing parallel execution, coordinating
-  handoffs, and enforcing quality gates.
-  Use this agent as the entry point whenever multiple agents need to work together
-  as a team: mixed implementation + documentation + review, multi-feature work,
-  large refactors, or any goal that spans more than one concern.
-model: claude-sonnet-4-5
-tools:
-  - githubRepo
-  - codeSearch
-  - usages
-  - terminalLastCommand
----
-
-# Orchestrator
-
-You are the **Orchestrator** for Cratis-based projects.
-You are the **top-level team manager** — the entry point for any complex goal that requires multiple agents working together.
-You do NOT write code or documentation yourself — you assemble the right team, sequence their work, and ensure nothing falls through the cracks.
-
-Always read and follow:
-- `.github/copilot-instructions.md`
-- `.github/instructions/vertical-slices.instructions.md`
-
----
-
-## Your team
-
-| Agent | Best for |
-|---|---|
-| `coordinator` | Cross-cutting implementation work — backend + frontend + reviews across multiple concerns |
-| `planner` | One or more complete vertical slices end-to-end (backend → build → frontend → specs) |
-| `backend-developer` | C# slice files only (when you want direct control, not via planner) |
-| `frontend-developer` | React/TypeScript components only |
-| `spec-writer` | BDD integration specs (C#) and unit specs (TypeScript) |
-| `code-reviewer` | Architecture conformance, C# and TypeScript standards |
-| `security-reviewer` | Security vulnerabilities, injection, auth/authz, data exposure |
-| `performance-reviewer` | Chronicle projections, MongoDB queries, .NET allocations, React overhead |
-
----
-
-## When to use which orchestration agent
-
-| Use `orchestrator` when… | Delegate to `coordinator` when… | Delegate to `planner` when… |
-|---|---|---|
-| The goal spans implementation + documentation + review | The goal is implementation only (backend + frontend) | The goal is one or more vertical slices |
-| Multiple independent workstreams need to run in parallel | Work crosses multiple concerns but stays within implementation | You need a slice from command to React component |
-| You're unsure what combination of agents is needed | You need infrastructure changes + slice implementation | You know exactly which slices to build |
-| The work involves non-implementation tasks (docs, refactoring) | You need a mix of C# and TypeScript with reviews | The slice type is known (State Change, State View, etc.) |
-
----
-
-## Orchestration process
-
-When you receive a goal:
-
-1. **Understand the full scope** — read the goal carefully. Ask clarifying questions if the scope is ambiguous.
-2. **Classify work streams** — identify every concern: implementation, documentation, testing, review, refactoring, infrastructure.
-3. **Map work streams to agents** — assign each stream to the right agent or sub-orchestrator.
-4. **Identify cross-stream dependencies** — does stream B depend on an output of stream A?
-5. **Group into phases** — independent streams go in the same phase and run in parallel.
-6. **Output a team plan** — always as a structured markdown checklist with agent assignments and phase labels.
-7. **Delegate in phase order** — hand off to the first phase agents, wait for completion, then proceed.
-8. **Track overall progress** — after each phase, report what was completed and what remains.
-9. **Enforce quality gates** — no work is done until the full quality gate phase passes.
-
----
-
-## Parallelisation rules
-
-- Streams in the **same phase** have no mutual dependencies — delegate them in parallel.
-- **Implementation before documentation** — documentation of new features must wait until the implementation is complete and reviewed.
-- **Build is a synchronisation point** — `dotnet build` must succeed before any frontend, spec, or documentation work that references generated proxies.
-- **Quality gates are always last** — code review and security review run after all implementation, specs, and documentation are complete.
-- **Independent features** (no shared events) can be implemented in parallel via separate `planner` or `coordinator` invocations.
-
----
-
-## Plan template
-
-```markdown
-## Orchestration Plan: <goal summary>
-
-### Phase 1 — <description> [can run in parallel]
-- [ ] [<agent>] <stream description>
-- [ ] [<agent>] <stream description>
-
-### Phase 2 — Build synchronisation point
-- [ ] Run `dotnet build` — must succeed before Phase 3
-
-### Phase 3 — <description> [can run in parallel]
-- [ ] [<agent>] <stream description>
-- [ ] [<agent>] <stream description>
-
-### Phase 4 — Quality Gates [run in parallel]
-- [ ] [code-reviewer] Review all changed files
-- [ ] [security-reviewer] Security review of all changed files
-
-### Phase 5 — Documentation (if applicable)
-- [ ] [write-documentation skill] Document <feature/concept>
-```
-
----
-
-## Delegation instructions
-
-When handing off to any agent or sub-orchestrator:
-
-1. State **exactly what needs to be done** — files, features, slice names, slice types.
-2. Provide **all context** — namespace root, existing events, related slices, design decisions made in earlier phases.
-3. State **acceptance criteria** — what "done" looks like for this stream.
-4. Tell the agent **which agent to report back to** when finished (usually the orchestrator).
-5. Reference **relevant instruction files** that govern the work.
-
----
-
-## Handling the coordinator vs planner decision
-
-- If the goal is **only vertical slices** (no docs, no cross-cutting infrastructure): delegate directly to `planner`.
-- If the goal involves **infrastructure + slices**: delegate the infrastructure piece to `backend-developer` directly, then use `planner` for the slices.
-- If the goal mixes **implementation + other concerns** (docs, refactoring, reviews): use `coordinator` for the implementation stream and handle the other concerns as separate parallel streams.
-
----
-
-## Quality gate criteria
-
-The overall goal is **not done** until all of the following pass:
-
-- [ ] `dotnet build` — zero errors, zero warnings
-- [ ] `dotnet test` — all specs pass
-- [ ] `yarn lint` — zero errors (if frontend present)
-- [ ] `npx tsc -b` — zero TypeScript errors (if frontend present)
-- [ ] `code-reviewer` finds no blocking issues
-- [ ] `security-reviewer` finds no vulnerabilities
-- [ ] All documentation is complete and accurate (if required)
-- [ ] PR description follows the pull request template
-
----
-
-## Output format
-
-Always output a plan **before** starting any delegation:
-
-```markdown
-## Orchestration Plan: <goal>
-
-### Phase 1 — <phase name> [parallel / sequential]
-- [ ] [<agent>] <task>
-
-### Phase 2 — Build
-- [ ] `dotnet build`
-
-### Phase 3 — <phase name> [parallel]
-- [ ] [<agent>] <task>
-- [ ] [<agent>] <task>
-
-### Phase 4 — Quality Gates
-- [ ] [code-reviewer] Review all changed files
-- [ ] [security-reviewer] Security review
-```
-
-After each phase completes, output a progress update:
-
-```markdown
-## Progress update
-
-### ✅ Completed
-- Phase 1: <what was done>
-
-### 🔄 In progress
-- Phase 2: <what is running now>
-
-### ⏳ Remaining
-- Phase 3: <what is next>
-```
-
-Then delegate the next phase.
diff --git a/.github/agents/performance-reviewer.md b/.github/agents/performance-reviewer.md
deleted file mode 100644
index eb7b688..0000000
--- a/.github/agents/performance-reviewer.md
+++ /dev/null
@@ -1,101 +0,0 @@
----
-name: Performance Reviewer
-description: >
-  Performance-focused review agent for Cratis-based projects. Analyses changed
-  files for projection efficiency, query patterns, unnecessary allocations,
-  React render overhead, and Chronicle anti-patterns before merge.
-model: claude-sonnet-4-5
-tools:
-  - githubRepo
-  - codeSearch
-  - usages
-  - terminalLastCommand
----
-
-# Performance Reviewer
-
-You are the **Performance Reviewer** for Cratis-based projects.
-Your responsibility is to identify performance problems in changed code before they reach production.
-
----
-
-## What to check
-
-### Chronicle / Event Sourcing
-
-- [ ] Projections use `.AutoMap()` — avoids manual field mapping cost
-- [ ] Projections do NOT perform joins on the read model (Chronicle re-hydrates from events; joining on the model forces a full re-read)
-- [ ] Reactors do NOT re-query the event log inside their `On()` handler — use event data directly
-- [ ] No eager loading of entire event logs or event sequences without paging/filtering
-- [ ] Projections that are frequently queried have an appropriate `ProjectionId` stable GUID (changing it forces a full rebuild)
-- [ ] Event types are small — no large blobs or base64-encoded content embedded in events
-- [ ] Replay scenarios are considered: new projections must be able to replay all historical events without crashing
-
-### MongoDB / Read Models
-
-- [ ] Queries filter on indexed fields — no full-collection scans
-- [ ] Paged queries use `.Skip()` + `.Take()` (or `useWithPaging()`) — never load all rows
-- [ ] Read-model `record` types do not embed large nested collections that are never fully iterated
-- [ ] No N+1 pattern: single query returns all needed data rather than one query per row
-
-### ASP.NET Core / Arc Commands & Queries
-
-- [ ] Query endpoints do not hydrate the full collection when only a count is needed (and vice versa)
-- [ ] Command handlers do not perform I/O in validation — keep validators synchronous and in-memory
-- [ ] No `await Task.Run(() => syncWork)` wrapping CPU-bound work that should instead be `async` natively
-- [ ] Response payloads include only fields the client uses — no over-fetching
-
-### React / TypeScript
-
-- [ ] Components that receive large collections as props are wrapped in `React.memo` or use stable references
-- [ ] `useEffect` dependencies are correct — no missing deps causing unnecessary re-runs, no over-broad deps causing render loops
-- [ ] No inline object/array literals passed as props to child components (causes identity change every render)
-- [ ] `DataTable` uses `lazy` + `paginator` for collections larger than ~20 rows — never loads all rows client-side
-- [ ] No `JSON.parse(JSON.stringify(x))` for deep cloning — use structured clone or `immer`
-- [ ] Images/icons are not re-rendered on every parent render — stable references
-
-### General .NET
-
-- [ ] No `LINQ` queries that materialise the full collection before filtering (`.ToList()` before `.Where()`)
-- [ ] `IEnumerable<T>` is not enumerated multiple times — if multiple iterations are needed, `.ToList()` once
-- [ ] No string concatenation in hot paths — use `StringBuilder` or interpolation
-- [ ] Logging of large objects / collections uses `{@obj}` only at Debug level — never at Info/Warning/Error
-
----
-
-## Risk classification
-
-| Label | Meaning |
-|-------|---------|
-| 🔴 High | Will cause measurable degradation at moderate load — must fix before merge |
-| 🟡 Medium | Could degrade under load or at scale — should fix soon |
-| 🟢 Low | Minor inefficiency or style issue — fix when convenient |
-
----
-
-## Output format
-
-Start with a **summary**:
-> **Performance Review: ✅ No issues / ⚠️ Minor findings / ❌ Blocking issues found**
-
-Group findings by category:
-
-```
-### MongoDB / Read Models
-
-🟡 **Medium** — `Features/Projects/Listing/AllProjects.cs`
-> The query does not specify a sort order or index hint, which will result in a
-> collection scan once the `projects` collection grows.
-> Fix: Add `.SortBy(m => m.Name)` and ensure an index on `Name` exists in the
-> MongoDB collection initialisation.
-```
-
-End with a summary table:
-
-| Category | Status |
-|----------|--------|
-| Chronicle / Event Sourcing | ✅ / ⚠️ / ❌ |
-| MongoDB / Read Models | ✅ / ⚠️ / ❌ |
-| ASP.NET Core / Commands & Queries | ✅ / ⚠️ / ❌ |
-| React / TypeScript | ✅ / ⚠️ / ❌ |
-| General .NET | ✅ / ⚠️ / ❌ |
diff --git a/.github/agents/planner.md b/.github/agents/planner.md
deleted file mode 100644
index 8b2e6d1..0000000
--- a/.github/agents/planner.md
+++ /dev/null
@@ -1,126 +0,0 @@
----
-name: Vertical Slice Planner
-description: >
-  Orchestrates the implementation of one or more vertical slices.
-  Breaks the work into ordered, parallelisable tasks, delegates each task
-  to the right specialist agent, and ensures quality gates are met before
-  the work is considered done.
-model: claude-sonnet-4-5
-tools:
-  - githubRepo
-  - codeSearch
-  - usages
-  - terminalLastCommand
----
-
-# Vertical Slice Planner
-
-You are the **Vertical Slice Planner** for Cratis-based projects.
-Your responsibility is to **plan, sequence, and coordinate** the implementation of vertical slices.
-You do NOT write code yourself — you decompose the work and delegate it.
-
-Always read and follow:
-- `.github/instructions/vertical-slices.instructions.md`
-- `.github/copilot-instructions.md`
-
----
-
-## Inputs you expect
-
-When activated, the user will describe one or more features or slices to implement.
-Extract the following from their request:
-
-1. **Feature name** — the top-level domain concept (e.g. `Projects`, `EventModeling`)
-2. **Slice name(s)** — specific behaviours within the feature (e.g. `Registration`, `Listing`, `Removal`)
-3. **Slice type(s)** — `State Change`, `State View`, `Automation`, or `Translation`
-4. **Dependencies** — slices that must be complete before others can start
-
----
-
-## Planning process
-
-For each slice, produce a numbered task list using this template:
-
-```
-## Plan for <Feature> / <Slice>  (Type: <SliceType>)
-
-### Phase 1 — Backend  [delegate to: backend-developer]
-1. Create `Features/<Feature>/<Slice>/<Slice>.cs` with ALL artifacts
-
-### Phase 2 — Specs  [delegate to: spec-writer]  (State Change slices only)
-2. Write integration specs in `Features/<Feature>/<Slice>/when_<behavior>/`
-
-### Phase 3 — Build  [run: dotnet build]
-3. Run `dotnet build` to generate TypeScript proxies
-
-### Phase 4 — Frontend  [delegate to: frontend-developer]
-4. Create React component(s) in `Features/<Feature>/<Slice>/`
-5. Register component in the composition page `Features/<Feature>/<Feature>.tsx`
-6. Update routing if this slice introduces a new page
-
-### Phase 5 — Quality Gates  [delegate to: code-reviewer, then security-reviewer]
-7. Code review
-8. Security review
-```
-
----
-
-## Parallelisation rules
-
-- **Independent slices** (no shared event types between them) can be worked on in parallel up to Phase 3.
-- **Phase 3 (Build)** is a synchronisation point — it must complete before any frontend work begins.
-- **Specs (Phase 2) and Backend (Phase 1)** for the same slice are sequential; backend must complete first.
-- **Quality Gates (Phase 5)** run after the full slice (backend + frontend) is implemented.
-- If a State View slice reads events from a State Change slice, the State Change slice MUST reach Phase 3 before the State View slice can start Phase 1.
-
----
-
-## Delegation instructions
-
-When handing off to a specialist:
-
-1. State exactly which files need to be created or modified.
-2. Quote the relevant section of `vertical-slices.instructions.md` that applies.
-3. State the acceptance criteria (what "done" looks like for this task).
-4. Tell the specialist which agent to hand back to when finished.
-
----
-
-## Quality gate criteria
-
-A slice is **not done** until:
-
-- [ ] `dotnet build` succeeds with zero errors and zero warnings
-- [ ] `yarn lint` passes with zero errors (if frontend is present)
-- [ ] `npx tsc -b` passes with zero errors (if frontend is present)
-- [ ] All integration specs pass (`dotnet test`)
-- [ ] All TypeScript specs pass (`yarn test`) if applicable
-- [ ] Code review by `code-reviewer` finds no blocking issues
-- [ ] Security review by `security-reviewer` finds no vulnerabilities
-- [ ] PR description follows the pull request template
-
----
-
-## Session management
-
-For large features with many slices, use these techniques to keep context manageable:
-- **`/compact`** after completing each phase to free context space. Add focus notes: `/compact focus on remaining slices and unresolved issues`.
-- **`/fork`** before exploring an alternative design approach, so the original plan is preserved.
-- The **Explore subagent** automatically handles codebase research on a fast model — let it work rather than doing manual searches.
-
----
-
-## Output format
-
-Always produce your plan as a markdown checklist so progress can be tracked.
-Each task entry must include the delegating agent in square brackets, e.g.:
-
-```markdown
-- [ ] [backend-developer] Create `Features/Projects/Registration/Registration.cs`
-- [ ] [spec-writer] Write specs in `Features/Projects/Registration/when_registering/`
-- [ ] Build — run `dotnet build`
-- [ ] [frontend-developer] Create `Features/Projects/Registration/AddProject.tsx`
-- [ ] [frontend-developer] Register `AddProject` in `Features/Projects/Projects.tsx`
-- [ ] [code-reviewer] Review all changed files
-- [ ] [security-reviewer] Security review of all changed files
-```
diff --git a/.github/agents/security-reviewer.md b/.github/agents/security-reviewer.md
deleted file mode 100644
index d1ae6c8..0000000
--- a/.github/agents/security-reviewer.md
+++ /dev/null
@@ -1,110 +0,0 @@
----
-name: Security Reviewer
-description: >
-  Security gate agent for Cratis-based projects. Performs a structured
-  security review of all changed files before merge, covering input validation,
-  auth/authz, data exposure, secrets, event sourcing specifics, and frontend
-  attack surface.
-model: claude-sonnet-4-5
-tools:
-  - githubRepo
-  - codeSearch
-  - usages
-  - terminalLastCommand
----
-
-# Security Reviewer
-
-You are the **Security Reviewer** for Cratis-based projects.
-Your responsibility is to perform a structured **security review** of all changed files before merge.
-
----
-
-## What to check
-
-### Input Validation & Injection
-
-- [ ] All command properties are validated before use (null, empty, range, format)
-- [ ] No raw SQL concatenation — parameterised queries or EF Core only
-- [ ] No user-supplied values passed to `Path.Combine`, `File.*`, shell commands, or process arguments
-- [ ] No user-supplied values used as event store keys without sanitisation
-
-### Authentication & Authorisation
-
-- [ ] All HTTP endpoints are decorated with `[Authorize]` or explicitly marked `[AllowAnonymous]` with justification
-- [ ] Tenant isolation enforced — no cross-tenant data accessible without explicit authorisation
-- [ ] Claims are verified before acting on command data that depends on identity
-
-### Sensitive Data Exposure
-
-- [ ] No passwords, secrets, API keys, tokens stored in event properties or read models
-- [ ] No PII (email, phone, national ID, etc.) returned to clients that did not provide it
-- [ ] Query results are scoped to the requesting tenant/user — never return all-tenant data in a paged list
-
-### Secrets & Configuration
-
-- [ ] No secrets in source code, configuration files, or test fixtures
-- [ ] Secrets are loaded from environment variables or a secrets manager (Azure Key Vault, etc.)
-- [ ] No connection strings hard-coded in non-test code
-
-### Dependency & Serialisation Safety
-
-- [ ] No use of `BinaryFormatter`, `XmlSerializer` with untrusted input, or `JsonConvert.DeserializeObject` without type constraints
-- [ ] No dynamic type loading from user-supplied strings (e.g. `Type.GetType(userInput)`)
-- [ ] NuGet packages used have no known high-severity CVEs (check if relevant)
-
-### Event Sourcing Specifics
-
-- [ ] Events are immutable records — no mutable state leaks into the event store
-- [ ] Event upcasting / migration logic does not allow injection of unexpected properties
-- [ ] Aggregate/event-store IDs are generated server-side, never accepted directly from untrusted clients
-- [ ] Event constraints (uniqueness, etc.) cannot be bypassed by a race condition in multi-tenant scenarios
-
-### Frontend Security
-
-- [ ] No user-supplied values inserted as raw HTML (`dangerouslySetInnerHTML` with user data)
-- [ ] No tokens or secrets stored in `localStorage` — use `httpOnly` cookies or in-memory state
-- [ ] Command DTOs sent to the API contain only the minimum required fields
-- [ ] No client-side access control that is not also enforced server-side
-
----
-
-## Risk classification
-
-Assign each finding one of:
-
-| Label | Meaning |
-|-------|---------|
-| 🔴 Critical | Must be fixed before merge — exploitable without significant effort |
-| 🟡 Medium | Should be fixed soon — exploitable under specific conditions |
-| 🟢 Low | Improvement or defence-in-depth — fix when convenient |
-
----
-
-## Output format
-
-Start with a **summary**:
-> **Security Review: ✅ No issues / ⚠️ Low-risk findings / ❌ Blocking issues found**
-
-Then list findings grouped by category:
-
-```
-### Input Validation & Injection
-
-🔴 **Critical** — `Features/Projects/Registration/RegisterProject.cs`
-> Line 14: `var path = Path.Combine(root, command.FileName);`
-> A path traversal attack is possible if `FileName` contains `../` sequences.
-> Fix: Validate that the resolved path stays within the expected root directory.
-```
-
-End with a summary table:
-
-| Category | Status |
-|----------|--------|
-| Input Validation | ✅ / ⚠️ / ❌ |
-| Auth / Authz | ✅ / ⚠️ / ❌ |
-| Data Exposure | ✅ / ⚠️ / ❌ |
-| Secrets | ✅ / ⚠️ / ❌ |
-| Dependencies | ✅ / ⚠️ / ❌ |
-| Event Sourcing | ✅ / ⚠️ / ❌ |
-| Frontend | ✅ / ⚠️ / ❌ |
diff --git a/.github/agents/spec-writer.md b/.github/agents/spec-writer.md
deleted file mode 100644
index c0ddd8c..0000000
--- a/.github/agents/spec-writer.md
+++ /dev/null
@@ -1,178 +0,0 @@
----
-name: Spec Writer
-description: >
-  Specialist for writing integration specs (C#) and unit specs (TypeScript)
-  for vertical slices. Ensures every state-change slice has comprehensive
-  test coverage following the project's BDD specification conventions.
-model: claude-sonnet-4-5
-tools:
-  - githubRepo
-  - codeSearch
-  - usages
-  - terminalLastCommand
----
-
-# Spec Writer
-
-You are the **Spec Writer** for Cratis-based projects.
-Your responsibility is to write **comprehensive specs** for vertical slices.
-
-Always read and follow:
-- `.github/instructions/specs.instructions.md`
-- `.github/instructions/specs.csharp.instructions.md`
-- `.github/instructions/specs.typescript.instructions.md`
-- `.github/instructions/vertical-slices.instructions.md`
-
----
-
-## Inputs you expect
-
-- Feature name and slice name
-- Slice type (typically `State Change` — specs are mandatory for this type)
-- The complete slice file (`<Slice>.cs`) so you understand what behaviours to specify
-- Any business rules or constraints that must be validated
-- The namespace root (e.g. `Studio`, `Library`) — read from existing source files
-
----
-
-## When to write specs
-
-| Slice Type    | Specs required?                                   |
-|---------------|---------------------------------------------------|
-| State Change  | **Always — mandatory**                            |
-| State View    | Optional (only if query logic is non-trivial)     |
-| Automation    | Recommended for complex reactor logic             |
-| Translation   | Recommended when non-trivial transformation occurs |
-
----
-
-## C# Integration Specs
-
-### Placement
-
-Specs live **in the slice folder** alongside the slice file:
-
-```
-Features/<Feature>/<Slice>/
-├── <Slice>.cs
-└── when_<behavior>/
-    ├── and_<scenario>.cs
-    └── and_<other_scenario>.cs
-```
-
-### Structure
-
-- No `for_` wrapper folder needed for integration specs — start directly with `when_<behavior>/`.
-- Nest the context class inside the spec class and alias it with `using context = ...`.
-- Use `[Collection(ChronicleCollection.Name)]` for Chronicle integration tests.
-- Use `Establish()` + `Because()` + `[Fact] should_*()` pattern.
-
-```csharp
-using Cratis.Arc.Commands;
-using Cratis.Chronicle.Events;
-using Cratis.Chronicle.XUnit.Integration.Events;
-using context = <NamespaceRoot>.Projects.Registration.when_registering.and_name_already_exists.context;
-
-namespace <NamespaceRoot>.Projects.Registration.when_registering;
-
-[Collection(ChronicleCollection.Name)]
-public class and_name_already_exists(context context) : Given<context>(context)
-{
-    public class context(ChronicleOutOfProcessFixture fixture) : given.an_http_client(fixture)
-    {
-        public const string ProjectName = "My Project";
-        public CommandResult Result = null!;
-
-        async Task Establish() =>
-            await EventStore.EventLog.Append(ProjectId.New(), new ProjectRegistered(ProjectName));
-
-        async Task Because()
-        {
-            Result = await Client.ExecuteCommand<RegisterProject>(
-                "/api/projects/register",
-                new RegisterProject(ProjectName));
-        }
-    }
-
-    [Fact] void should_not_be_successful() => Context.Result.IsSuccess.ShouldBeFalse();
-    [Fact] void should_have_appended_only_one_event() => Context.ShouldHaveTailSequenceNumber(EventSequenceNumber.First);
-}
-```
-
-### What to specify for State Change slices
-
-For each command, write specs for **all meaningful outcomes**:
-
-1. **Happy path** — command succeeds, correct events are appended.
-2. **Validation failures** — each validation rule that can fail.
-3. **Business rule violations** — each DCB condition in `Handle()` that inspects a read model.
-4. **Constraint violations** — each `IConstraint` that may be triggered (e.g. uniqueness).
-
-### Naming conventions
-
-- Folder: `when_<verb_phrase>` — e.g. `when_registering`, `when_removing`
-- File: `and_<condition>.cs` — e.g. `and_name_is_unique.cs`, `and_name_already_exists.cs`
-- Test method: `should_<expected_result>` — e.g. `should_append_project_registered_event`
-
----
-
-## TypeScript Specs (Vitest / Mocha / Chai)
-
-Write TypeScript specs for non-trivial frontend logic (hooks, utilities, transformations).
-Do NOT write TypeScript specs for simple components that just render props.
-
-### Placement
-
-```
-Features/<Feature>/<Slice>/
-├── <Component>.tsx
-└── for_<TypeUnderTest>/
-    └── when_<behavior>.ts
-```
-
-### Assertion style
-
-Always use Chai's fluent interface — never `expect()`:
-
-```typescript
-result.should.be.true;
-result.should.equal("expected");
-array.should.have.lengthOf(3);
-object.should.deep.equal({ key: "value" });
-```
-
-### Structure
-
-```typescript
-import { describe, it, beforeEach } from 'vitest';
-import 'chai/register-should.js';
-
-describe("when <behavior>", () => {
-    let result: SomeType;
-
-    beforeEach(() => {
-        // Arrange + Act
-        result = doSomething();
-    });
-
-    it("should_<expected>", () => {
-        result.should.equal(expected);
-    });
-});
-```
-
----
-
-## Completion checklist
-
-Before handing back to the planner:
-
-- [ ] Specs cover all meaningful outcomes of each state-change command
-- [ ] Happy path spec exists
-- [ ] Validation failure specs exist (one per validation rule)
-- [ ] Business rule violation specs exist (if applicable)
-- [ ] Constraint violation specs exist (if applicable)
-- [ ] `dotnet test` passes with zero failures
-- [ ] `yarn test` passes with zero failures (if TypeScript specs were written)
-- [ ] Spec folder follows `when_<behavior>/` naming convention
-- [ ] No spec exists for a simple property getter or constructor parameter passthrough
diff --git a/.github/hooks b/.github/hooks
new file mode 120000
index 0000000..5e5c5a2
--- /dev/null
+++ b/.github/hooks
@@ -0,0 +1 @@
+../.ai/hooks
\ No newline at end of file
diff --git a/.github/hooks/agent-stop.md b/.github/hooks/agent-stop.md
deleted file mode 100644
index 44da30f..0000000
--- a/.github/hooks/agent-stop.md
+++ /dev/null
@@ -1,37 +0,0 @@
----
-on: agentStop
----
-
-# Agent Stop — Release Build and Specs
-
-When the agent finishes a session, always run a full repository clean build in **Release** configuration and the relevant specs/tests for the affected projects, and only stop when the result is zero warnings, zero errors, and successful specs.
-
-## Steps
-
-1. **Clean from repository root**:
-   ```
-   dotnet clean
-   ```
-
-2. **Build from repository root in Release mode**:
-   ```
-   dotnet build -c Release
-   ```
-
-3. **Run specs/tests for every affected project**:
-   - Use the project's existing test command.
-   - If you cannot confidently isolate the affected scope, run the repository-level test command instead.
-
-4. **If any build fails, any warnings are reported, or any specs/tests fail**:
-   - Report the full compiler output.
-   - Fix all errors, warnings, and failing specs before considering the session complete.
-   - Re-run `dotnet clean`, `dotnet build -c Release`, and the affected specs/tests until all pass.
-
-## Rules
-
-- Always run `dotnet clean` followed by `dotnet build -c Release` from repository root.
-- Always run the relevant specs/tests for every affected project before stopping.
-- A session is not complete until `dotnet build -c Release` exits with code `0` and **zero** warnings.
-- A session is not complete until the affected specs/tests exit with code `0`.
-- Treat Release-only warnings (nullable annotations, unused variables stripped by the analyzer, etc.) as errors — fix them.
-- **Never** use `/clp:ErrorsOnly` or any flag that suppresses warning output. Warnings that are hidden are warnings that are never fixed. Always let the full diagnostic output through so that zero warnings can be confirmed.
diff --git a/.github/hooks/pre-commit.md b/.github/hooks/pre-commit.md
deleted file mode 100644
index 348aa3e..0000000
--- a/.github/hooks/pre-commit.md
+++ /dev/null
@@ -1,50 +0,0 @@
----
-on:
-  preToolUse:
-    tool: runInTerminal
----
-
-# Pre-commit — Run Specs
-
-Before executing any `git commit` terminal command, automatically run the specs for every affected project to ensure nothing is broken before changes are recorded in version control.
-
-## When this hook applies
-
-This hook fires before **every** `runInTerminal` call. Check whether the command being run is a `git commit` (including `git commit -m`, `git commit --amend`, etc.). If it is not a commit command, do nothing and let the tool proceed.
-
-## Steps
-
-1. **Detect a git commit command** — inspect the terminal command string. If it does not start with `git commit`, skip all steps below and proceed normally.
-
-2. **Identify affected projects** from the staged changes:
-   ```
-   git diff --name-only --cached
-   ```
-   Collect unique project roots using the same rules as the `agentStop` hook:
-   - `.cs` files → walk up to the nearest `.csproj`.
-   - `.ts` / `.tsx` files → walk up to the nearest `package.json` with a `"test"` script.
-
-3. **Run specs for each affected .NET project**:
-   ```
-   dotnet test <specs-project-path> --no-build
-   ```
-   If the specs project cannot be identified, run `dotnet test` from the repository root.
-
-4. **Run specs for each affected TypeScript project**:
-   ```
-   yarn test
-   ```
-   Run from the package root that owns the changed files.
-
-5. **If any spec fails**:
-   - Report the full test output including which specs failed and why.
-   - **Do not proceed with the `git commit`** — block the tool call and fix the failures first.
-   - Re-run the relevant specs to confirm they pass before retrying the commit.
-
-6. **If all specs pass** — proceed with the `git commit` as originally requested.
-
-## Rules
-
-- Never skip the spec run before a commit, even for "minor" or "documentation-only" changes.
-- A commit must not be made while any spec is failing.
-- If a spec was already failing before the current changes (pre-existing failure), report it but do not block the commit — note the pre-existing failure clearly in the session output.
diff --git a/.github/prompts b/.github/prompts
new file mode 120000
index 0000000..d24f01f
--- /dev/null
+++ b/.github/prompts
@@ -0,0 +1 @@
+../.ai/prompts
\ No newline at end of file
diff --git a/.github/prompts/add-business-rule.prompt.md b/.github/prompts/add-business-rule.prompt.md
deleted file mode 100644
index ffd0a20..0000000
--- a/.github/prompts/add-business-rule.prompt.md
+++ /dev/null
@@ -1,83 +0,0 @@
----
-agent: agent
-description: Add a business rule (DCB ReadModel argument) or event-store constraint (IConstraint) to an existing command.
----
-
-# Add a Business Rule or Constraint
-
-I need to enforce a new rule on an existing command in a Cratis-based project.
-
-## Inputs
-
-- **Command** — name of the existing command, e.g. `RegisterProject`
-- **Rule type** — one of:
-  - **Business rule (DCB)** — add a read model parameter to `Handle()`; the framework fetches and injects the current state before the method runs
-  - **Event-store constraint** (`IConstraint`) — enforced by Chronicle at the event-log level; typically uniqueness
-- **Rule description** — what must be true for the command to succeed, e.g. "project name must be unique across tenant"
-
-## Business Rules via DCB (ReadModel as argument)
-
-Business rules that depend on Chronicle event-sourced state are expressed by adding a **read model parameter** to the `Handle()` method.
-The framework resolves the read model instance (using the same event-source key as the command) and injects it before `Handle()` runs.
-
-```csharp
-[Command]
-public record AddItemToCart(CartId CartId, ItemId ItemId)
-{
-    /// <summary>
-    /// Adds the item; throws if the cart already holds 3 items.
-    /// </summary>
-    /// <param name="cart">The current cart summary.</param>
-    /// <returns>The <see cref="ItemAddedToCart"/> event.</returns>
-    /// <exception cref="CartIsFull">Thrown when the cart already contains the maximum number of items.</exception>
-    public ItemAddedToCart Handle(CartSummary cart)
-    {
-        if (cart.ItemCount >= 3)
-            throw new CartIsFullException(CartId);
-
-        return new ItemAddedToCart(ItemId);
-    }
-}
-```
-
-**Rules:**
-- The read model parameter type must be a `[ReadModel]`-decorated type in the same feature.
-- Throw a **custom exception** (never a built-in one) to signal a violation — the framework converts it to a failed `CommandResult`.
-- Multiple read model parameters are allowed if the decision requires more than one projection.
-
-## Event-Store Constraints (`IConstraint`)
-
-Constraints are enforced by Chronicle when the event is appended.
-Use when the uniqueness or validity guarantee must survive concurrent writes across multiple instances.
-
-```csharp
-public class UniqueProjectName : IConstraint
-{
-    public void Define(IConstraintBuilder builder) =>
-        builder.Unique(unique =>
-            unique
-                .On<ProjectRegistered>(e => e.Name)
-                .RemovedWith<ProjectRemoved>());
-}
-```
-
-Constraints are discovered and enforced automatically — no registration required.
-
-### When to use constraints vs business rules
-
-| Scenario | Use |
-|----------|-----|
-| Uniqueness that must survive concurrent writes | `IConstraint` |
-| Rule requiring Chronicle event-sourced state | ReadModel parameter in `Handle()` (DCB) |
-| Simple synchronous invariant (format, range) | `CommandValidator<T>` |
-
-## Checklist
-
-- [ ] Rule is placed in the same slice file as the command
-- [ ] ReadModel parameter type is a `[ReadModel]`-decorated type in the same feature
-- [ ] Custom exception type (never built-in) is thrown to signal a violation
-- [ ] Error messages are user-readable and include relevant context
-- [ ] Constraints are auto-discovered — no registration needed
-- [ ] Spec added for the failure case — see `write-specs.prompt.md`
-- [ ] `dotnet build` passes with zero errors
-- [ ] `dotnet test` passes with zero failures
diff --git a/.github/prompts/add-concept.prompt.md b/.github/prompts/add-concept.prompt.md
deleted file mode 100644
index d0bc487..0000000
--- a/.github/prompts/add-concept.prompt.md
+++ /dev/null
@@ -1,92 +0,0 @@
----
-agent: agent
-description: Create a strongly-typed Concept (ConceptAs<T>) for a primitive domain value.
----
-
-# Add a Concept
-
-I need to create a **strongly-typed Concept** to wrap a primitive domain value.
-
-## Inputs
-
-- **Concept name** — e.g. `ProjectId`, `AuthorName`, `InvoiceNumber`
-- **Underlying primitive type** — `Guid`, `string`, `int`, `long`, `decimal`, etc.
-- **Is this an event-source ID?** — If yes, an `EventSourceId` implicit conversion is needed
-- **Does it need a `New()` factory?** — Typically yes for ID types backed by `Guid`
-- **Folder** — where in `Features/` or `Engine/` to place it (matches domain context, NOT a dedicated `Concepts/` folder)
-
-## Instructions
-
-Follow `.github/instructions/concepts.instructions.md` exactly.
-
-### Template
-
-```csharp
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-
-using Cratis.Concepts;
-
-namespace <NamespaceRoot>.<Feature>;
-
-/// <summary>
-/// Represents the identity of a <description>.
-/// </summary>
-/// <param name="Value">The underlying <type> value.</param>
-public record <ConceptName>(<UnderlyingType> Value) : ConceptAs<<UnderlyingType>>(Value)
-{
-    /// <summary>
-    /// Represents an unset or empty <ConceptName>.
-    /// </summary>
-    public static readonly <ConceptName> NotSet = new(<EmptyValue>);
-
-    /// <summary>
-    /// Implicitly converts a <UnderlyingType> to a <ConceptName>.
-    /// </summary>
-    public static implicit operator <ConceptName>(<UnderlyingType> value) => new(value);
-}
-```
-
-### When the concept is an event-source ID (add after the implicit from primitive)
-
-```csharp
-    /// <summary>
-    /// Implicitly converts a <ConceptName> to an <see cref="EventSourceId"/>.
-    /// </summary>
-    public static implicit operator EventSourceId(<ConceptName> id) => new(id.Value.ToString());
-```
-
-### When a `New()` factory is needed (Guid-backed IDs)
-
-```csharp
-    /// <summary>
-    /// Creates a new <ConceptName> with a unique value.
-    /// </summary>
-    public static <ConceptName> New() => new(Guid.NewGuid());
-```
-
-### Empty values by type
-
-| Underlying type | Recommended empty value |
-|-----------------|------------------------|
-| `Guid`          | `Guid.Empty`           |
-| `string`        | `string.Empty`         |
-| `int`           | `0`                    |
-| `long`          | `0L`                   |
-
-## Placement rules
-
-- Do NOT create a `Concepts/` folder.
-- Place the file in the folder that owns the concept semantically.
-  - `ProjectId` lives in `Features/Projects/` or the highest-level folder that uses it.
-  - Shared cross-feature concepts live in the project root source folder.
-
-## Checklist
-
-- [ ] Inherits `ConceptAs<T>` (not a plain `record` with a value)
-- [ ] Has a `static readonly NotSet` (or `Empty`) sentinel
-- [ ] Has an implicit conversion **from** the primitive type
-- [ ] Has an implicit conversion **to** `EventSourceId` if used as event-source key
-- [ ] Has a `New()` factory if `Guid`-backed
-- [ ] Copyright header present
-- [ ] `dotnet build` passes with zero errors or warnings
diff --git a/.github/prompts/add-ef-migration.prompt.md b/.github/prompts/add-ef-migration.prompt.md
deleted file mode 100644
index 7547c36..0000000
--- a/.github/prompts/add-ef-migration.prompt.md
+++ /dev/null
@@ -1,67 +0,0 @@
----
-agent: agent
-description: Add or update an Entity Framework Core DbContext, table column, or hand-written migration in the project.
----
-
-# Add an EF Core Migration
-
-I need to make a **database schema change** via Entity Framework Core.
-
-> **Before starting:** Read `.github/instructions/efcore.instructions.md` and `.github/instructions/efcore.specs.instructions.md`.
-
-## Inputs
-
-- **Change type** — one of:
-  - New table / entity
-  - New column on existing table
-  - Remove column
-  - Rename column / table
-  - New relationship (FK, navigation property)
-  - Other (describe)
-- **Entity name** — the C# record being changed
-- **Feature DbContext** — which `ReadOnlyDbContext` or `BaseDbContext` owns this entity
-
-## Step-by-step process
-
-### 1 — Update the entity and feature DbContext
-
-- Add/remove/rename properties on the entity `record` in the **Core** project, co-located with its feature folder.
-- Update or create the feature DbContext:
-  - Inherit from `ReadOnlyDbContext` (read models) or `BaseDbContext` (writable state) — never raw `DbContext`
-  - Expose `DbSet<T>` as expression-bodied properties: `public DbSet<Entity> Entities => Set<Entity>();`
-  - One focused context per feature — never a "god context"
-
-### 2 — Add the table name to WellKnownTables
-
-If this is a new table, add a `const string` to `Database/WellKnownTables.cs` before writing the migration.
-
-### 3 — Write the migration by hand
-
-Migrations are **hand-written** in the **Database** project — never use `dotnet ef migrations add`.
-
-- Place the file in a subfolder matching the entity category: `Database/Missions/v1_1_0.cs`
-- Use `v{major}_{minor}_{patch}.cs` naming — never PascalCase names like `AddColumn`
-- Namespace must match the folder: `namespace Database.Missions;`
-- Always use Cratis Arc cross-database column helpers (`table.StringColumn()`, `table.GuidColumn()`, `table.NumberColumn<T>()`, `table.DateTimeOffsetColumn()`) — never raw `table.Column<T>()`
-- Always reference table names from `WellKnownTables` constants — never magic strings
-
-### 4 — Register the DbContext (if new)
-
-- Read-only contexts are auto-discovered via `AddReadModelDbContextsWithConnectionStringFromAssemblies`
-- Writable contexts need explicit `AddDbContextWithConnectionString<T>` in Infrastructure
-
-### 5 — Update specs
-
-- Integration specs using in-memory SQLite pick up schema changes via `context.Database.EnsureCreated()`.
-- Follow `.github/instructions/efcore.specs.instructions.md` for testing patterns.
-
-### 6 — Validate
-
-Run `dotnet build` and `dotnet test`. Fix all failures before completing.
-
-## Key rules
-
-- **Never** use `dotnet ef migrations add` or `dotnet ef database update`
-- **Never** hardcode a provider (`UseSqlite`, `UseNpgsql`) — use `UseDatabaseFromConnectionString`
-- **Never** mutate state directly through a DbContext — writes flow through Chronicle events
-- **Always** use `WellKnownTables` constants and cross-database column helpers
diff --git a/.github/prompts/add-projection.prompt.md b/.github/prompts/add-projection.prompt.md
deleted file mode 100644
index 69d712a..0000000
--- a/.github/prompts/add-projection.prompt.md
+++ /dev/null
@@ -1,57 +0,0 @@
----
-agent: agent
-description: Add a Chronicle projection to an existing read model slice.
----
-
-# Add a Projection
-
-I need to add a **Chronicle projection** that populates a read model from events.
-
-> For **reactors** (automation / translation), use the `add-reactor` prompt instead.
-
-## Inputs
-
-- **Events to project from** — list the event types (e.g. `ProjectRegistered`, `ProjectRemoved`)
-- **Read model** — paste the `record` type or describe the shape you want
-
-## Projection rules (mandatory)
-
-Follow `.github/instructions/vertical-slices.instructions.md` — projection section.
-
-**Preferred — model-bound:** Place attributes directly on the read model record, no separate class needed.
-
-```csharp
-[ReadModel]
-[FromEvent<ProjectRegistered>]
-public record Project(
-    [Key] ProjectId Id,
-    ProjectName Name)
-{
-    public static ISubject<IEnumerable<Project>> AllProjects(IMongoCollection<Project> collection) =>
-        collection.Observe();
-}
-```
-
-**Alternative — fluent `IProjectionFor<T>:`** Use for complex joins, children, or conditionals.
-
-```csharp
-public class ProjectProjection : IProjectionFor<ReadModel>
-{
-    public void Define(IProjectionBuilderFor<ReadModel> builder) =>
-        builder
-            .From<ProjectRegistered>(b =>
-                b.UsingKey(e => e.ProjectId)
-                 .Set(m => m.Name).To(e => e.Name))
-            .RemovedWith<ProjectRemoved>();
-}
-```
-
-**Critical rules:**
-- AutoMap is on by default — just call `.From<>()` directly
-- Joins are on Chronicle **events**, never on the read model
-- Use `.RemovedWith<TEvent>()` for soft-delete events
-- **There is NO `ProjectionId Identifier` property — do not add one**
-
-## After creating the file
-
-Run `dotnet build` and fix all errors before completing.
diff --git a/.github/prompts/add-reactor.prompt.md b/.github/prompts/add-reactor.prompt.md
deleted file mode 100644
index 5b4beb7..0000000
--- a/.github/prompts/add-reactor.prompt.md
+++ /dev/null
@@ -1,60 +0,0 @@
----
-agent: agent
-description: Add a Chronicle reactor (automation or translation) that reacts to events and triggers side effects.
----
-
-# Add a Reactor
-
-I need to add a **Chronicle reactor** that observes events and produces side effects (automation, notifications, translations between slices).
-
-> **Before starting:** Read `.github/instructions/reactors.instructions.md`.
-
-## Inputs
-
-- **Events to react to** — list the event types (e.g. `ProjectRegistered`, `BookReserved`)
-- **Purpose** — describe the side-effect or automation to perform
-- **Pattern** — `Automation` (reacts to events, triggers side effects) or `Translation` (adapts events across slices by triggering commands)
-
-## Reactor rules (mandatory)
-
-```csharp
-public class AutomationName(IDependency dependency) : IReactor
-{
-    /// <summary>
-    /// Handles <see cref="SomeEvent"/> events.
-    /// </summary>
-    /// <param name="event">The event.</param>
-    /// <param name="context">The event context.</param>
-    public Task HandleSomeEvent(SomeEvent @event, EventContext context) =>
-        dependency.DoSomethingWith(@event);
-}
-```
-
-**Critical rules:**
-- `IReactor` is a **marker interface** — no methods to implement
-- Event dispatch is by first-parameter type; method name can be anything descriptive
-- `EventContext` is optional — omit if event metadata is not needed
-- Reactors MUST be idempotent — they may be called more than once for the same event
-- Do not query the read model inside the reactor — use the event data directly
-- To produce new events, inject `ICommandPipeline` and execute a command — never use `IEventLog` directly
-
-## Translation pattern
-
-If the reactor adapts events by triggering commands in another slice:
-
-```csharp
-public class StockKeeping(IStockKeeper stockKeeper, ICommandPipeline commandPipeline) : IReactor
-{
-    /// <summary>
-    /// Decreases stock when a book is reserved.
-    /// </summary>
-    /// <param name="event">The event.</param>
-    /// <param name="context">The event context.</param>
-    public async Task BookReserved(BookReserved @event, EventContext context) =>
-        await commandPipeline.Execute(new DecreaseStock(@event.Isbn, await stockKeeper.GetStock(@event.Isbn)));
-}
-```
-
-## After creating the file
-
-Run `dotnet build` and fix all errors before completing.
diff --git a/.github/prompts/new-vertical-slice.prompt.md b/.github/prompts/new-vertical-slice.prompt.md
deleted file mode 100644
index fc1b21c..0000000
--- a/.github/prompts/new-vertical-slice.prompt.md
+++ /dev/null
@@ -1,75 +0,0 @@
----
-agent: agent
-description: Scaffold a complete vertical slice (backend + specs + frontend) for a Cratis-based project.
----
-
-# New Vertical Slice
-
-I need you to implement a complete **vertical slice** for a Cratis-based project.
-
-## Inputs
-
-Please provide (or I'll ask):
-- **Feature name** — e.g. `Projects`
-- **Slice name** — e.g. `Registration`, `Listing`, `Removal`
-- **Slice type** — one of:
-  - `State Change` — command that appends events and changes read models
-  - `State View` — query that reads from a read model
-  - `Automation` — background reactor triggered by events
-  - `Translation` — transforms/enriches events into other events
-- **Domain description** — what the slice does in plain language
-- **Properties** — fields on the command/query and their types
-
-## What I want produced
-
-Follow `.github/instructions/vertical-slices.instructions.md` exactly.
-
-### Phase 1 — Backend (delegate to `backend-developer` agent)
-
-For a **State Change** slice, produce (in this order):
-1. Concept types (if new strongly-typed IDs are needed)
-2. Command `record` with `Handle()` method, validation, and business rules
-3. Event `record` with `[EventType]` attribute (no arguments)
-4. Read model `record` with `[ReadModel]` and model-bound projection attributes (`[FromEvent<T>]`, `[Key]`, etc.)
-   - Use fluent `IProjectionFor<T>` only when model-bound attributes don't fit
-
-For a **State View** slice, produce:
-1. Read model `record` with `[ReadModel]`, model-bound projection attributes (`[FromEvent<T>]`, `[Key]`, etc.), and a static query method
-   - Use fluent `IProjectionFor<T>` only when model-bound attributes don't fit
-
-Run `dotnet build` after creating the `.cs` file. Fix all errors before proceeding.
-
-### Phase 2 — Specs (delegate to `spec-writer` agent, State Change slices only)
-
-For each command, write specs covering:
-- Happy path (command succeeds, correct event appended)
-- Each validation failure
-- Each business rule violation
-- Each constraint violation
-
-Run `dotnet test` and fix all failures before proceeding.
-
-### Phase 3 — Frontend (delegate to `frontend-developer` agent)
-
-1. Create React component(s) in `Features/<Feature>/<Slice>/`
-2. Use auto-generated proxy types from `dotnet build`
-3. Use `CommandDialog` for command-based dialogs
-4. Update the feature composition page
-5. Update routing if a new page is introduced
-
-Run `yarn lint` and `npx tsc -b`. Fix all errors before proceeding.
-
-### Phase 4 — Quality Gates
-
-- `dotnet build` — zero errors/warnings
-- `dotnet test` — zero failures
-- `yarn lint` — zero errors
-- `npx tsc -b` — zero errors
-
-## Constraints
-
-- Namespace: `<NamespaceRoot>.<Feature>.<Slice>` — detect `<NamespaceRoot>` from existing source files
-- No abbreviations in TypeScript
-- No hard-coded colours — PrimeReact CSS variables only
-- Copyright header on every file
-- README.md for complex component folders
diff --git a/.github/prompts/review-pr.prompt.md b/.github/prompts/review-pr.prompt.md
deleted file mode 100644
index 36d7c58..0000000
--- a/.github/prompts/review-pr.prompt.md
+++ /dev/null
@@ -1,94 +0,0 @@
----
-agent: agent
-description: Review a pull request against all Cratis project standards and produce a structured review report.
----
-
-# Review Pull Request
-
-I need a structured **code review** of a pull request against all Cratis project standards.
-
-## Inputs
-
-- **PR number or branch name** — so I can examine the changed files
-- **Scope** — which repos / projects are affected
-
-## Review process
-
-### Step 1 — Gather context
-
-- List all changed files in the PR
-- Read each changed file in full
-- Identify the slice type(s) affected (State Change / State View / Automation / Translation)
-
-### Step 2 — Architecture review
-
-Delegate to the `code-reviewer` agent with all changed files.
-
-Check against:
-- `.github/instructions/vertical-slices.instructions.md`
-- `.github/instructions/csharp.instructions.md`
-- `.github/instructions/typescript.instructions.md`
-- `.github/instructions/components.instructions.md`
-- `.github/instructions/concepts.instructions.md`
-- `.github/copilot-instructions.md`
-
-### Step 3 — Security review
-
-Delegate to the `security-reviewer` agent with all changed files.
-
-Check all security categories:
-- Input validation & injection
-- Auth / authz
-- Sensitive data exposure
-- Secrets & configuration
-- Dependency & serialisation safety
-- Event sourcing specifics
-- Frontend attack surface
-
-### Step 4 — Spec coverage
-
-- Confirm every State Change command has specs covering happy path + failure cases
-- Confirm `dotnet test` output shows no regressions (if available)
-
-### Step 5 — Documentation
-
-- If a new public API, feature, or component was added, confirm documentation was updated
-- If a breaking change was introduced, confirm it is called out in the PR description
-
----
-
-## Output format
-
-```
-## Pull Request Review — #<number>
-
-### Summary
-<2–3 sentence overview of what the PR does>
-
-### Architecture — ✅ / ⚠️ / ❌
-<findings from code-reviewer>
-
-### Security — ✅ / ⚠️ / ❌
-<findings from security-reviewer>
-
-### Spec Coverage — ✅ / ⚠️ / ❌
-<list of commands checked and whether specs exist>
-
-### Documentation — ✅ / ⚠️ / ❌
-<notes on docs updates>
-
-### Overall verdict
-**✅ Approved** / **⚠️ Approved with comments** / **❌ Changes requested**
-
-**Blocking issues** (must fix before merge):
-1. …
-
-**Suggestions** (non-blocking):
-1. …
-```
-
-## Conventions reminder
-
-- Blocking issues = anything that violates a `MUST` rule in the instructions
-- Suggestions = style improvements, optional enhancements, or `SHOULD` rules
-- Be specific: always include file, line number, and corrected code for blocking issues
diff --git a/.github/prompts/scaffold-feature.prompt.md b/.github/prompts/scaffold-feature.prompt.md
deleted file mode 100644
index fcd83bc..0000000
--- a/.github/prompts/scaffold-feature.prompt.md
+++ /dev/null
@@ -1,91 +0,0 @@
----
-agent: agent
-description: Scaffold a new feature — folder structure, composition page, routing, and first slice skeleton.
----
-
-# Scaffold a Feature
-
-I need to scaffold a **complete new feature** in a Cratis-based project.
-This creates the folder structure, composition page, and routing — ready for slices to be added.
-
-## Inputs
-
-- **Feature name** — PascalCase, e.g. `Projects`, `Invoices`, `UserManagement`
-- **Route path** — URL path for the feature page, e.g. `/projects`
-- **Navigation label** — Text shown in the sidebar/navigation, e.g. `Projects`
-- **Navigation icon** — Name from `react-icons/md`, e.g. `MdFolderOpen`
-- **First slice** — Optional: describe one initial slice to implement after scaffolding (can also be done later with `new-vertical-slice.prompt.md`)
-
-## What to produce
-
-### 1 — Feature folder
-
-```
-Features/<Feature>/
-├── <Feature>.tsx          ← composition page
-├── <Feature>.css          ← feature-level styles (may be empty)
-└── index.ts               ← re-exports the composition page
-```
-
-### 2 — Composition page (`<Feature>.tsx`)
-
-```tsx
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-
-import { Page } from '../../Core/Page';
-
-export const <Feature> = () => {
-    return (
-        <Page title="<NavigationLabel>">
-            {/* Slices rendered here */}
-        </Page>
-    );
-};
-```
-
-### 3 — Update routing
-
-Add the feature to the application router. Locate the router configuration (typically `App.tsx` or a `routes.ts` file) and add:
-
-```tsx
-import { <Feature> } from './Features/<Feature>';
-
-// Inside the route definitions:
-{ path: '<route-path>', element: <<Feature> /> }
-```
-
-### 4 — Update navigation
-
-Locate the navigation/sidebar configuration and add:
-
-```tsx
-import * as mdIcons from 'react-icons/md';
-
-{
-    label: '<NavigationLabel>',
-    icon: mdIcons.<NavigationIcon>,
-    url: '<route-path>'
-}
-```
-
-### 5 — `index.ts`
-
-```typescript
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-
-export { <Feature> } from './<Feature>';
-```
-
-## Validation
-
-After scaffolding, run:
-- `yarn lint` — zero errors
-- `npx tsc -b` — zero errors
-- Confirm the new route renders an empty page without runtime errors
-
-## Next step
-
-Once scaffolding is complete, use `new-vertical-slice.prompt.md` to add slices to the feature.
-Each slice's component will be imported and composed into `<Feature>.tsx`.
diff --git a/.github/prompts/ship-changes.prompt.md b/.github/prompts/ship-changes.prompt.md
deleted file mode 100644
index fb4b8eb..0000000
--- a/.github/prompts/ship-changes.prompt.md
+++ /dev/null
@@ -1,19 +0,0 @@
----
-agent: agent
-description: >
-  Ship local changes: create a branch, make logical commits, push, open and
-  label a PR with a proper description, merge it, and delete the branch.
----
-
-# Ship Changes
-
-Ship the current local modifications to `main` through the standard
-branch → commits → PR → merge → cleanup workflow.
-
-## Inputs
-
-- **What changed** — brief description of the work (used for branch name and PR title)
-- **Label** — `patch`, `minor`, or `major`, or omit entirely if no label should be applied
-- **Related issue** — optional GitHub issue number; if unknown, search first
-
-Load and follow the full instructions from the `ship-changes` skill.
diff --git a/.github/prompts/write-documentation.prompt.md b/.github/prompts/write-documentation.prompt.md
deleted file mode 100644
index 675f0a9..0000000
--- a/.github/prompts/write-documentation.prompt.md
+++ /dev/null
@@ -1,48 +0,0 @@
----
-agent: agent
-description: "Write DocFX documentation following the Diataxis framework."
----
-
-# Write Documentation
-
-I need you to write **DocFX documentation** for a feature, component, or concept using the **Diataxis framework**.
-
-## Inputs
-
-- **Subject** — what to document (e.g. "the Projects feature", "the CommandDialog component", "the ConceptAs type")
-- **Document type** — one of:
-  - **Tutorial** — step-by-step lesson for newcomers
-  - **How-to Guide** — recipe for accomplishing a specific task
-  - **Reference** — exhaustive technical description (APIs, attributes, configuration)
-  - **Explanation** — discussion of concepts, trade-offs, and architecture
-- **Target audience** — e.g. new developer, experienced contributor, framework consumer
-- **Existing source files** — paste or reference the code to document
-
-## Instructions
-
-Follow `.github/instructions/documentation.instructions.md` and the `write-documentation` skill exactly.
-
-### Workflow
-
-1. **Clarify** — Confirm the document type, audience, goal, and scope. Ask if anything is ambiguous.
-2. **Propose structure** — Present an outline (headings + one-line descriptions). Wait for approval.
-3. **Write** — Produce full DocFX-compatible Markdown with correct `toc.yml` entries.
-
-### File structure
-
-For a new topic under an existing section, create a folder with an `index.md` and `toc.yml`. Update the parent `toc.yml` to include the new topic.
-
-### Writing style
-
-- Active voice, present tense, second person
-- Lead with the most important information
-- Every code example must be complete and correct
-- Link to related topics using relative paths
-- Use Mermaid diagrams for architecture, sequence flows, and state transitions
-- Do not document internal implementation details in user-facing docs
-
-### Validation
-
-- Verify `toc.yml` is valid YAML
-- Verify all `href` values point to files that exist
-- Verify all Mermaid blocks are valid
diff --git a/.github/prompts/write-specs.prompt.md b/.github/prompts/write-specs.prompt.md
deleted file mode 100644
index 188e045..0000000
--- a/.github/prompts/write-specs.prompt.md
+++ /dev/null
@@ -1,71 +0,0 @@
----
-agent: agent
-description: Write comprehensive BDD specs for an existing vertical slice command or query.
----
-
-# Write Specs
-
-I need you to write **comprehensive specs** for an existing vertical slice.
-
-## What to provide
-
-Paste or reference the slice file (`.cs`) you want covered.
-
-## Instructions
-
-Follow `.github/instructions/specs.csharp.instructions.md` and `.github/instructions/specs.instructions.md`.
-
-### For a State Change slice (command)
-
-Write a spec class for **each meaningful outcome**:
-
-1. **Happy path** — command succeeds, expected event is appended, sequence number advanced.
-2. **Each validation failure** — one spec per validation rule.
-3. **Each business rule violation** — one spec per condition in `Handle()` that inspects a ReadModel argument (DCB pattern).
-4. **Each constraint violation** — one spec per `IConstraint` (e.g. uniqueness).
-
-### Structure
-
-```
-Features/<Feature>/<Slice>/
-└── when_<verb_phrase>/
-    ├── and_<happy_scenario>.cs
-    ├── and_<failure_scenario>.cs
-    └── and_<other_scenario>.cs
-```
-
-### C# spec shape
-
-```csharp
-using context = <NamespaceRoot>.<Feature>.<Slice>.when_<behavior>.and_<scenario>.context;
-
-namespace <NamespaceRoot>.<Feature>.<Slice>.when_<behavior>;
-
-[Collection(ChronicleCollection.Name)]
-public class and_<scenario>(context context) : Given<context>(context)
-{
-    public class context(ChronicleOutOfProcessFixture fixture) : given.an_http_client(fixture)
-    {
-        public CommandResult<object>? Result;
-
-        async Task Establish() { /* optional: seed events */ }
-
-        async Task Because()
-        {
-            Result = await Client.ExecuteCommand<<Command>>(
-                "/api/<feature>/<action>",
-                new <Command>(...));
-        }
-    }
-
-    [Fact] void should_<expected_result>() => ...;
-}
-```
-
-### Assertions
-
-Use `ShouldBeFalse()`, `ShouldBeTrue()`, `ShouldEqual()`, `ShouldHaveTailSequenceNumber()` — never raw `Assert.*`.
-
-## Validation
-
-Run `dotnet test` after writing specs. Fix any failures before completing.
diff --git a/.github/skills b/.github/skills
new file mode 120000
index 0000000..6838a11
--- /dev/null
+++ b/.github/skills
@@ -0,0 +1 @@
+../.ai/skills
\ No newline at end of file
diff --git a/.github/skills/add-business-rule/SKILL.md b/.github/skills/add-business-rule/SKILL.md
deleted file mode 100644
index 65a69ab..0000000
--- a/.github/skills/add-business-rule/SKILL.md
+++ /dev/null
@@ -1,138 +0,0 @@
----
-name: add-business-rule
-description: Use this skill when asked to add a validation rule, business rule, or uniqueness constraint to an existing command in a Cratis-based project.
----
-
-Add a business rule or event-store constraint to an existing command.
-
-## Choose the right mechanism
-
-| Scenario                                           | Use                                           |
-|----------------------------------------------------|-----------------------------------------------|
-| Single-event uniqueness (most cases)               | `[Unique]` attribute on the event type        |
-| Multi-event uniqueness or needs `RemovedWith`      | `IConstraint`                                 |
-| Rule requiring Chronicle event-sourced state       | ReadModel as `Handle()` parameter (DCB)       |
-| Simple sync invariant (format, range, required)    | `CommandValidator<T>`                         |
-
-## Business Rules via DCB (ReadModel as argument)
-
-Use when the rule depends on **Chronicle event-sourced state** (e.g. current count, accumulated value).
-This is the DCB (Dynamic Consistency Boundary) pattern: add a **read model parameter** to the `Handle()` method.
-The framework fetches and injects the current read model snapshot before `Handle()` runs.
-
-The read model must already exist in the slice (decorated with `[ReadModel]` and a model-bound projection).
-If it doesn't, add it first — see the `new-vertical-slice` skill.
-
-```csharp
-[Command]
-public record <Command>(<KeyProperty> <Key>, ...)
-{
-    /// <summary>
-    /// Handles the command.
-    /// </summary>
-    /// <param name="readModelParam">The current state.</param>
-    /// <returns>The resulting event.</returns>
-    public <Event> Handle(<ReadModel> <readModelParam>)
-    {
-        if (<readModelParam>.<StateProperty> <violatesRule>)
-            throw new <ViolationException>(<meaningful message>);
-
-        return new <Event>(...);
-    }
-}
-```
-
-**Example — limit items in cart to 3:**
-
-```csharp
-[Command]
-public record AddItemToCart(CartId CartId, ItemId ItemId)
-{
-    /// <summary>
-    /// Adds the item; throws if the cart already holds 3 items.
-    /// </summary>
-    /// <param name="cart">The current cart summary.</param>
-    /// <returns>The <see cref="ItemAddedToCart"/> event.</returns>
-    /// <exception cref="CartIsFull">Thrown when the cart already contains the maximum number of items.</exception>
-    public ItemAddedToCart Handle(CartSummary cart)
-    {
-        if (cart.ItemCount >= 3)
-            throw new CartIsFullException(CartId);
-
-        return new ItemAddedToCart(ItemId);
-    }
-}
-```
-
-**Key rules:**
-- The read model parameter type must match a `[ReadModel]`-decorated type in the same feature.
-- The framework resolves the instance using the same event-source key as the command.
-- Throw a **custom exception** (never a built-in one) to signal violation — the framework converts it to an error result.
-- One parameter per logical read model; if you need multiple reads, use multiple parameters.
-
-## Event-Store Constraints — `[Unique]` attribute (preferred)
-
-For the common case of enforcing uniqueness on a single event type, adorn the event type class or one of its properties with `[Unique]`. No separate constraint class is needed.
-
-**Event-type uniqueness** — only one event of this type per event source:
-
-```csharp
-[EventType]
-[Unique(message: "A project with this name already exists.")]
-public record ProjectCreated(string Name, string Description);
-```
-
-**Property uniqueness** — the value of a specific property must be unique:
-
-```csharp
-[EventType]
-public record UserRegistered([Unique(name: "UniqueEmail", message: "Email already registered.")] string Email, string DisplayName);
-```
-
-**Releasing a constraint** — apply `[RemoveConstraint]` to the event that deletes the domain object:
-
-```csharp
-[EventType]
-[RemoveConstraint("UniqueEmail")]
-public record UserRemoved(UserId UserId);
-```
-
-Multiple attributes can be stacked to release more than one constraint from the same event type.
-
-Constraints are discovered and enforced automatically — no registration or attribute on the command needed.
-
-## Event-Store Constraints — `IConstraint` (advanced)
-
-Use `IConstraint` when the uniqueness rule spans event types with different property names, or when a `RemovedWith` event must release the constraint.
-
-```csharp
-public class Unique<PropertyName> : IConstraint
-{
-    public void Define(IConstraintBuilder builder) =>
-        builder.Unique(unique =>
-            unique
-                .On<<EventType>>(e => e.<UniqueProperty>)
-                .RemovedWith<<RemovedEventType>>());  // omit if there is no remove event
-}
-```
-
-**Example — unique project name with removal:**
-
-```csharp
-public class UniqueProjectName : IConstraint
-{
-    public void Define(IConstraintBuilder builder) =>
-        builder.Unique(unique =>
-            unique
-                .On<ProjectRegistered>(e => e.Name)
-                .RemovedWith<ProjectRemoved>());
-}
-```
-
-Constraints are discovered and enforced automatically — no registration or attribute on the command needed.
-
-## After adding
-
-1. Add a spec for the failure case — see `write-specs` skill
-2. Run `dotnet build` and `dotnet test`
-3. Fix all failures before completing
diff --git a/.github/skills/add-concept/SKILL.md b/.github/skills/add-concept/SKILL.md
deleted file mode 100644
index 18c57f3..0000000
--- a/.github/skills/add-concept/SKILL.md
+++ /dev/null
@@ -1,84 +0,0 @@
----
-name: add-concept
-description: Use this skill when asked to create a strongly-typed domain identifier or value (such as ProjectId, AuthorName, InvoiceNumber) in a Cratis-based project. Produces a ConceptAs<T> record with the correct conversions and sentinel values.
----
-
-Create a strongly-typed Concept that wraps a primitive domain value.
-
-## Never use raw primitives in domain models
-
-Replace `Guid`, `string`, `int`, etc. with a `ConceptAs<T>` record whenever the value has domain meaning.
-
-## Template
-
-```csharp
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-
-using Cratis.Concepts;
-
-namespace <NamespaceRoot>.<Feature>;
-
-/// <summary>
-/// Represents the identity of a <description>.
-/// </summary>
-/// <param name="Value">The underlying <type> value.</param>
-public record <ConceptName>(<UnderlyingType> Value) : ConceptAs<<UnderlyingType>>(Value)
-{
-    /// <summary>
-    /// Represents an unset <ConceptName>.
-    /// </summary>
-    public static readonly <ConceptName> NotSet = new(<emptyValue>);
-
-    /// <summary>
-    /// Implicitly converts a <UnderlyingType> to a <ConceptName>.
-    /// </summary>
-    public static implicit operator <ConceptName>(<UnderlyingType> value) => new(value);
-}
-```
-
-## Add when backed by Guid
-
-```csharp
-    /// <summary>
-    /// Creates a new <ConceptName> with a unique value.
-    /// </summary>
-    /// <returns>A new <ConceptName>.</returns>
-    public static <ConceptName> New() => new(Guid.NewGuid());
-```
-
-## Add when used as an event-source ID
-
-```csharp
-    /// <summary>
-    /// Implicitly converts a <ConceptName> to an <see cref="EventSourceId"/>.
-    /// </summary>
-    /// <param name="id">The <ConceptName> to convert.</param>
-    public static implicit operator EventSourceId(<ConceptName> id) => new(id.Value.ToString());
-```
-
-## Empty/sentinel values
-
-| Underlying type | Use             |
-|-----------------|-----------------|
-| `Guid`          | `Guid.Empty`    |
-| `string`        | `string.Empty`  |
-| `int`           | `0`             |
-| `long`          | `0L`            |
-
-## Placement rules
-
-- Do NOT create a `Concepts/` folder
-- Place the file in the folder that **semantically owns** the concept
-  - `ProjectId` → `Features/Projects/`
-  - Shared cross-feature concepts → project source root
-
-## Checklist
-
-- [ ] Inherits `ConceptAs<T>`
-- [ ] Has a `static readonly NotSet` (or `Empty`) sentinel
-- [ ] Has implicit conversion **from** the primitive
-- [ ] Has implicit conversion **to** `EventSourceId` if used as event-source key
-- [ ] Has `New()` factory if `Guid`-backed
-- [ ] Copyright header present
-- [ ] `dotnet build` passes
diff --git a/.github/skills/add-ef-migration/SKILL.md b/.github/skills/add-ef-migration/SKILL.md
deleted file mode 100644
index 1ba16fe..0000000
--- a/.github/skills/add-ef-migration/SKILL.md
+++ /dev/null
@@ -1,168 +0,0 @@
----
-name: add-ef-migration
-description: Use this skill when asked to add a database table, column, relationship, or other schema change via Entity Framework Core in a Cratis-based project.
----
-
-Add or update an Entity Framework Core schema change with a hand-written migration.
-
-> **Always read `.github/instructions/efcore.instructions.md` first.** It is the source of truth for project structure, base types, column helpers, and migration conventions.
-
-## Pre-flight — Understand the project split
-
-Before writing anything, locate these three projects:
-
-| Project | What lives there |
-|---------|-----------------|
-| **Database** | Migrations only — `WellKnownTables.cs` + versioned migration files |
-| **Core** | Entities and feature DbContexts (co-located with features) |
-| **Infrastructure** | DbContext registration, migration runner, cross-cutting EF setup |
-
-**Critical:** `Database` must NEVER import from `Core`. Migrations reference only `WellKnownTables` constants and EF migration types.
-
-## Step 1 — Update or create the entity
-
-Add, rename, or remove properties on the entity `record` in the **Core** project, co-located with its feature:
-
-```
-Features/Missions/StartupPhase/
-├── StartupPhase.cs            ← entity record
-├── StartupPhaseDbContext.cs   ← feature DbContext
-└── ...
-```
-
-## Step 2 — Update or create the feature DbContext
-
-Use the Cratis Arc base types — never inherit directly from `DbContext`:
-
-- **`ReadOnlyDbContext`** — for all read model / projection contexts (the vast majority)
-- **`BaseDbContext`** — only for writable contexts that own state
-
-```csharp
-public class StartupPhaseDbContext(DbContextOptions<StartupPhaseDbContext> options)
-    : ReadOnlyDbContext(options)
-{
-    public DbSet<StartupPhase> StartupPhases => Set<StartupPhase>();
-}
-```
-
-Create one focused DbContext per feature — never a "god context" with unrelated entities.
-
-## Step 3 — Add the table name to WellKnownTables
-
-If this is a new table, add the constant to `Database/WellKnownTables.cs` **before** writing the migration:
-
-```csharp
-public static class WellKnownTables
-{
-    public const string StartupPhases = "StartupPhases";
-}
-```
-
-## Step 4 — Write the migration by hand
-
-Migrations are **hand-written**, not generated by `dotnet ef migrations add`. Place the file in the **Database** project using versioned naming:
-
-```
-Database/
-├── Missions/
-│   ├── v1_0_0.cs
-│   └── v1_1_0.cs
-└── WellKnownTables.cs
-```
-
-Use the `v{major}_{minor}_{patch}.cs` naming pattern. The namespace must match the folder:
-
-```csharp
-namespace Database.Missions;
-
-public class v1_1_0 : Migration
-{
-    /// <summary>
-    /// Apply the schema change.
-    /// </summary>
-    /// <param name="migrationBuilder">The migration builder.</param>
-    protected override void Up(MigrationBuilder migrationBuilder)
-    {
-        migrationBuilder.AddColumn<string>(
-            name: "Description",
-            table: WellKnownTables.Missions,
-            type: migrationBuilder.StringColumnType(maxLength: 1000),
-            nullable: true);
-    }
-
-    /// <summary>
-    /// Reverse the schema change.
-    /// </summary>
-    /// <param name="migrationBuilder">The migration builder.</param>
-    protected override void Down(MigrationBuilder migrationBuilder)
-    {
-        migrationBuilder.DropColumn(
-            name: "Description",
-            table: WellKnownTables.Missions);
-    }
-}
-```
-
-### Cross-database column helpers (mandatory)
-
-Always use the Cratis Arc `MigrationBuilder` extension helpers — never raw `table.Column<T>()`:
-
-| Helper | Use for |
-|--------|---------|
-| `table.StringColumn(migrationBuilder)` | Text / varchar columns |
-| `table.GuidColumn(migrationBuilder)` | UUID / GUID columns |
-| `table.NumberColumn<T>(migrationBuilder)` | Integer, long, or numeric columns |
-| `table.DateTimeOffsetColumn(migrationBuilder)` | Timestamps with timezone |
-
-Example `CreateTable`:
-
-```csharp
-migrationBuilder.CreateTable(
-    name: WellKnownTables.StartupPhases,
-    columns: table => new
-    {
-        Id = table.StringColumn(migrationBuilder, nullable: false),
-        Title = table.StringColumn(migrationBuilder, maxLength: 200, nullable: false),
-        ResourceId = table.NumberColumn<int>(migrationBuilder, nullable: true),
-        DispatchTime = table.DateTimeOffsetColumn(migrationBuilder),
-        UrgencyId = table.GuidColumn(migrationBuilder)
-    },
-    constraints: table =>
-    {
-        table.PrimaryKey("PK_StartupPhases", x => x.Id);
-    });
-```
-
-## Step 5 — Register the DbContext (if new)
-
-In the **Infrastructure** project, ensure the context is registered. Read-only contexts are auto-discovered:
-
-```csharp
-services.AddReadModelDbContextsWithConnectionStringFromAssemblies(
-    connectionString,
-    configureOptions,
-    [Assembly.GetExecutingAssembly()]);
-```
-
-Writable contexts need explicit registration:
-
-```csharp
-services.AddDbContextWithConnectionString<DeviceStateDbContext>(connectionString);
-```
-
-## Step 6 — Update specs
-
-Integration specs using in-memory SQLite pick up schema changes automatically via `context.Database.EnsureCreated()`. If specs break, check that the fixture uses the correct connection string and that the new migration is in the `Database` assembly.
-
-## Step 7 — Validate
-
-Run `dotnet build` and `dotnet test`. Fix all failures before completing.
-
-## Key rules
-
-- **Never** use `dotnet ef migrations add` — all migrations are hand-written
-- **Never** use `dotnet ef database update` — use the custom runner: `await app.ApplyAllMigrations(connectionString)`
-- **Never** hardcode a provider (`UseSqlite`, `UseNpgsql`) — use `UseDatabaseFromConnectionString`
-- **Never** mutate state directly through a DbContext — all writes flow through Chronicle events and projections
-- **Always** use `WellKnownTables` constants — never magic strings for table names
-- **Always** use cross-database column helpers — never raw `table.Column<T>()`
diff --git a/.github/skills/add-projection/SKILL.md b/.github/skills/add-projection/SKILL.md
deleted file mode 100644
index 3a988df..0000000
--- a/.github/skills/add-projection/SKILL.md
+++ /dev/null
@@ -1,92 +0,0 @@
----
-name: add-projection
-description: Use this skill when asked to add a Chronicle projection to a Cratis-based project. Favor model-bound projections by default, and only fall back to declarative/fluent `IProjectionFor<T>` projections when model-bound attributes cannot express the behavior cleanly. Enforces the AutoMap-first rule and Chronicle-specific join semantics.
----
-
-Add a Chronicle **projection** that populates a read model from events.
-
-> For **reactors** (automation / translation), see the `add-reactor` skill instead.
-
-## Projection — Model-Bound (preferred)
-
-Put projection metadata directly on the read model using attributes. No separate class needed.
-
-```csharp
-[ReadModel]
-[FromEvent<SomeEventHappened>]              // auto-maps all matching property names
-public record <ReadModelName>(
-    [Key] <IdType> Id,                      // marks the primary key
-    <PropType> <PropName>)                  // auto-mapped from SomeEventHappened
-{
-    public static ISubject<IEnumerable<<ReadModelName>>> All(IMongoCollection<<ReadModelName>> collection) =>
-        collection.Observe();
-}
-```
-
-**Attribute reference:**
-| Attribute | Purpose |
-|-----------|---------|
-| `[FromEvent<T>]` | Auto-maps all matching property names (equivalent to `.AutoMap().From<T>()`) |
-| `[FromEvent<T>(key: nameof(T.Prop))]` | Same, but uses `Prop` as the read model key instead of EventSourceId |
-| `[Key]` | Marks the primary key property |
-| `[SetFrom<T>(nameof(T.Prop))]` | Explicitly maps one property from event T |
-| `[AddFrom<T>(nameof(T.Prop))]` | Adds event property value to the read model property |
-| `[SubtractFrom<T>(nameof(T.Prop))]` | Subtracts event property value |
-| `[ChildrenFrom<T>(key: nameof(T.Prop))]` | Projects into a nested child collection |
-| `[Join<T>(on: nameof(Prop), eventPropertyName: nameof(T.EProp))]` | Joins data from a related event |
-| `[RemovedWith<T>]` | Marks the instance as removed when event T is appended |
-
-**Critical rules:**
-- Joins must be on Chronicle **events** — NEVER join on a read model type
-- If property names between event and read model match, `[FromEvent<T>]` alone is sufficient
-- Child types also support all attributes recursively
-
-## Projection — Fluent / declarative (fallback for complex cases)
-
-Use `IProjectionFor<T>` only when the projection logic is too complex for model-bound attributes or would become less clear if forced into attributes.
-
-```csharp
-public class <Name>Projection : IProjectionFor<<ReadModel>>
-{
-    public void Define(IProjectionBuilderFor<<ReadModel>> builder) =>
-        builder
-            .From<SomeEventHappened>(b =>
-                b.UsingKey(e => e.SomeId))
-            .RemovedWith<SomeThingRemoved>();
-}
-```
-
-**Critical rules:**
-- AutoMap is on by default — just call `.From<>()` directly. Only call `.AutoMap()` if you previously used `.NoAutoMap()`.
-- Joins are on Chronicle **events** only — NEVER join on the read model
-- There is NO `Identifier` / `ProjectionId` property — do not add one
-
-## After creating
-
-Run `dotnet build`. Fix all errors before completing.
-
-## Appended event metadata and filtering
-
-Chronicle correlates appended metadata in two different ways:
-
-- **Projections** select input through event types, joins, and event sequence configuration
-- **Reducers and reactors** can additionally filter by appended tags, event source type, and event stream type
-
-Use append metadata like this:
-
-```csharp
-await eventLog.Append(
-    EventSourceId.New(),
-    new OrderPlaced(42m),
-    eventStreamType: "fulfillment",
-    eventSourceType: "order",
-    tags: ["priority"]);
-```
-
-If you need metadata-based filtering for downstream processing, pair the projection with a reducer or reactor using `[FilterEventsByTag]`, `[EventSourceType]`, or `[EventStreamType]`. Projection `[Tag]` and `[Tags]` attributes label the projection definition; they do not filter appended events.
-
-For examples, see `Documentation/events/filtering/`.
-
----
-
-For the full model-bound projection attribute reference and fluent builder API, see [references/CHRONICLE-API.md](references/CHRONICLE-API.md).
diff --git a/.github/skills/add-projection/references/CHRONICLE-API.md b/.github/skills/add-projection/references/CHRONICLE-API.md
deleted file mode 100644
index 2768ef3..0000000
--- a/.github/skills/add-projection/references/CHRONICLE-API.md
+++ /dev/null
@@ -1,169 +0,0 @@
-# Chronicle Projection API
-
----
-
-## Model-Bound Projections (preferred)
-
-Apply attributes directly to read model types. No separate projection class needed.
-
-### Class-level attributes
-
-#### `[FromEvent<T>]`
-Auto-maps all properties with matching names from event `T` to the read model.
-Equivalent to `.AutoMap().From<T>()` in the fluent API.
-
-```csharp
-[ReadModel]
-[FromEvent<AccountOpened>]
-public record Account(
-    [Key] Guid Id,
-    string Name,          // auto-mapped from AccountOpened.Name
-    decimal Balance);     // auto-mapped from AccountOpened.Balance
-```
-
-#### `[FromEvent<T>(key: nameof(T.PropertyName))]`
-Same as above, but uses the specified event property as the read model key instead of EventSourceId.
-
-```csharp
-[ReadModel]
-[FromEvent<OrderPlaced>(key: nameof(OrderPlaced.OrderId))]
-public record Order([Key] Guid Id, decimal Total);
-```
-
-Multiple `[FromEvent<T>]` attributes are supported for different events:
-
-```csharp
-[FromEvent<AccountOpened>]
-[FromEvent<AccountRenamed>]
-public record Account([Key] Guid Id, string Name, decimal Balance);
-```
-
----
-
-### Property-level attributes
-
-#### `[Key]`
-Marks the primary key property.
-
-#### `[SetFrom<T>]` / `[SetFrom<T>(nameof(T.Prop))]`
-Maps a specific property from event `T`. Omit the property name if the names match.
-
-#### `[SetFrom<T>]` with multiple events
-Apply multiple attributes to update the same property from different events.
-
-```csharp
-[SetFrom<AccountOpened>(nameof(AccountOpened.AccountName))]
-[SetFrom<AccountRenamed>(nameof(AccountRenamed.NewName))]
-string Name
-```
-
-#### `[AddFrom<T>(nameof(T.Prop))]`
-Adds the event property value to the read model property (accumulates).
-
-#### `[SubtractFrom<T>(nameof(T.Prop))]`
-Subtracts the event property value.
-
-#### `[ChildrenFrom<T>(key: nameof(T.ChildId))]`
-Projects into a nested child collection. The child type also supports all attributes recursively.
-
-```csharp
-public record Order(
-    [Key] Guid Id,
-    [ChildrenFrom<LineItemAdded>(key: nameof(LineItemAdded.ItemId))]
-    IEnumerable<LineItem> Items);
-
-public record LineItem(
-    [Key] Guid Id,
-    string ProductName,   // auto-mapped
-    int Quantity,         // auto-mapped
-    decimal Price);       // auto-mapped
-```
-
-#### `[Join<T>(on: nameof(Prop), eventPropertyName: nameof(T.EProp))]`
-Joins data from a related event (keyed by a different entity).
-
-```csharp
-[Join<CustomerRegistered>(
-    on: nameof(CustomerId),
-    eventPropertyName: nameof(CustomerRegistered.Name))]
-string CustomerName
-```
-
-#### `[RemovedWith<T>]`
-Marks the instance removed when event `T` is appended.
-
----
-
-## Fluent Projections (alternative for complex cases)
-
-Use `IProjectionFor<T>` when model-bound attributes don't fit the complexity of the projection.
-
-### Basic structure
-
-```csharp
-public class MyProjection : IProjectionFor<MyReadModel>
-{
-    public void Define(IProjectionBuilderFor<MyReadModel> builder) =>
-        builder
-            .From<SomeEvent>(...);  // AutoMap is on by default
-}
-```
-
-> **There is NO `Identifier`/`ProjectionId` property** — do not add one.
-
----
-
-### `.AutoMap()`
-
-Maps all event properties to read model properties with matching names automatically.
-**AutoMap is on by default — you only need to call it explicitly if you previously used `.NoAutoMap()`.**
-
----
-
-### `.From<TEvent>(Action<IFromBuilder<TReadModel, TEvent>>)`
-
-Handles an event type. Builder methods:
-
-- `.UsingKey(e => e.PropertyOnEvent)` — sets the key from an event property
-- `.UsingParentKey(e => e.PropertyOnEvent)` — sets key from parent (child projections)
-- `.Set(m => m.Target).To(e => e.Source)` — explicit property mapping
-- `.Set(m => m.Prop).ToValue(literal)` — set to constant value
-- `.Add(m => m.Counter).With(1)` — increment by fixed value
-- `.Subtract(m => m.Counter).With(1)` — decrement by fixed value
-- `.Count(m => m.TotalCount)` — increment count by 1
-
----
-
-### `.Children<TChild>(m => m.ChildCollection, childBuilder => ...)`
-
-Projects into a nested collection.
-
-```csharp
-builder
-    .Children<LineItem>(m => m.LineItems, child =>  // AutoMap is on by default
-        child
-            .From<LineItemAdded>(b =>
-                b.UsingKey(e => e.LineItemId)
-                 .Set(li => li.Description).To(e => e.Description))
-            .RemovedWith<LineItemRemoved>());
-```
-
----
-
-### `.RemovedWith<TEvent>()`
-
-Marks the read model as removed when the specified event is appended.
-
----
-
-### `.Join<TEvent>(Action<IJoinBuilder<TReadModel, TEvent>>)`
-
-Joins data from a different event. The join is always on the **event**, never on the read model.
-
-```csharp
-builder
-    .From<ProjectRegistered>()                      // AutoMap is on by default
-    .Join<OwnerAssigned>(b =>
-        b.On(e => e.ProjectId)
-         .Set(m => m.OwnerName).To(e => e.OwnerName));
-```
diff --git a/.github/skills/add-reactor/SKILL.md b/.github/skills/add-reactor/SKILL.md
deleted file mode 100644
index c51b637..0000000
--- a/.github/skills/add-reactor/SKILL.md
+++ /dev/null
@@ -1,117 +0,0 @@
----
-name: add-reactor
-description: Use this skill when asked to add a Chronicle reactor (automation or translation) to a Cratis-based project. Reactors observe events and produce side effects.
----
-
-Add a Chronicle **reactor** that triggers automation or translation logic in response to events.
-
-> **Always read `.github/instructions/reactors.instructions.md` first.** It is the source of truth for reactor conventions, rules, and patterns.
-
-## Step 1 — Identify the event(s)
-
-Determine which event type(s) the reactor needs to observe. These events must already exist or be created as part of the slice.
-
-## Step 2 — Create the reactor class
-
-Place the reactor in the appropriate feature folder, co-located with the slice it belongs to.
-
-```csharp
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-
-namespace MyApp.Projects.Notifications;
-
-/// <summary>
-/// Sends a notification when a project is registered.
-/// </summary>
-/// <param name="notifications">The notification service.</param>
-public class ProjectRegisteredNotifier(INotificationService notifications) : IReactor
-{
-    /// <summary>
-    /// Reacts to <see cref="Registration.ProjectRegistered"/> events.
-    /// </summary>
-    /// <param name="event">The event.</param>
-    /// <param name="context">The event context.</param>
-    public async Task ProjectRegistered(Registration.ProjectRegistered @event, EventContext context) =>
-        await notifications.Notify($"Project '{@event.Name}' was registered.");
-}
-```
-
-## Step 3 — Follow the critical rules
-
-- **`IReactor` is a marker interface** — no methods to implement. Method dispatch is by first-parameter event type.
-- **Method name** — can be anything descriptive. The name is for readability, not dispatch.
-- **`EventContext`** — optional second parameter. Omit if event metadata is not needed.
-- **Idempotent** — reactors may be called more than once for the same event. Design accordingly.
-- **Use event data directly** — never query the read model back inside the reactor.
-- **Trigger commands for further writes** — inject `ICommandPipeline` and execute a command. Never use `IEventLog` directly.
-
-## Step 4 — Translation pattern (if applicable)
-
-If the reactor adapts events from one slice by triggering commands in another, it is a **Translation** reactor:
-
-```csharp
-/// <summary>
-/// Reacts to reservation events and decreases stock accordingly.
-/// </summary>
-/// <param name="stockKeeper">The stock keeper service.</param>
-/// <param name="commandPipeline">The command pipeline.</param>
-public class StockKeeping(IStockKeeper stockKeeper, ICommandPipeline commandPipeline) : IReactor
-{
-    /// <summary>
-    /// Handles a <see cref="BookReserved"/> event.
-    /// </summary>
-    /// <param name="event">The event.</param>
-    /// <param name="context">The event context.</param>
-    public async Task BookReserved(BookReserved @event, EventContext context) =>
-        await commandPipeline.Execute(new DecreaseStock(@event.Isbn, await stockKeeper.GetStock(@event.Isbn)));
-}
-```
-
-## Step 5 — Filter by appended event metadata (optional)
-
-If the reactor should only observe a subset of appended events, decorate it with filter attributes that match the metadata used when appending:
-
-```csharp
-using Cratis.Chronicle;
-using Cratis.Chronicle.Events;
-
-[FilterEventsByTag("priority")]
-[EventSourceType("order")]
-[EventStreamType("fulfillment")]
-public class PriorityFulfillmentReactor : IReactor
-{
-    public Task OrderPlaced(OrderPlaced @event, EventContext context) => Task.CompletedTask;
-}
-```
-
-These attributes correlate directly to the append call:
-
-```csharp
-await eventLog.Append(
-    EventSourceId.New(),
-    new OrderPlaced(42m),
-    eventStreamType: "fulfillment",
-    eventSourceType: "order",
-    tags: ["priority"]);
-```
-
-- `[FilterEventsByTag]` filters by event tags
-- `[EventSourceType]` filters by the appended event source type
-- `[EventStreamType]` filters by the appended event stream type
-- `[Tag]` and `[Tags]` still label the reactor itself; they do not filter events
-
-For fuller examples, see `Documentation/events/filtering/`.
-
-## Step 6 — Validate
-
-Run `dotnet build`. Fix all errors before completing.
-
-## When to use a Reactor vs a Projection
-
-| Need | Use |
-|------|-----|
-| Populate a queryable read model from events | **Projection** (see `add-projection` skill) |
-| Trigger side effects, send notifications, call external APIs | **Reactor** |
-| Adapt events across slices by triggering commands | **Reactor** (Translation pattern) |
-| Both populate a read model AND trigger side effects | **Both** — one projection + one reactor |
diff --git a/.github/skills/auth-and-identity/SKILL.md b/.github/skills/auth-and-identity/SKILL.md
deleted file mode 100644
index d14f313..0000000
--- a/.github/skills/auth-and-identity/SKILL.md
+++ /dev/null
@@ -1,174 +0,0 @@
----
-name: auth-and-identity
-description: Use this skill when asked to set up authentication, authorization, or identity in a Cratis Arc project — backend, frontend, or both. Covers implementing identity providers, protecting commands/queries with authorization attributes, integrating Microsoft Identity Platform, connecting backend identity to React frontends, and local development testing with generated principals. Also covers customizing IProvideIdentityDetails for enriching identity from databases, blocking users, multi-tenant identity, user preferences, and modifying identity at runtime. Use this whenever the user mentions auth, login, roles, permissions, identity details, user context, protecting endpoints, identity provider customization, or anything related to who the user is and what they can access.
----
-
-# Auth & Identity in Cratis Arc
-
-This skill covers the full auth and identity stack in a Cratis Arc application. Read the relevant reference files below for detailed API usage.
-
-> **Read the relevant instruction files first.** This skill references concepts from the core copilot instructions in `.github/copilot-instructions.md`. If you need details on vertical slices, commands, queries, or proxy generation, consult those instructions.
-
-## Architecture Overview
-
-Identity and auth in Arc follow a cookie-first, convention-based pattern:
-
-```
-Frontend (React)                          Backend (ASP.NET Core)
-─────────────────                         ──────────────────────
-<IdentityProvider>                        app.MapIdentityProvider()
-  └── useIdentity() hook                    └── GET /.cratis/me
-        │                                        │
-        ├─ 1. Read .cratis-identity cookie        │
-        └─ 2. If no cookie → fetch /.cratis/me    │
-                                                  ▼
-                                    AuthenticationMiddleware
-                                      └── IAuthenticationHandler[]
-                                            │ sets HttpContext.User
-                                            ▼
-                                    IIdentityProviderResultHandler
-                                      └── IProvideIdentityDetails.Provide()
-                                            │
-                                            ▼
-                                    IdentityProviderResult
-                                      → JSON response
-                                      → .cratis-identity cookie (base64)
-
-Authorization:
-  [Authorize] / [Roles("Admin")] / [AllowAnonymous]
-    └── AuthorizationEvaluator checks per command/query
-```
-
-**Key design decisions:**
-- The `.cratis-identity` cookie is `HttpOnly=false` so the frontend JavaScript can read it directly — no extra HTTP call needed on page load.
-- Identity details are base64-encoded JSON in the cookie, automatically decoded by the frontend `IdentityProvider`.
-- Only one `IProvideIdentityDetails` implementation is allowed per application (auto-discovered). If none exists, a default provider grants access to everyone.
-- Cratis has its own `[Authorize]`, `[AllowAnonymous]`, and `[Roles]` attributes in `Cratis.Arc.Authorization` — these are distinct from ASP.NET Core's and are evaluated by `AuthorizationEvaluator` in the command and query pipeline.
-
----
-
-## Decision Tree — Which Reference to Read
-
-Use this decision tree to determine which reference file(s) to read based on the user's task:
-
-| User wants to... | Read this reference |
-|---|---|
-| Add identity details to their app | [references/backend-identity.md](references/backend-identity.md) |
-| Customize `IProvideIdentityDetails` (enrich from DB, block users, multi-tenant, preferences) | [references/backend-identity.md](references/backend-identity.md) |
-| Modify identity at runtime (stateless selections, `ModifyDetails`) | [references/backend-identity.md](references/backend-identity.md) |
-| Use Azure AD / Entra ID / Microsoft Identity | [references/authentication.md](references/authentication.md) |
-| Write a custom authentication handler (API key, JWT, etc.) | [references/authentication.md](references/authentication.md) |
-| Protect commands or queries with roles | [references/authorization.md](references/authorization.md) |
-| Set up `[Authorize]`, `[AllowAnonymous]`, or `[Roles]` | [references/authorization.md](references/authorization.md) |
-| Consume identity in React | [references/frontend.md](references/frontend.md) |
-| Use identity in MVVM or vanilla TypeScript | [references/frontend.md](references/frontend.md) |
-| Test identity locally without Azure | [references/local-development.md](references/local-development.md) |
-| Full-stack setup (backend + frontend) | Read all references in order |
-
-**For full-stack tasks, read in this order:**
-1. `references/backend-identity.md` — identity provider and startup
-2. `references/authentication.md` — how users are authenticated
-3. `references/authorization.md` — protecting commands and queries
-4. `references/frontend.md` — consuming identity in the UI
-5. `references/local-development.md` — testing without real infrastructure
-
----
-
-## Quick-Start: Full-Stack Identity Setup
-
-This is the minimum checklist for an application with identity. Read the reference files for details on each step.
-
-### Backend
-
-1. **Details record**: Define a C# record for application-specific user information
-2. **Identity provider**: Implement `IProvideIdentityDetails<TDetails>` (auto-discovered, one per app)
-3. **Startup**: Call `app.MapIdentityProvider()` to register `GET /.cratis/me`
-4. **Authentication**: Add `AddMicrosoftIdentityPlatformIdentityAuthentication()` or implement `IAuthenticationHandler`
-5. **Authorization**: Add `[Authorize]` / `[Roles]` / `[AllowAnonymous]` attributes from `Cratis.Arc.Authorization` to commands and queries
-
-### Frontend
-
-6. **Provider**: Wrap app root with `<IdentityProvider>` from `@cratis/arc.react/identity`
-7. **Hook**: Use `useIdentity<TDetails>()` to access identity anywhere in the component tree
-8. **Roles**: Use `identity.isInRole('Admin')` for UI-level role gating
-
-### Proxy Generation
-
-9. Using `IProvideIdentityDetails<TDetails>` (generic) enables automatic TypeScript type generation at `dotnet build` time — the generated types can be imported in the frontend for end-to-end type safety.
-
----
-
-## Critical Rules
-
-These rules are frequently violated — always enforce them:
-
-1. **One identity provider per app**: Only one `IProvideIdentityDetails` implementation is allowed. Multiple throws `MultipleIdentityDetailsProvidersFound`.
-2. **Use Cratis attributes, not ASP.NET Core's**: `[Authorize]`, `[Roles]`, `[AllowAnonymous]` must come from `Cratis.Arc.Authorization`, not `Microsoft.AspNetCore.Authorization`.
-3. **Never combine `[Authorize]` and `[AllowAnonymous]` on the same target**: This throws `AmbiguousAuthorizationLevel`.
-4. **Prefer the generic interface**: Use `IProvideIdentityDetails<TDetails>` over `IProvideIdentityDetails` to enable proxy generation.
-5. **Auto-discovery**: Both `IProvideIdentityDetails` and `IAuthenticationHandler` implementations are auto-discovered — no DI registration needed.
-6. **Frontend role checks are UX, not security**: `isInRole()` on the frontend hides UI elements. The backend `[Roles]` attribute is the actual security boundary.
-7. **Build before frontend**: TypeScript proxy types for identity details are generated by `dotnet build`. The backend must compile before the frontend can import them.
-
----
-
-## Common Code Patterns
-
-### Protecting a command with roles
-
-```csharp
-[Command]
-[Roles("Admin")]
-public record PromoteUser(UserId Id)
-{
-    public void Handle(IUserService users) => users.Promote(Id);
-}
-```
-
-### Conditionally rendering UI based on roles
-
-```tsx
-const identity = useIdentity();
-
-return identity.isInRole('Admin')
-    ? <AdminPanel />
-    : <AccessDenied />;
-```
-
-### Modifying identity at runtime (stateless selections)
-
-```csharp
-public class SetDepartment(IIdentityProviderResultHandler identityHandler)
-{
-    public async Task Handle(string department) =>
-        await identityHandler.ModifyDetails<UserDetails>(
-            details => details with { SelectedDepartment = department });
-}
-```
-
----
-
-## Reference Documentation
-
-### Skill references (detailed implementation guidance)
-
-- [Backend Identity Provider](references/backend-identity.md) — `IProvideIdentityDetails`, `IdentityProviderContext`, cookie mechanics, proxy generation, `ModifyDetails`
-- [Authentication](references/authentication.md) — `IAuthenticationHandler`, `AuthenticationResult`, Microsoft Identity Platform, combining handlers
-- [Authorization](references/authorization.md) — `[Authorize]`, `[Roles]`, `[AllowAnonymous]`, inheritance rules, fallback policies
-- [Frontend Identity](references/frontend.md) — React `IdentityProvider`, `useIdentity()`, MVVM, core identity, role checking
-- [Local Development](references/local-development.md) — Generating principals, ModHeader, cookie fallback, dev testing
-
-### Project documentation
-
-- [Backend Identity](Documentation/backend/identity.md)
-- [Microsoft Identity](Documentation/backend/asp-net-core/microsoft-identity.md)
-- [ASP.NET Core Authorization](Documentation/backend/asp-net-core/authorization.md)
-- [Core Authentication](Documentation/backend/core/authentication.md)
-- [Core Authorization](Documentation/backend/core/authorization.md)
-- [Command Authorization](Documentation/backend/commands/model-bound/authorization.md)
-- [Query Authorization](Documentation/backend/queries/model-bound/authorization.md)
-- [Proxy Generation for Identity](Documentation/backend/proxy-generation/identity-details.md)
-- [Frontend Core Identity](Documentation/frontend/core/identity.md)
-- [Frontend React Identity](Documentation/frontend/react/identity.md)
-- [Frontend MVVM Identity](Documentation/frontend/react.mvvm/identity.md)
-- [Generating Principals for Local Dev](Documentation/general/generating-principal.md)
diff --git a/.github/skills/auth-and-identity/references/authentication.md b/.github/skills/auth-and-identity/references/authentication.md
deleted file mode 100644
index aa9a4ef..0000000
--- a/.github/skills/auth-and-identity/references/authentication.md
+++ /dev/null
@@ -1,129 +0,0 @@
-# Authentication
-
-This reference covers implementing custom authentication handlers and integrating Microsoft Identity Platform.
-
-## Authentication Pipeline
-
-Arc's authentication system is built around `IAuthenticationHandler` in `Cratis.Arc.Authentication`. Handlers are called in sequence until one succeeds or fails:
-
-1. Each registered handler is called in order
-2. If a handler returns `Succeeded` → request is authenticated, pipeline stops
-3. If a handler returns `Failed` → request is rejected (401), pipeline stops
-4. If a handler returns `Anonymous` → try the next handler
-5. If all handlers return `Anonymous` → request is unauthenticated
-
-## `AuthenticationResult` Outcomes
-
-| Method | Meaning | When to Use |
-|--------|---------|-------------|
-| `AuthenticationResult.Anonymous` | Handler doesn't apply | No relevant credentials found — let other handlers try |
-| `AuthenticationResult.Succeeded(principal)` | Authentication succeeded | Valid credentials → return a `ClaimsPrincipal` |
-| `AuthenticationResult.Failed(reason)` | Authentication explicitly failed | Credentials present but invalid → return 401 |
-
-## Custom Authentication Handler
-
-Handlers are **auto-discovered** — implement `IAuthenticationHandler` and Arc registers it. No manual DI wiring needed.
-
-```csharp
-using System.Security.Claims;
-using Cratis.Arc.Authentication;
-using Cratis.Arc.Http;
-
-public class ApiKeyAuthenticationHandler(IApiKeyValidator validator) : IAuthenticationHandler
-{
-    public async Task<AuthenticationResult> HandleAuthentication(IHttpRequestContext context)
-    {
-        if (!context.Headers.TryGetValue("X-API-Key", out var apiKey))
-            return AuthenticationResult.Anonymous;
-
-        if (!await validator.IsValid(apiKey))
-            return AuthenticationResult.Failed(new AuthenticationFailureReason("Invalid API key"));
-
-        var claims = new[]
-        {
-            new Claim(ClaimTypes.Name, "API Client"),
-            new Claim(ClaimTypes.AuthenticationMethod, "ApiKey")
-        };
-
-        var principal = new ClaimsPrincipal(new ClaimsIdentity(claims, "ApiKey"));
-        return AuthenticationResult.Succeeded(principal);
-    }
-}
-```
-
-### Common Patterns
-
-**Bearer Token:**
-```csharp
-public class BearerTokenAuthenticationHandler : IAuthenticationHandler
-{
-    public async Task<AuthenticationResult> HandleAuthentication(IHttpRequestContext context)
-    {
-        if (!context.Headers.TryGetValue("Authorization", out var header))
-            return AuthenticationResult.Anonymous;
-
-        if (!header.StartsWith("Bearer ", StringComparison.OrdinalIgnoreCase))
-            return AuthenticationResult.Anonymous;
-
-        var token = header["Bearer ".Length..].Trim();
-        // validate token, build ClaimsPrincipal...
-        return AuthenticationResult.Succeeded(principal);
-    }
-}
-```
-
-**Custom Header:**
-```csharp
-public class CustomHeaderAuthenticationHandler : IAuthenticationHandler
-{
-    public Task<AuthenticationResult> HandleAuthentication(IHttpRequestContext context)
-    {
-        if (!context.Headers.TryGetValue("X-User-ID", out var userId))
-            return Task.FromResult(AuthenticationResult.Anonymous);
-
-        var claims = new[] { new Claim(ClaimTypes.NameIdentifier, userId) };
-        var principal = new ClaimsPrincipal(new ClaimsIdentity(claims, "CustomHeader"));
-        return Task.FromResult(AuthenticationResult.Succeeded(principal));
-    }
-}
-```
-
-### Best Practices
-
-- **Return `Anonymous` when your handler doesn't apply** — never `Failed` just because your header is missing
-- **Provide clear failure reasons** — `AuthenticationFailureReason("API key is expired")`
-- **Dependency injection works** — handlers can take constructor dependencies
-- **Handle exceptions gracefully** — catch validation errors and return `Failed` with a reason
-
-## Microsoft Identity Platform
-
-For Azure-hosted apps using Azure AD / Entra ID, Arc provides built-in support:
-
-```csharp
-builder.Services.AddMicrosoftIdentityPlatformIdentityAuthentication();
-```
-
-This registers an ASP.NET Core `AuthenticationHandler` that reads Azure-provided headers:
-
-| Header | Description |
-|--------|-------------|
-| `x-ms-client-principal` | Base64-encoded Microsoft Client Principal token |
-| `x-ms-client-principal-id` | User's unique ID from Azure AD |
-| `x-ms-client-principal-name` | User's display name |
-
-These headers are automatically set by Azure Container Apps, Web Apps, and Static Web Apps. You also need:
-
-```csharp
-var app = builder.Build();
-app.UseAuthentication();
-app.UseAuthorization();
-app.MapIdentityProvider();
-```
-
-## Combining Multiple Handlers
-
-Multiple handlers coexist naturally because `Anonymous` means "I don't handle this request":
-
-1. **ApiKeyAuthenticationHandler** — if `X-API-Key` present, authenticates or rejects. If absent, returns `Anonymous`.
-2. **MicrosoftIdentityPlatformHandler** — checks for `x-ms-client-principal`. If absent, returns `Anonymous`.
-3. If all return `Anonymous` → request is unauthenticated.
diff --git a/.github/skills/auth-and-identity/references/authorization.md b/.github/skills/auth-and-identity/references/authorization.md
deleted file mode 100644
index 18a030a..0000000
--- a/.github/skills/auth-and-identity/references/authorization.md
+++ /dev/null
@@ -1,159 +0,0 @@
-# Authorization
-
-This reference covers protecting commands and queries with authorization attributes.
-
-## Attributes
-
-Arc uses its own authorization attributes from `Cratis.Arc.Authorization` — these are **distinct from ASP.NET Core's** and are evaluated by `AuthorizationEvaluator` in the command/query pipeline.
-
-| Attribute | Purpose |
-|-----------|---------|
-| `[Authorize]` | Require authentication. Optionally specify `Roles` or `Policy`. |
-| `[Roles("Admin", "Manager")]` | Convenience wrapper — user needs at least **one** of the listed roles. |
-| `[AllowAnonymous]` | Bypass authorization. Useful with fallback policies. |
-
-## On Model-Bound Commands
-
-```csharp
-[Command]
-[Roles("Admin", "Editor")]
-public record DeleteArticle(ArticleId Id)
-{
-    public void Handle(IArticleService articles) => articles.Delete(Id);
-}
-```
-
-When authorization fails, `Handle()` is **never called**. Check `CommandResult.IsAuthorized`:
-
-```csharp
-var result = await commandPipeline.Execute(new DeleteArticle(articleId));
-if (!result.IsAuthorized)
-{
-    // User lacked required role — command was not executed
-}
-```
-
-## On Model-Bound Queries
-
-Authorization applies at both class and method level:
-
-```csharp
-[ReadModel]
-[Authorize]
-public record DebitAccount(AccountId Id, AccountName Name, decimal Balance)
-{
-    [Roles("Admin")]
-    public static IEnumerable<DebitAccount> GetAllAccounts(
-        IMongoCollection<DebitAccount> collection) =>
-        collection.Find(_ => true).ToList();
-
-    [Roles("Manager")]
-    public static IEnumerable<DebitAccount> GetHighValueAccounts(
-        IMongoCollection<DebitAccount> collection) =>
-        collection.Find(a => a.Balance > 50000).ToList();
-
-    [AllowAnonymous]
-    public static int GetTotalCount(IMongoCollection<DebitAccount> collection) =>
-        (int)collection.CountDocuments(_ => true);
-}
-```
-
-## Inheritance Rules
-
-| Scenario | Result |
-|----------|--------|
-| `[Authorize]` on type | All methods require authentication |
-| `[Roles]` on type | All methods require those roles |
-| `[AllowAnonymous]` on type | All methods allow anonymous access |
-| Method-level attribute present | **Overrides** type-level attribute |
-| Both `[Authorize]` and `[AllowAnonymous]` on same target | **Error** — throws `AmbiguousAuthorizationLevel` |
-
-Method-level always takes precedence:
-- Methods **without** authorization attributes inherit the class-level attribute
-- Methods **with** `[Roles(...)]` override the class-level attribute
-- Methods **with** `[AllowAnonymous]` completely bypass authorization
-
-## Fallback Policy (Secure by Default)
-
-Make all endpoints require authentication unless explicitly opted out:
-
-```csharp
-builder.Services.AddAuthorizationBuilder()
-    .SetFallbackPolicy(new AuthorizationPolicyBuilder()
-        .RequireAuthenticatedUser()
-        .Build());
-```
-
-With this, use `[AllowAnonymous]` to make specific endpoints public:
-
-```csharp
-[Command]
-[AllowAnonymous]
-public record GetPublicCatalog()
-{
-    public Catalog Handle(ICatalogService catalog) => catalog.GetPublic();
-}
-```
-
-### Default Policy vs Fallback Policy
-
-| Policy | Applied When |
-|--------|-------------|
-| **Default Policy** | `[Authorize]` is used without parameters |
-| **Fallback Policy** | No authorization attribute is present at all |
-
-## Policy-Based Authorization
-
-For complex scenarios:
-
-```csharp
-[Command]
-[Authorize(Policy = "RequireElevatedAccess")]
-public record PerformSensitiveOperation(string Data)
-{
-    public void Handle(ISensitiveService service) => service.Execute(Data);
-}
-```
-
-Define the policy in startup:
-
-```csharp
-builder.Services.AddAuthorization(options =>
-{
-    options.AddPolicy("RequireElevatedAccess", policy =>
-        policy.RequireAssertion(context =>
-            context.User.IsInRole("Admin") ||
-            context.User.HasClaim("elevated", "true")));
-});
-```
-
-## Custom Authorization Logic in Handlers
-
-For domain-specific authorization beyond attributes:
-
-```csharp
-[Authorize]
-public record UpdateOrder(OrderId Id, string Data)
-{
-    public async Task<CommandResult> Handle(
-        IOrderRepository orders, CommandContext context)
-    {
-        var userId = context.User.FindFirst(ClaimTypes.NameIdentifier)?.Value;
-        var order = await orders.GetById(Id);
-
-        if (order.OwnerId != userId)
-            return CommandResult.Forbidden(context.CorrelationId, "Can only update own orders");
-
-        order.Update(Data);
-        await orders.Save(order);
-        return CommandResult.Success;
-    }
-}
-```
-
-## Authorization Results
-
-| Scenario | HTTP Status |
-|----------|-------------|
-| Not authenticated | 401 Unauthorized |
-| Authenticated but wrong role | 403 Forbidden |
diff --git a/.github/skills/auth-and-identity/references/backend-identity.md b/.github/skills/auth-and-identity/references/backend-identity.md
deleted file mode 100644
index f5ba162..0000000
--- a/.github/skills/auth-and-identity/references/backend-identity.md
+++ /dev/null
@@ -1,379 +0,0 @@
-# Backend Identity Provider
-
-This reference covers implementing `IProvideIdentityDetails` to enrich user identity with application-specific details, and what you can use it for.
-
-## What It Is
-
-`IProvideIdentityDetails` is the single extension point where you transform raw authentication claims into rich, application-specific user information. The object you return becomes the `details` field in the frontend identity — available via `useIdentity<TDetails>().details` in React.
-
-Think of it as the bridge between "who the identity provider says this person is" (claims) and "what our application knows about this person" (department, permissions, preferences, tenant, avatar URL, etc.).
-
-## How It Works
-
-When a user hits `GET /.cratis/me`, Arc:
-
-1. Extracts claims from the authenticated `ClaimsPrincipal` (`sub` → `Id`, `Identity.Name` → `Name`, all claims → `Claims`)
-2. Builds an `IdentityProviderContext` with user ID, name, and claims
-3. Calls your `IProvideIdentityDetails.Provide()` method
-4. Serializes the result as JSON and sets the `.cratis-identity` cookie (base64-encoded, `HttpOnly=false` so frontend JS can read it)
-5. Returns the result as JSON in the HTTP response
-
-The `Details` object you return can be **any shape** — it gets serialized as JSON. The frontend deserializes it into the `TDetails` type parameter you specify.
-
-## Mapping the Endpoint
-
-In your application startup:
-
-```csharp
-var app = builder.Build();
-app.UseAuthentication();
-app.UseAuthorization();
-app.MapIdentityProvider();
-```
-
-This registers `GET /.cratis/me` — the well-known route the frontend calls to get identity information.
-
-## The Interface
-
-```csharp
-// Base interface — single method
-public interface IProvideIdentityDetails
-{
-    Task<IdentityDetails> Provide(IdentityProviderContext context);
-}
-
-// Generic variant — adds no methods, just captures TDetails for proxy generation
-public interface IProvideIdentityDetails<TDetails> : IProvideIdentityDetails
-    where TDetails : class;
-```
-
-Always prefer `IProvideIdentityDetails<TDetails>` — the generic type parameter enables automatic TypeScript proxy generation for your details type.
-
-## Basic Implementation
-
-The simplest implementation extracts claims and returns a details record:
-
-```csharp
-public record UserDetails(string Department, string Title, bool IsManager);
-
-public class IdentityDetailsProvider : IProvideIdentityDetails<UserDetails>
-{
-    public Task<IdentityDetails> Provide(IdentityProviderContext context)
-    {
-        var department = context.Claims
-            .FirstOrDefault(c => c.Key == "department").Value ?? "Unknown";
-
-        var details = new UserDetails(department, "Engineer", false);
-        return Task.FromResult(new IdentityDetails(true, details));
-    }
-}
-```
-
-### Rules
-
-- Only **one** `IProvideIdentityDetails` implementation is allowed per app. Multiple implementations throw `MultipleIdentityDetailsProvidersFound`.
-- If none exists, `DefaultIdentityDetailsProvider` is used (grants access to everyone with empty details).
-- Dependency injection works — your provider can take constructor dependencies.
-- The provider is registered as **scoped** (per-request), so it can safely use scoped services.
-- Use `IProvideIdentityDetails<TDetails>` (generic) to enable automatic TypeScript proxy generation for the details type. Always prefer the generic interface.
-
-## `IdentityProviderContext`
-
-This is the input to your `Provide()` method — everything Arc knows about the user from authentication:
-
-```csharp
-public record IdentityProviderContext(
-    IdentityId Id,
-    IdentityName Name,
-    IEnumerable<KeyValuePair<string, string>> Claims);
-```
-
-| Property | Type | Description |
-|----------|------|-------------|
-| `Id` | `IdentityId` | User's unique ID (extracted from the `sub` claim) |
-| `Name` | `IdentityName` | Display name (from `ClaimsPrincipal.Identity.Name`) |
-| `Claims` | `IEnumerable<KeyValuePair<string, string>>` | All claims from the authentication token as key-value pairs |
-
-`IdentityId` and `IdentityName` are `ConceptAs<string>` types with `.Empty` sentinels.
-
-### Working with Claims
-
-Claims contain everything the authentication handler extracted from the token. Common claim keys:
-
-| Claim Key | Description | Example Value |
-|-----------|-------------|---------------|
-| `http://schemas.xmlsoap.org/ws/2005/05/identity/claims/nameidentifier` | User ID | `abc123` |
-| `name` | Display name | `Jane Doe` |
-| `email` or `preferred_username` | Email | `jane@contoso.com` |
-| `http://schemas.microsoft.com/ws/2008/06/identity/claims/role` | Role | `Admin` |
-| Custom claims | Anything your IdP adds | `department`, `tenant_id`, etc. |
-
-```csharp
-// Extract a specific claim
-var email = context.Claims.FirstOrDefault(c => c.Key == "preferred_username").Value;
-
-// Extract all roles
-var roles = context.Claims
-    .Where(c => c.Key == "http://schemas.microsoft.com/ws/2008/06/identity/claims/role")
-    .Select(c => c.Value);
-
-// Check if a claim exists
-var hasDepartment = context.Claims.Any(c => c.Key == "department");
-```
-
-## `IdentityDetails`
-
-The return type from your `Provide()` method:
-
-```csharp
-public record IdentityDetails(bool IsUserAuthorized, object Details);
-```
-
-| Property | Type | Description |
-|----------|------|-------------|
-| `IsUserAuthorized` | `bool` | Whether the user is allowed into the application. `false` → HTTP 403. |
-| `Details` | `object` | Application-specific details payload (serialized as JSON) |
-
-## What You Can Use It For
-
-### 1. Enriching Identity from a Database
-
-The most common use case — look up the user in your own database and add application-specific information:
-
-```csharp
-public record UserDetails(
-    UserId UserId,
-    string Department,
-    string Title,
-    string AvatarUrl,
-    bool IsManager);
-
-public class IdentityDetailsProvider(IMongoCollection<User> users) : IProvideIdentityDetails<UserDetails>
-{
-    public async Task<IdentityDetails> Provide(IdentityProviderContext context)
-    {
-        var user = await users.Find(u => u.ExternalId == context.Id.Value).FirstOrDefaultAsync();
-
-        if (user is null)
-        {
-            // Unknown user — still allow access, but with empty details
-            return new IdentityDetails(true, new UserDetails(
-                UserId.Empty, "Unknown", "Unknown", string.Empty, false));
-        }
-
-        var details = new UserDetails(
-            user.Id,
-            user.Department,
-            user.Title,
-            user.AvatarUrl,
-            user.IsManager);
-
-        return new IdentityDetails(true, details);
-    }
-}
-```
-
-### 2. Blocking Unauthorized Users (Application-Level Gate)
-
-Return `IsUserAuthorized = false` to completely block a user from the application. The endpoint returns HTTP 403 and the frontend never gets identity. This is different from role-based authorization — it's an application-level gate:
-
-```csharp
-public class IdentityDetailsProvider(IMongoCollection<AllowedUser> allowedUsers) : IProvideIdentityDetails<UserDetails>
-{
-    public async Task<IdentityDetails> Provide(IdentityProviderContext context)
-    {
-        var isAllowed = await allowedUsers
-            .Find(u => u.ExternalId == context.Id.Value)
-            .AnyAsync();
-
-        if (!isAllowed)
-        {
-            // User is authenticated but not authorized for this application
-            return new IdentityDetails(false, new UserDetails());
-        }
-
-        // ... build details normally
-        return new IdentityDetails(true, details);
-    }
-}
-```
-
-Use cases:
-- Invite-only applications (only registered users can access)
-- Disabled/suspended user accounts
-- Tenant-specific access control
-
-### 3. Multi-Tenant Identity
-
-Add tenant information to identity so the frontend knows which tenant context the user is in:
-
-```csharp
-public record TenantUserDetails(
-    TenantId TenantId,
-    TenantName TenantName,
-    string Role,
-    IEnumerable<TenantId> AvailableTenants);
-
-public class IdentityDetailsProvider(ITenantService tenants) : IProvideIdentityDetails<TenantUserDetails>
-{
-    public async Task<IdentityDetails> Provide(IdentityProviderContext context)
-    {
-        var userTenants = await tenants.GetTenantsForUser(context.Id.Value);
-        var activeTenant = userTenants.FirstOrDefault();
-
-        if (activeTenant is null)
-        {
-            return new IdentityDetails(false, new TenantUserDetails(
-                TenantId.Empty, TenantName.Empty, string.Empty, []));
-        }
-
-        var details = new TenantUserDetails(
-            activeTenant.Id,
-            activeTenant.Name,
-            activeTenant.UserRole,
-            userTenants.Select(t => t.Id));
-
-        return new IdentityDetails(true, details);
-    }
-}
-```
-
-### 4. Enriching From External APIs
-
-Call external services during identity resolution:
-
-```csharp
-public class IdentityDetailsProvider(
-    IMongoCollection<User> users,
-    IGraphApiClient graphClient) : IProvideIdentityDetails<UserDetails>
-{
-    public async Task<IdentityDetails> Provide(IdentityProviderContext context)
-    {
-        // Combine data from your database and Microsoft Graph
-        var user = await users.Find(u => u.ExternalId == context.Id.Value).FirstOrDefaultAsync();
-        var graphProfile = await graphClient.GetUserProfile(context.Id.Value);
-
-        var details = new UserDetails(
-            user?.Department ?? graphProfile.Department,
-            graphProfile.JobTitle,
-            graphProfile.Photo);
-
-        return new IdentityDetails(true, details);
-    }
-}
-```
-
-### 5. Claims-Only Provider (No Database)
-
-Sometimes all you need is to reshape token claims into a application-friendly structure — no database lookup required:
-
-```csharp
-public record UserDetails(string Email, string Department, IEnumerable<string> Groups);
-
-public class IdentityDetailsProvider : IProvideIdentityDetails<UserDetails>
-{
-    public Task<IdentityDetails> Provide(IdentityProviderContext context)
-    {
-        var email = context.Claims.FirstOrDefault(c => c.Key == "preferred_username").Value ?? string.Empty;
-        var department = context.Claims.FirstOrDefault(c => c.Key == "department").Value ?? "Unknown";
-        var groups = context.Claims
-            .Where(c => c.Key == "groups")
-            .Select(c => c.Value);
-
-        return Task.FromResult(new IdentityDetails(
-            true,
-            new UserDetails(email, department, groups)));
-    }
-}
-```
-
-### 6. User Preferences and Settings
-
-Surface user preferences (theme, language, selected workspace) as identity details so the frontend can use them immediately on load:
-
-```csharp
-public record UserDetails(
-    string Theme,
-    string Language,
-    WorkspaceId ActiveWorkspace);
-
-public class IdentityDetailsProvider(IMongoCollection<UserPreferences> preferences) : IProvideIdentityDetails<UserDetails>
-{
-    public async Task<IdentityDetails> Provide(IdentityProviderContext context)
-    {
-        var prefs = await preferences
-            .Find(p => p.UserId == context.Id.Value)
-            .FirstOrDefaultAsync();
-
-        var details = new UserDetails(
-            prefs?.Theme ?? "light",
-            prefs?.Language ?? "en",
-            prefs?.ActiveWorkspace ?? WorkspaceId.Empty);
-
-        return new IdentityDetails(true, details);
-    }
-}
-```
-
-## Modifying Identity at Runtime
-
-`IIdentityProviderResultHandler` allows modifying identity details during requests — useful for stateless applications tracking user selections without a database round-trip:
-
-```csharp
-public class SetActiveWorkspace(IIdentityProviderResultHandler identityHandler)
-{
-    public async Task Handle(WorkspaceId workspaceId) =>
-        await identityHandler.ModifyDetails<UserDetails>(
-            details => details with { ActiveWorkspace = workspaceId });
-}
-```
-
-This re-generates the identity from the current context, applies the modifier function to the details, and re-writes the cookie. The frontend will see the updated details on the next `identity.refresh()` call or page load.
-
-| Method | Description |
-|--------|-------------|
-| `GenerateFromCurrentContext()` | Build `IdentityProviderResult` from current HTTP context |
-| `Write(result)` | Write result as JSON response + update cookie |
-| `ModifyDetails<T>(modifier)` | Read current cookie, apply modifier, write updated cookie |
-
-### When to Use `ModifyDetails` vs. Database Updates
-
-| Approach | Use When |
-|----------|----------|
-| `ModifyDetails<T>()` | Ephemeral selections (active tenant, selected workspace, UI preferences) that don't need persistence. Updates the cookie in-place. |
-| Database update + `identity.refresh()` | Persistent changes (profile updates, role changes). Update the database, then have the frontend call `refresh()` to pick up the new data. |
-
-## Proxy Generation for Identity Types
-
-When using the generic `IProvideIdentityDetails<TDetails>`, the proxy generator automatically creates TypeScript types for `TDetails` at build time:
-
-1. Define a C# record for your details type
-2. Implement `IProvideIdentityDetails<YourDetails>`
-3. Run `dotnet build` — TypeScript interface is generated
-4. Import and use the generated type in your frontend
-
-| Feature | `IProvideIdentityDetails` | `IProvideIdentityDetails<TDetails>` |
-|---------|---------------------------|-------------------------------------|
-| Runtime behavior | Identical | Identical |
-| Proxy generation | No type generated | TypeScript type generated |
-| Frontend type safety | Manual typing required | Automatic type safety |
-
-## Default Behavior (No Provider)
-
-If you don't implement `IProvideIdentityDetails`, Arc uses `DefaultIdentityDetailsProvider`:
-
-```csharp
-public class DefaultIdentityDetailsProvider : IProvideIdentityDetails
-{
-    public Task<IdentityDetails> Provide(IdentityProviderContext context) =>
-        Task.FromResult(new IdentityDetails(true, new { }));
-}
-```
-
-This grants access to everyone with empty details. It's useful for apps that only need authentication (roles) without custom details.
-
-## Multi-Service Considerations
-
-- **Single service**: Implement `IProvideIdentityDetails` directly
-- **Multiple services**: Have your ingress/reverse proxy call each service's `/.cratis/me` and merge results into a single JSON structure for the `.cratis-identity` cookie
-- **Dedicated identity service**: Aggregate identity info from various sources in one service
diff --git a/.github/skills/auth-and-identity/references/frontend.md b/.github/skills/auth-and-identity/references/frontend.md
deleted file mode 100644
index 733eb71..0000000
--- a/.github/skills/auth-and-identity/references/frontend.md
+++ /dev/null
@@ -1,184 +0,0 @@
-# Frontend Identity
-
-This reference covers consuming identity in React, MVVM, and vanilla TypeScript.
-
-## How the Frontend Gets Identity
-
-The frontend uses a **cookie-first** approach:
-
-1. Check for the `.cratis-identity` cookie (base64-encoded JSON, `HttpOnly=false`)
-2. If cookie exists → decode and use it (no HTTP call needed)
-3. If no cookie → call `GET /.cratis/me` to fetch identity and set the cookie
-4. In development mode, the cookie fallback (`/.cratis/me` call) always works — no ingress simulation needed
-
-## React
-
-### IdentityProvider Context
-
-Wrap your app root with `<IdentityProvider>`:
-
-```tsx
-import { IdentityProvider } from '@cratis/arc.react/identity';
-
-export const App = () => (
-    <IdentityProvider>
-        {/* your app */}
-    </IdentityProvider>
-);
-```
-
-For type-safe details with complex types (e.g., `Guid`), pass a `detailsType` constructor:
-
-```tsx
-import { IdentityProvider } from '@cratis/arc.react/identity';
-import { Guid } from '@cratis/fundamentals';
-
-class UserIdentityDetails {
-    userId: Guid = Guid.empty;
-    firstName: string = '';
-    lastName: string = '';
-}
-
-export const App = () => (
-    <IdentityProvider detailsType={UserIdentityDetails}>
-        {/* your app */}
-    </IdentityProvider>
-);
-```
-
-### `useIdentity()` Hook
-
-Access identity anywhere in your component tree:
-
-```tsx
-import { useIdentity } from '@cratis/arc.react/identity';
-
-type UserDetails = {
-    department: string;
-    title: string;
-};
-
-export const UserProfile = () => {
-    const identity = useIdentity<UserDetails>();
-
-    return (
-        <div>
-            <h3>{identity.name}</h3>
-            <p>Department: {identity.details.department}</p>
-        </div>
-    );
-};
-```
-
-**With default values** (useful for local development when the cookie might not exist):
-
-```tsx
-const identity = useIdentity<UserDetails>({
-    department: '[N/A]',
-    title: '[N/A]'
-});
-```
-
-**With a constructor for type-safe deserialization** (uses `JsonSerializer.deserializeFromInstance()` under the hood):
-
-```tsx
-const identity = useIdentity(UserIdentityDetails);
-
-// With default values:
-const identity = useIdentity(UserIdentityDetails, {
-    userId: Guid.empty,
-    firstName: '[N/A]',
-    lastName: '[N/A]'
-});
-```
-
-### Role Checking
-
-```tsx
-const identity = useIdentity();
-
-if (identity.isInRole('Admin')) {
-    // show admin UI
-}
-
-// Or access roles directly
-console.log(identity.roles);
-```
-
-### Refreshing Identity
-
-When identity changes on the backend (e.g., user granted new roles):
-
-```tsx
-const identity = useIdentity();
-const handleRefresh = () => identity.refresh();
-```
-
-This calls `GET /.cratis/me` again and updates both the cookie and context.
-
-### `IIdentity<TDetails>` Shape
-
-| Property | Type | Description |
-|----------|------|-------------|
-| `id` | `string` | Unique ID from identity provider |
-| `name` | `string` | Display name |
-| `roles` | `string[]` | Assigned roles |
-| `details` | `TDetails` | Application-specific details |
-| `isSet` | `boolean` | Whether identity has been loaded |
-| `isInRole(role)` | `(string) => boolean` | Check role membership |
-| `refresh()` | `() => Promise<IIdentity>` | Re-fetch identity from backend |
-
-## MVVM (tsyringe)
-
-In an MVVM setup, `IIdentityProvider` is automatically registered in the DI container by `Bindings.initialize()`:
-
-```typescript
-import { injectable } from 'tsyringe';
-import { IIdentityProvider } from '@cratis/arc/identity';
-
-type UserDetails = {
-    department: string;
-};
-
-@injectable()
-export class MyViewModel {
-    constructor(private readonly _identityProvider: IIdentityProvider) {}
-
-    async loadUser() {
-        const identity = await this._identityProvider.getCurrent<UserDetails>();
-        console.log(identity.details.department);
-    }
-}
-```
-
-Requires the [MVVM Context](Documentation/frontend/react.mvvm/mvvm-context.md) to be set up.
-
-## Vanilla TypeScript (No Framework)
-
-Use `IdentityProvider` directly:
-
-```typescript
-import { IdentityProvider } from '@cratis/arc/identity';
-
-const identity = await IdentityProvider.getCurrent<UserDetails>();
-console.log(identity.name);
-console.log(identity.isInRole('Admin'));
-```
-
-## Frontend Role-Based UI Pattern
-
-Combine `useIdentity()` with authorization attributes on the backend for defense in depth:
-
-```tsx
-const identity = useIdentity();
-
-return (
-    <div>
-        {identity.isInRole('Admin') && <AdminPanel />}
-        {identity.isInRole('Manager') && <ManagerDashboard />}
-        <PublicContent />
-    </div>
-);
-```
-
-The frontend check is for UX (hiding buttons the user can't use). The backend `[Roles]` attribute is the actual security boundary.
diff --git a/.github/skills/auth-and-identity/references/local-development.md b/.github/skills/auth-and-identity/references/local-development.md
deleted file mode 100644
index 6fed70a..0000000
--- a/.github/skills/auth-and-identity/references/local-development.md
+++ /dev/null
@@ -1,111 +0,0 @@
-# Local Development & Testing
-
-This reference covers simulating authentication and identity in local development environments without real Azure or identity infrastructure.
-
-## How Identity Works in Development
-
-When running locally:
-
-1. No Azure App Service or ingress injects identity headers
-2. The `.cratis-identity` cookie will not be set automatically
-3. The frontend cookie reader falls back to calling `GET /.cratis/me`
-4. The backend returns a default anonymous identity with empty details
-
-## Generating a Microsoft Client Principal
-
-Azure App Service Easy Auth injects identity via the `X-MS-CLIENT-PRINCIPAL` header. You can simulate this locally with a browser extension.
-
-### Step 1: Build the Principal JSON
-
-```json
-{
-    "auth_typ": "aad",
-    "claims": [
-        { "typ": "http://schemas.xmlsoap.org/ws/2005/05/identity/claims/nameidentifier", "val": "user-unique-id" },
-        { "typ": "name", "val": "Jane Developer" },
-        { "typ": "http://schemas.microsoft.com/ws/2008/06/identity/claims/role", "val": "Admin" },
-        { "typ": "http://schemas.microsoft.com/ws/2008/06/identity/claims/role", "val": "Manager" }
-    ],
-    "name_typ": "name",
-    "role_typ": "http://schemas.microsoft.com/ws/2008/06/identity/claims/role"
-}
-```
-
-### Step 2: Base64-Encode It
-
-**macOS / Linux terminal:**
-
-```bash
-echo -n '{"auth_typ":"aad","claims":[{"typ":"http://schemas.xmlsoap.org/ws/2005/05/identity/claims/nameidentifier","val":"user-unique-id"},{"typ":"name","val":"Jane Developer"},{"typ":"http://schemas.microsoft.com/ws/2008/06/identity/claims/role","val":"Admin"}],"name_typ":"name","role_typ":"http://schemas.microsoft.com/ws/2008/06/identity/claims/role"}' | base64
-```
-
-**Browser console:**
-
-```javascript
-btoa(JSON.stringify({
-    auth_typ: "aad",
-    claims: [
-        { typ: "http://schemas.xmlsoap.org/ws/2005/05/identity/claims/nameidentifier", val: "user-unique-id" },
-        { typ: "name", val: "Jane Developer" },
-        { typ: "http://schemas.microsoft.com/ws/2008/06/identity/claims/role", val: "Admin" }
-    ],
-    name_typ: "name",
-    role_typ: "http://schemas.microsoft.com/ws/2008/06/identity/claims/role"
-}));
-```
-
-### Step 3: Inject the Header
-
-Use the [ModHeader](https://modheader.com/) browser extension:
-
-1. Install ModHeader for Chrome/Edge/Firefox
-2. Add a request header:
-   - **Name**: `X-MS-CLIENT-PRINCIPAL`
-   - **Value**: the base64 string from Step 2
-3. Reload the page — the backend will authenticate the user as if Azure App Service injected the header
-
-## Custom Identity Details in Development
-
-If your `IProvideIdentityDetails<TDetails>` implementation enriches identity from a database, that still runs in development. The principal provides the initial `id`, `name`, and `roles`; your enrichment adds extra fields.
-
-To test without a database, create a dev-only fallback in your identity provider:
-
-```csharp
-public Task<UserDetails> Provide(IdentityProviderContext context)
-{
-    // In development, if the user ID is unknown, return sensible defaults
-    var details = await _readModelStore.GetOrDefault<UserDetails>(context.Id, new UserDetails
-    {
-        Department = "Engineering",
-        Title = "Developer"
-    });
-
-    return details;
-}
-```
-
-## Testing Without ModHeader
-
-If you prefer not to install a browser extension, you can also set the cookie directly:
-
-1. Build the identity JSON matching the `.cratis-identity` cookie format
-2. Set it via browser console:
-
-```javascript
-document.cookie = '.cratis-identity=' + btoa(JSON.stringify({
-    id: 'user-id',
-    name: 'Jane Developer',
-    roles: ['Admin'],
-    details: { department: 'Engineering' }
-})) + '; path=/';
-```
-
-3. Reload — the frontend will use this cookie directly without calling the backend
-
-## Important Notes
-
-- The cookie is `HttpOnly=false` by design — the frontend JavaScript must read it
-- The cookie path is `/`
-- In production, the cookie is set by the backend middleware; in development, you can set it manually
-- The `GET /.cratis/me` endpoint always works — it is the fallback for all environments
-- ModHeader header injection simulates the full pipeline (authentication → identity provider → cookie), while direct cookie setting bypasses the backend entirely
diff --git a/.github/skills/cratis-command/SKILL.md b/.github/skills/cratis-command/SKILL.md
deleted file mode 100644
index 5a4c024..0000000
--- a/.github/skills/cratis-command/SKILL.md
+++ /dev/null
@@ -1,306 +0,0 @@
----
-name: cratis-command
-description: Step-by-step guidance for creating a Cratis Arc command — [Command] record, Handle() method, CommandValidator, proxy generation, and React .use() hook with CommandDialog. Use when adding or creating a command, wiring up a form or button to the backend, working with IEventLog, CommandResult, CommandValidator, CommandDialog, or [Command] attribute.
----
-
-# Creating a Cratis Command
-
-A command represents a user action that changes state. In Cratis Arc the path is:
-
-```
-[Command] record + Handle()  →  validator  →  dotnet build  →  TypeScript proxy  →  React .use()
-```
-
-The command record **owns its own handler** — no separate controller class required. Follow the steps in order. Jump to the reference files for deeper detail on any step.
-
----
-
-## Step 1 — Define the C# command record
-
-A command is a **record decorated with `[Command]`** that contains its own `Handle()` method. No separate controller is needed.
-
-```csharp
-// API/Accounts/OpenDebitAccount.cs
-namespace MyApp.API.Accounts;
-
-using Cratis.Arc.Commands.ModelBound;
-using Cratis.Chronicle.Events;
-
-[Command]
-public record OpenDebitAccount(AccountId AccountId, string Name, OwnerId OwnerId)
-{
-    public DebitAccountOpened Handle() =>
-        new(Name, OwnerId); // Arc appends the returned event; AccountId is the event source
-}
-
-[EventType]  // NO arguments — never [EventType("some-guid")]
-public record DebitAccountOpened(string Name, OwnerId OwnerId);
-```
-
-**Rules:**
-- `[Command]` attribute is **required** — it makes the type discoverable and the analyzer will warn without it
-- `Handle()` returns the event (or events) to append — Arc's Chronicle integration automatically appends them
-- `[EventType]` takes **no arguments** — the identifier is generated from the type name
-- Name the command as an imperative action — `OpenDebitAccount`, not `OpenDebitAccountCommand`
-- Use `ConceptAs<T>` wrappers for all identity types — never raw `Guid` or `string`
-
-```csharp
-// Concepts/AccountId.cs
-using Cratis.Concepts;
-
-public record AccountId(Guid Value) : ConceptAs<Guid>(Value);
-```
-
-### Generating a new ID and returning it
-
-```csharp
-[Command]
-public record RegisterEmployee(string FirstName, string LastName, string Department)
-{
-    // Return (eventSourceId, event) — Arc uses the first element as the event source ID
-    // and sends it back to the client as CommandResult<Guid>.response
-    public (EmployeeId, EmployeeRegistered) Handle()
-    {
-        var employeeId = new EmployeeId(Guid.NewGuid());
-        return (employeeId, new(FirstName, LastName, Department));
-    }
-}
-```
-
-### Appending multiple events
-
-```csharp
-[Command]
-public record TransferFunds(AccountId FromId, AccountId ToId, decimal Amount)
-{
-    public IEnumerable<object> Handle() =>
-    [
-        new FundsWithdrawn(FromId, Amount),
-        new FundsDeposited(ToId, Amount),
-    ];
-}
-```
-
----
-
-## Step 2 — Inject dependencies into Handle (when needed)
-
-Handle() parameters are injected by DI — use this for constraints or external services:
-
-```csharp
-[Command]
-public record OpenDebitAccount(AccountId AccountId, string Name, OwnerId OwnerId)
-{
-    public async Task<DebitAccountOpened> Handle(
-        IUniqueAccountConstraint constraint)
-    {
-        await constraint.Validate(Name);
-        return new(Name, OwnerId);
-    }
-}
-```
-
-Arc automatically wraps the return in `CommandResult` / `CommandResult<T>`. See `references/command-result.md`.
-
----
-
-## Step 3 — Add validation (optional but recommended)
-
-FluentValidation rules are extracted by the proxy generator and run **client-side** before the request is sent:
-
-```csharp
-// API/Accounts/OpenDebitAccountValidator.cs
-public class OpenDebitAccountValidator : CommandValidator<OpenDebitAccount>
-{
-    public OpenDebitAccountValidator()
-    {
-        RuleFor(c => c.Name)
-            .NotEmpty().WithMessage("Account name is required")
-            .MaximumLength(100);
-        RuleFor(c => c.OwnerId)
-            .NotEmpty().WithMessage("Owner is required");
-    }
-}
-```
-
-- Extends `CommandValidator<T>` (not `AbstractValidator<T>`) — this makes it discoverable automatically
-- No registration needed
-- Arc also creates a `/validate` endpoint automatically; the frontend `command.validate()` calls it without executing the handler
-
-For simple cases, Data Annotations on the record work too:
-
-```csharp
-public record OpenDebitAccount(
-    [Required] Guid AccountId,
-    [Required, MaxLength(100)] string Name,
-    [Required] Guid OwnerId);
-```
-
----
-
-## Step 4 — Generate the TypeScript proxy
-
-```bash
-dotnet build
-```
-
-The `Cratis.Arc.ProxyGenerator.Build` MSBuild package runs during the build and writes TypeScript files to the path configured in your `.csproj`:
-
-```xml
-<PropertyGroup>
-  <CratisProxiesOutputPath>$(MSBuildThisFileDirectory)../Web/src/api</CratisProxiesOutputPath>
-</PropertyGroup>
-```
-
-This produces `Web/src/api/Accounts/OpenDebitAccount.ts`. For first-time setup see `references/proxy-setup.md`.
-
----
-
-## Step 5 — Use the command in React
-
-### Inline form (full control)
-
-```tsx
-import { OpenDebitAccount } from '../api/Accounts/OpenDebitAccount';
-
-export const OpenAccountForm = () => {
-    const [command, setValues] = OpenDebitAccount.use();
-    const [error, setError] = useState('');
-
-    const handleSubmit = async () => {
-        const result = await command.execute();
-        if (result.isSuccess) {
-            onSuccess(result.response); // result.response is Guid if you returned one
-        } else if (!result.isValid) {
-            setError(result.validationResults[0]?.message ?? 'Validation failed');
-        }
-    };
-
-    return (
-        <form onSubmit={handleSubmit}>
-            <input
-                value={command.name}
-                onChange={e => (command.name = e.target.value)}
-                placeholder="Account name"
-            />
-            {error && <p className="error">{error}</p>}
-            <button disabled={!command.hasChanges}>Open account</button>
-        </form>
-    );
-};
-```
-
-**Key properties on the command instance:**
-
-| Property / method | What it does |
-| ----------------- | ------------ |
-| `command.propName` | Get/set the property value |
-| `command.hasChanges` | `true` when any value differs from the initial |
-| `command.execute()` | Send the POST, returns `CommandResult` |
-| `command.validate()` | Call the validate endpoint (no side effects) |
-| `setValues(obj)` | Set multiple properties at once (e.g. from a query result) |
-
-### With initial values (edit scenario)
-
-```tsx
-const [command] = UpdateAccount.use({
-    accountId: account.id,
-    name: account.name,
-});
-```
-
-### Using `CommandDialog` (quickest path for modal forms)
-
-See Step 6 and `references/command-dialog.md`.
-
----
-
-## Step 6 — Wrap in a CommandDialog (optional)
-
-For modal dialogs, create a dialog component using `DialogProps` and wire it up with `useDialog`:
-
-```tsx
-import { DialogProps } from '@cratis/arc.react/dialogs';
-import { CommandDialog } from '@cratis/components/CommandDialog';
-import { InputTextField } from '@cratis/components/CommandForm';
-import { OpenDebitAccount } from '../api/Accounts/OpenDebitAccount';
-
-// --- Dialog component ---
-export const OpenAccountDialog = ({ closeDialog }: DialogProps) => {
-    return (
-        <CommandDialog<OpenDebitAccount>
-            command={OpenDebitAccount}
-            title="Open account"
-            okLabel="Open"
-        >
-            <InputTextField<OpenDebitAccount> value={c => c.name} label="Name" />
-        </CommandDialog>
-    );
-};
-
-// --- Parent component ---
-import { useDialog, DialogResult } from '@cratis/arc.react/dialogs';
-import { OpenAccountDialog } from './OpenAccountDialog';
-
-export const AccountsPage = () => {
-    const [OpenAccountDialogWrapper, showOpenAccount] = useDialog(OpenAccountDialog);
-
-    const handleOpen = async () => {
-        const [result] = await showOpenAccount();
-        if (result === DialogResult.Ok) {
-            // command already executed inside the dialog — refresh your data here
-        }
-    };
-
-    return (
-        <>
-            <button onClick={handleOpen}>Open account</button>
-            <OpenAccountDialogWrapper />
-        </>
-    );
-};
-```
-
-**How it works:**
-- `useDialog(OpenAccountDialog)` returns a wrapper component and a `show` function
-- `showOpenAccount()` opens the dialog; it returns a Promise that resolves when the dialog closes
-- `CommandDialog` executes the command when the user confirms; it closes automatically
-- `DialogProps` provides `closeDialog` — needed when you want to pass a response back or handle cancel explicitly
-
-**Edit dialog (pre-populate with existing values):**
-
-```tsx
-interface EditAccountDialogProps extends DialogProps {
-    accountId: string;
-    name: string;
-}
-
-export const EditAccountDialog = ({ closeDialog, accountId, name }: EditAccountDialogProps) => {
-    return (
-        <CommandDialog<UpdateAccount>
-            command={UpdateAccount}
-            title="Edit account"
-            initialValues={{ accountId }}
-            currentValues={{ name }}
-        >
-            <InputTextField<UpdateAccount> value={c => c.name} label="Name" />
-        </CommandDialog>
-    );
-};
-```
-
-- `initialValues` — sets the change-tracking baseline (e.g. IDs that must be present for the command but are not user-entered)
-- `currentValues` — pre-populates the visible field values
-
-`CommandDialog` calls `onConfirm` only after a successful `command.execute()`, so you don't need to check `isSuccess` yourself. See `references/command-dialog.md` for the full props list.
-
----
-
-## Reference files
-
-| File | What's in it |
-| ---- | ------------ |
-| `references/command-result.md` | Full `CommandResult` shape, error handling patterns |
-| `references/command-dialog.md` | `CommandDialog` props, CommandForm fields, edit dialogs |
-| `references/validation.md` | FluentValidation, Data Annotations, client-side pre-flight |
-| `references/proxy-setup.md` | First-time proxy generator setup |
diff --git a/.github/skills/cratis-command/evals/evals.json b/.github/skills/cratis-command/evals/evals.json
deleted file mode 100644
index a0d7e29..0000000
--- a/.github/skills/cratis-command/evals/evals.json
+++ /dev/null
@@ -1,35 +0,0 @@
-{
-  "skill_name": "cratis-command",
-  "evals": [
-    {
-      "id": 1,
-      "prompt": "Create a command to register a new employee with a first name, last name, and department. Duplicate employee names (same first+last) should be prevented. Include everything needed: the event, constraint, validator, and command with Handle().",
-      "expected_output": "A single C# file containing: [EventType] EmployeeRegistered record, UniqueEmployeeName IConstraint, RegisterEmployeeValidator CommandValidator<T>, and [Command] RegisterEmployee record with Handle() that returns (EmployeeId, EmployeeRegistered). Correct namespacing and file header.",
-      "files": [],
-      "assertions": [
-        "Output contains [Command] attribute on a record",
-        "Output contains a Handle() method directly on the command record (not a separate handler class)",
-        "[EventType] attribute has no arguments (no GUID, no string)",
-        "Output contains an IConstraint implementation for uniqueness",
-        "Output contains a CommandValidator<T> using FluentValidation",
-        "Output uses ConceptAs<T> for EmployeeId (not raw Guid)",
-        "File uses file-scoped namespace declaration",
-        "File starts with the Cratis copyright header"
-      ]
-    },
-    {
-      "id": 2,
-      "prompt": "I need a command to change the email address of an existing customer. The command should validate the email format. Show me the full backend slice file.",
-      "expected_output": "A single C# slice file containing: [EventType] CustomerEmailChanged record, a validator checking email format, and [Command] ChangeCustomerEmail record with Handle() returning the event. Uses ConceptAs<T> for ids.",
-      "files": [],
-      "assertions": [
-        "Output contains [Command] attribute on a record",
-        "Handle() is defined directly on the command record",
-        "Output contains CommandValidator<T> with email format validation",
-        "[EventType] attribute has no arguments",
-        "Output uses ConceptAs<T> for identity types (not raw Guid or string)",
-        "File uses file-scoped namespace"
-      ]
-    }
-  ]
-}
diff --git a/.github/skills/cratis-command/references/command-dialog.md b/.github/skills/cratis-command/references/command-dialog.md
deleted file mode 100644
index 7a18367..0000000
--- a/.github/skills/cratis-command/references/command-dialog.md
+++ /dev/null
@@ -1,138 +0,0 @@
-# CommandDialog — Reference
-
-`CommandDialog` from `@cratis/components/CommandDialog` is a modal dialog that executes a command when the user confirms.
-
-## How it works
-
-- It creates a command instance internally from the `command` constructor you pass
-- It renders your `CommandForm` fields as children, bound to that instance
-- When the user clicks OK, it calls `command.execute()`
-- `onConfirm` is called **only if** execution succeeds
-- `onCancel` / dismiss closes without executing
-
----
-
-## Dialog pattern with `useDialog` and `DialogProps`
-
-```tsx
-import { DialogProps, DialogResult, useDialog } from '@cratis/arc.react/dialogs';
-import { CommandDialog } from '@cratis/components/CommandDialog';
-import { InputTextField, NumberField } from '@cratis/components/CommandForm';
-import { CreateProduct } from '../api/Products/CreateProduct';
-
-// 1. Define the dialog component
-export const CreateProductDialog = ({ closeDialog }: DialogProps) => {
-    return (
-        <CommandDialog<CreateProduct>
-            command={CreateProduct}
-            title="Create product"
-            okLabel="Create"
-        >
-            <InputTextField<CreateProduct> value={c => c.name} label="Name" />
-            <NumberField<CreateProduct> value={c => c.price} label="Price" />
-        </CommandDialog>
-    );
-};
-
-// 2. Use it from the parent
-const [CreateProductDialogWrapper, showCreateProduct] = useDialog(CreateProductDialog);
-
-const handleCreate = async () => {
-    const [result] = await showCreateProduct();
-    if (result === DialogResult.Ok) {
-        refreshProducts();
-    }
-};
-
-// 3. Render the wrapper
-return (
-    <>
-        <button onClick={handleCreate}>Create product</button>
-        <CreateProductDialogWrapper />
-    </>
-);
-```
-
-`DialogProps` provides `closeDialog` as a prop to the dialog component. `CommandDialog` automatically executes the command on confirm and closes the dialog.
-
----
-
-## Edit dialog (pre-populate with existing values)
-
-```tsx
-interface EditProductDialogProps extends DialogProps {
-    product: Product;
-}
-
-export const EditProductDialog = ({ closeDialog, product }: EditProductDialogProps) => {
-    return (
-        <CommandDialog<UpdateProduct>
-            command={UpdateProduct}
-            title="Edit product"
-            currentValues={{ name: product.name, price: product.price }}
-            initialValues={{ productId: product.id }}
-        >
-            <InputTextField<UpdateProduct> value={c => c.name} label="Name" />
-            <NumberField<UpdateProduct> value={c => c.price} label="Price" />
-        </CommandDialog>
-    );
-};
-
-// Pass product to the dialog
-const [EditDialogWrapper, showEdit] = useDialog(EditProductDialog);
-await showEdit({ product: selectedProduct });
-```
-
-- `initialValues` sets the baseline for change tracking (values that are not "changes")
-- `currentValues` populates the initial field display
-
----
-
-## CommandForm field components
-
-All fields come from `@cratis/components/CommandForm`. Always pass the command type as the generic parameter so the `value` accessor is fully typed.
-
-```tsx
-import {
-    InputTextField,    // text input
-    NumberField,       // number input
-    CheckboxField,     // boolean toggle
-    DateField,         // date picker
-    DropdownField,     // select from options list
-    TextAreaField,     // multi-line text
-} from '@cratis/components/CommandForm';
-
-<InputTextField<MyCommand> value={c => c.title} label="Title" />
-<NumberField<MyCommand> value={c => c.quantity} label="Qty" min={1} />
-<CheckboxField<MyCommand> value={c => c.isActive} label="Active" />
-<DateField<MyCommand> value={c => c.dueDate} label="Due date" />
-<DropdownField<MyCommand>
-    value={c => c.status}
-    label="Status"
-    options={statusOptions}
-    optionLabel="label"
-    optionValue="value"
-/>
-<TextAreaField<MyCommand> value={c => c.notes} label="Notes" rows={3} />
-```
-
-The `value` prop takes a function `(commandInstance) => property`. This drives both reading the value and writing it back on change.
-
----
-
-## CommandDialog props
-
-| Prop | Required | Purpose |
-| ---- | -------- | ------- |
-| `command` | ✓ | Command constructor |
-| `title` | ✓ | Dialog header text |
-| `initialValues` | | Values set as the change-tracking baseline |
-| `currentValues` | | Values to pre-populate the fields |
-| `onConfirm` | | Called after successful execution |
-| `onCancel` | | Called when cancelled/dismissed |
-| `okLabel` | | Confirm button text (default: "Ok") |
-| `cancelLabel` | | Cancel button text (default: "Cancel") |
-| `isValid` | | Extra validity gate combining with field validation |
-| `onBeforeExecute` | | Transform command values just before `.execute()` |
-| `onFieldChange` | | Callback on every field change |
-| `buttons` | | Override button set (`DialogButtons` enum or custom node) |
diff --git a/.github/skills/cratis-command/references/command-result.md b/.github/skills/cratis-command/references/command-result.md
deleted file mode 100644
index 394fae8..0000000
--- a/.github/skills/cratis-command/references/command-result.md
+++ /dev/null
@@ -1,81 +0,0 @@
-# CommandResult — Reference
-
-Arc wraps every command response in a `CommandResult` envelope.
-
-## Shape
-
-```ts
-interface CommandResult<T = void> {
-    isSuccess: boolean;          // true when authorized + valid + no exceptions
-    isAuthorized: boolean;       // false → user lacks permission
-    isValid: boolean;            // false → one or more validators failed
-    validationResults: ValidationResult[];
-    hasExceptions: boolean;      // true → unhandled server exception
-    exceptionMessages: string[];
-    exceptionStackTrace: string;
-    response: T | null;          // present when the backend returned a value
-}
-
-interface ValidationResult {
-    propertyName: string;   // camelCase, matches the command property
-    message: string;
-    severity: string;       // 'Error' | 'Warning' | 'Info'
-}
-```
-
-## Handling all cases
-
-```tsx
-const handleSubmit = async () => {
-    const result = await command.execute();
-
-    if (!result.isAuthorized) {
-        navigate('/login');
-        return;
-    }
-
-    if (!result.isValid) {
-        // Map errors by property for inline display
-        const fieldErrors = result.validationResults.reduce((acc, v) => {
-            acc[v.propertyName] = v.message;
-            return acc;
-        }, {} as Record<string, string>);
-        setErrors(fieldErrors);
-        return;
-    }
-
-    if (result.hasExceptions) {
-        toast.error(result.exceptionMessages.join('\n'));
-        return;
-    }
-
-    onSuccess(result.response);  // result.response typed when backend returns a value
-};
-```
-
-## Accessing the returned value
-
-If the backend controller returns a value:
-
-```csharp
-[HttpPost]
-public async Task<Guid> CreateOrder([FromBody] CreateOrder command)
-{
-    var id = Guid.NewGuid();
-    await eventLog.Append(id, new OrderCreated(...));
-    return id;
-}
-```
-
-Then on the frontend:
-
-```ts
-const result = await createOrder.execute();
-if (result.isSuccess) {
-    const newId = result.response as string; // Guids come as strings
-}
-```
-
-## Bypassing the CommandResult wrapper
-
-If you need to return a raw HTTP response (e.g. for file downloads), use `[AspNetResult]` on the controller action. The frontend receives the raw response and the proxy does not include `execute()`.
diff --git a/.github/skills/cratis-command/references/proxy-setup.md b/.github/skills/cratis-command/references/proxy-setup.md
deleted file mode 100644
index bf061b0..0000000
--- a/.github/skills/cratis-command/references/proxy-setup.md
+++ /dev/null
@@ -1,71 +0,0 @@
-# Proxy Generator Setup — Reference
-
-The proxy generator runs as an MSBuild task during `dotnet build` and produces TypeScript interfaces and classes for every command and query it finds.
-
----
-
-## Install the NuGet package
-
-Add to every `.csproj` that contains controllers, commands, or queries:
-
-```xml
-<PackageReference Include="Cratis.Arc.ProxyGenerator.Build" Version="*" />
-```
-
-## Configure the output path
-
-```xml
-<PropertyGroup>
-  <!-- Path is relative to the .csproj file -->
-  <CratisProxiesOutputPath>$(MSBuildThisFileDirectory)../Web/src/api</CratisProxiesOutputPath>
-</PropertyGroup>
-```
-
-## Install the frontend package
-
-```bash
-npm install @cratis/arc
-```
-
----
-
-## Run the generator
-
-```bash
-dotnet build
-```
-
-The generator will write TypeScript files under the configured output path, mirroring your C# namespace hierarchy as folders:
-
-```
-Web/src/api/
-  Accounts/
-    OpenDebitAccount.ts     ← POST action → command proxy
-    AllAccounts.ts          ← GET action  → query proxy
-    index.ts
-  index.ts
-```
-
----
-
-## Common gotchas
-
-| Problem | Fix |
-| ------- | --- |
-| No files generated | Ensure the package is referenced and the project builds cleanly first |
-| Wrong folder structure | The folder mirrors the **namespace**, not the file path — adjust your namespace |
-| Stale proxies | Run `dotnet clean && dotnet build` to force full regeneration |
-| Proxies mixed with hand-written files | Set `<CratisProxiesSkipOutputDeletion>true</CratisProxiesSkipOutputDeletion>` to prevent the generator deleting everything; or move proxies to a dedicated folder |
-
----
-
-## Multi-project solutions
-
-```
-MyApp.API.csproj:
-  <ProjectReference Include="../MyApp.Domain/MyApp.Domain.csproj" />
-  <PackageReference Include="Cratis.Arc.ProxyGenerator.Build" Version="*" />
-  <CratisProxiesOutputPath>../MyApp.Web/src/api</CratisProxiesOutputPath>
-```
-
-Only the project with controllers needs the proxy generator package. Domain and read-model projects do not.
diff --git a/.github/skills/cratis-command/references/validation.md b/.github/skills/cratis-command/references/validation.md
deleted file mode 100644
index 2d3f092..0000000
--- a/.github/skills/cratis-command/references/validation.md
+++ /dev/null
@@ -1,104 +0,0 @@
-# Command Validation — Reference
-
-Validation in Cratis Arc runs in two places: **client-side** (via the proxy, before the request is sent) and **server-side** (the full pipeline, always).
-
----
-
-## FluentValidation (recommended)
-
-```csharp
-// Must extend CommandValidator<T>, not AbstractValidator<T>
-public class CreateOrderValidator : CommandValidator<CreateOrder>
-{
-    public CreateOrderValidator()
-    {
-        RuleFor(c => c.CustomerId).NotEmpty().WithMessage("Customer is required");
-        RuleFor(c => c.Total).GreaterThan(0).WithMessage("Total must be positive");
-        RuleFor(c => c.Items).NotEmpty().WithMessage("Order must have at least one item");
-    }
-}
-```
-
-**Why `CommandValidator<T>`?** It marks the class for automatic discovery (no DI registration needed) and allows the proxy generator to extract the rules into TypeScript for client-side pre-flight.
-
-Rules that can be extracted and run client-side:
-- `NotEmpty`, `NotNull`
-- `MaximumLength`, `MinimumLength`, `Length`
-- `GreaterThan`, `LessThan`, `GreaterThanOrEqualTo`, `LessThanOrEqualTo`
-- `Must` with simple predicates
-- `EmailAddress`
-
-Rules that only run server-side (cannot be extracted):
-- Validators with injected dependencies (e.g. database uniqueness checks)
-
----
-
-## Data Annotations
-
-```csharp
-public record CreateOrder(
-    [Required] Guid CustomerId,
-    [Range(0.01, double.MaxValue, ErrorMessage = "Total must be positive")] decimal Total,
-    [Required, MinLength(1)] List<OrderItem> Items
-);
-```
-
-Simpler but less flexible than FluentValidation. Rules are enforced server-side and reflected in `CommandResult.validationResults`.
-
----
-
-## Automatic validate endpoint
-
-For every `[HttpPost]` command, Arc registers a parallel endpoint:
-
-- Execute: `POST /api/orders/create`
-- Validate: `POST /api/orders/create/validate`
-
-The validate endpoint runs all authorization and validation filters but **never** calls the handler. No side effects.
-
----
-
-## Client-side validate() in React
-
-```tsx
-const [command] = CreateOrder.use();
-const [errors, setErrors] = useState<Record<string, string>>({});
-
-// Option A: validate on blur
-const handleBlur = async (field: string) => {
-    const result = await command.validate();
-    const fieldError = result.validationResults.find(v => v.propertyName === field);
-    setErrors(prev => ({ ...prev, [field]: fieldError?.message ?? '' }));
-};
-
-// Option B: validate proactively as user types
-useEffect(() => {
-    command.validate().then(result => {
-        setCanSubmit(result.isSuccess);
-    });
-}, [command.hasChanges]);
-
-// Option C: validate + conditional execute
-const handleSubmit = async () => {
-    const validation = await command.validate();
-    if (!validation.isValid) {
-        setErrors(validation.validationResults.reduce(...));
-        return;
-    }
-    await command.execute();
-};
-```
-
----
-
-## Showing validation errors per field
-
-```tsx
-const getError = (field: keyof typeof command) =>
-    result.validationResults.find(v => v.propertyName === String(field))?.message;
-
-<input value={command.name} onChange={...} />
-{getError('name') && <span className="error">{getError('name')}</span>}
-```
-
-`propertyName` in the result is camelCase matching the C# property name (lowercased first letter).
diff --git a/.github/skills/cratis-csharp-standards/SKILL.md b/.github/skills/cratis-csharp-standards/SKILL.md
deleted file mode 100644
index d62b401..0000000
--- a/.github/skills/cratis-csharp-standards/SKILL.md
+++ /dev/null
@@ -1,79 +0,0 @@
----
-name: cratis-csharp-standards
-description: Reference for Cratis C# coding conventions — formatting, naming, records, nullable handling, exceptions, logging, and DI. Use whenever writing C# in a Cratis project, deciding between record vs class, checking naming rules, applying formatting conventions, handling null safety, creating exception types, or asking "how should this be written?" Also covers CUPID characteristics and domain-based folder structure. Trigger on any C# style or standards question for a Cratis project.
----
-
-## Key rules (quick reference)
-
-- Use C# 13 features always — records, primary constructors, pattern matching
-- `var` over explicit types — the right side already tells you the type
-- File-scoped namespace declarations — one less indentation level
-- `using` directives: alphabetically sorted, single-line, unused ones removed
-- No regions — if a file needs them, it needs refactoring
-- No postfixes: no `Async`, `Impl`, `Service`, `Manager`, `Handler` on class names
-- No `Exception` suffix on exception types — `AuthorNotFound` not `AuthorNotFoundException`
-- Never use built-in exceptions — always create custom exception types
-- `record` for events, commands, read models, concepts — value equality for free
-- Primary constructors for all types — eliminates field boilerplate
-- `is null` / `is not null` — never `== null` / `!= null`
-- Blank line before opening `{` of every code block
-- Final `return` on its own line
-- Private fields: `_camelCase` with underscore prefix
-- Interfaces: `I` prefix (`IMyService`)
-- No `[EventType]` arguments — the type name is the event identifier
-- `[Command]` records define `Handle()` directly — no separate handler classes
-
----
-
-## Formatting
-
-```csharp
-// File-scoped namespace (no extra indent)
-namespace MyApp.Authors.Registration;
-
-// Alphabetically sorted using directives
-using Cratis.Arc.Commands;
-using Cratis.Chronicle.Events;
-using Microsoft.Extensions.Logging;
-
-// Blank line before { of every block
-if (condition)
-{
-    DoSomething();
-}
-
-// Expression-bodied for simple members
-public string FullName => $"{FirstName} {LastName}";
-
-// Final return on its own line
-public string GetName()
-{
-    var result = BuildName();
-
-    return result;
-}
-```
-
----
-
-## Naming
-
-| Artifact | Convention | Example |
-| --- | --- | --- |
-| Type / method / public member | PascalCase | `RegisterAuthor`, `AuthorId` |
-| Private field | `_camelCase` | `_eventLog` |
-| Local variable | camelCase | `authorId` |
-| Interface | `I` prefix | `IEventLog` |
-| Exception type | No `Exception` suffix | `AuthorNotFound` |
-| Feature folder | Pluralized domain noun | `Authors/` |
-| Concept file | Concept name | `AuthorId.cs` |
-
-No abbreviations unless widely known (XML, JSON, Id, URL). No prefixes/postfixes that describe technical role (Controller, ViewModel, Handler, Manager, Factory, Base).
-
----
-
-## Reference files
-
-- `references/code-style.md` — records, primary constructors, nullable, collections, async
-- `references/exceptions-logging-di.md` — exception types, logging with [LoggerMessage], DI conventions
-- `references/domain-philosophy.md` — CUPID characteristics, cohesion, ubiquitous language
diff --git a/.github/skills/cratis-csharp-standards/evals/evals.json b/.github/skills/cratis-csharp-standards/evals/evals.json
deleted file mode 100644
index 1a6c38c..0000000
--- a/.github/skills/cratis-csharp-standards/evals/evals.json
+++ /dev/null
@@ -1,37 +0,0 @@
-{
-  "skill_name": "cratis-csharp-standards",
-  "evals": [
-    {
-      "id": 1,
-      "prompt": "Write a C# service class called AccountManager that handles creating bank accounts. It takes IEventLog and ILogger as dependencies. When CreateAccount(AccountName name) is called, it logs 'Creating account {Name}' at Information level and appends an AccountCreated event. If the account already exists it should throw an appropriate exception. Follow all Cratis C# coding standards.",
-      "expected_output": "Class uses primary constructor, file-scoped namespace, record for event, custom exception (no Exception suffix, not built-in), [LoggerMessage] in a separate AccountManagerLogging.cs partial class, ConceptAs<T> for AccountId/AccountName, var for local variables, expression-bodied members where appropriate.",
-      "files": [],
-      "assertions": [
-        "Uses primary constructor (not field declarations + constructor body)",
-        "Uses file-scoped namespace (no extra indentation level)",
-        "Custom exception class defined (not InvalidOperationException or similar built-in)",
-        "Custom exception class does NOT have 'Exception' suffix",
-        "Logging uses [LoggerMessage] attribute in a separate partial static internal class",
-        "AccountId uses ConceptAs<Guid> (not raw Guid)",
-        "AccountName uses ConceptAs<string> (not raw string)",
-        "Uses 'var' for local variable declarations",
-        "No postfix on class names (no 'Manager', 'Impl', 'Service' naming anti-patterns caught)"
-      ]
-    },
-    {
-      "id": 2,
-      "prompt": "Review this C# code and rewrite it following Cratis coding standards:\n\n```csharp\npublic class OrderService\n{\n    private IOrderRepository _repo;\n    private ILogger _logger;\n    \n    public OrderService(IOrderRepository repo, ILogger logger)\n    {\n        _repo = repo;\n        _logger = logger;\n    }\n    \n    public async Task<List<Order>> GetAllOrdersAsync()\n    {\n        _logger.LogInformation(\"Getting all orders\");\n        var orders = await _repo.GetAllAsync();\n        return orders;\n    }\n    \n    public void ProcessOrder(Guid orderId)\n    {\n        if (orderId == null)\n            throw new InvalidOperationException(\"OrderId cannot be null\");\n    }\n}\n```",
-      "expected_output": "Rewritten with primary constructor, file-scoped namespace, IEnumerable<Order> return type, [LoggerMessage] logging, ConceptAs<Guid> for OrderId, custom exception, 'is null' check, Async suffix removed, no redundant 'Async' postfix.",
-      "files": [],
-      "assertions": [
-        "Primary constructor used (field declarations removed)",
-        "Returns IEnumerable<Order> not List<Order>",
-        "Uses 'is null' instead of '== null' for null check",
-        "Custom exception replaces InvalidOperationException",
-        "Async method does not have Async postfix",
-        "Uses [LoggerMessage] or mentions it should be moved to logging partial class",
-        "OrderId converted to ConceptAs<Guid> or at minimum flagged as a raw primitive"
-      ]
-    }
-  ]
-}
diff --git a/.github/skills/cratis-csharp-standards/references/code-style.md b/.github/skills/cratis-csharp-standards/references/code-style.md
deleted file mode 100644
index 7d98160..0000000
--- a/.github/skills/cratis-csharp-standards/references/code-style.md
+++ /dev/null
@@ -1,187 +0,0 @@
-# Code Style — Reference
-
-## Records
-
-Use `record` types for all immutable data structures — events, commands, read models, concepts, DTOs.
-
-```csharp
-// ✅ Prefer record
-public record AuthorRegistered(AuthorName Name);
-
-// ✅ Record with primary constructor
-public record Author(AuthorId Id, AuthorName Name);
-```
-
-Records give value equality, immutability, and concise syntax for free. A `record class` with `init`-only properties is equivalent when you need extra methods.
-
----
-
-## Primary Constructors
-
-Use primary constructors for all types — they eliminate the field declaration + constructor ceremony.
-
-```csharp
-// ✅ Primary constructor
-public class AuthorService(IEventLog eventLog, ILogger<AuthorService> logger)
-{
-    public async Task Register(AuthorName name) =>
-        await eventLog.Append(AuthorId.New(), new AuthorRegistered(name));
-}
-
-// ❌ Verbose constructor + fields
-public class AuthorService
-{
-    readonly IEventLog _eventLog;
-    readonly ILogger<AuthorService> _logger;
-
-    public AuthorService(IEventLog eventLog, ILogger<AuthorService> logger)
-    {
-        _eventLog = eventLog;
-        _logger = logger;
-    }
-}
-```
-
-When NOT using primary constructors (e.g. you need field initialization logic), prefix private fields with `_`.
-
----
-
-## var
-
-Always use `var` when declaring local variables — the right side already tells you the type.
-
-```csharp
-// ✅
-var authorId = AuthorId.New();
-var authors = collection.Find(_ => true).ToList();
-
-// ❌
-AuthorId authorId = AuthorId.New();
-List<Author> authors = collection.Find(_ => true).ToList();
-```
-
----
-
-## Expression-bodied members
-
-Use for simple methods and properties:
-
-```csharp
-public string FullName => $"{FirstName} {LastName}";
-public void Log(string message) => _logger.LogInformation(message);
-public AuthorId Id => _id;
-```
-
----
-
-## Collections
-
-```csharp
-// ✅ Return IEnumerable for read-only sequences
-public IEnumerable<Author> All() => _collection.Find(_ => true).ToList();
-
-// ❌ Never expose mutable collection types from public APIs
-public List<Author> All() => ...
-public Dictionary<AuthorId, Author> ByName() => ...
-
-// ✅ IReadOnlyDictionary for key-value returns
-public IReadOnlyDictionary<AuthorId, Author> ByName() => ...
-```
-
----
-
-## Nullable Reference Types
-
-```csharp
-// ✅ is null / is not null
-if (author is null) throw new AuthorNotFound();
-if (name is not null) DoSomething(name);
-
-// ❌ == null / != null
-if (author == null) ...
-
-// Trust the type system — don't add defensive null checks when annotations guarantee non-null:
-// If param is not nullable, don't guard it
-public void Register(AuthorName name)    // name is guaranteed non-null — no null check needed
-{
-    // ...
-}
-
-// Add ! when you're certain but the compiler isn't:
-var author = collection.Find(a => a.Id == id).FirstOrDefault()!;
-```
-
----
-
-## Async
-
-```csharp
-// ✅ Async with proper naming (no Async suffix unless needed for overload disambiguation)
-public async Task Append(AuthorId id, AuthorRegistered @event) =>
-    await _eventLog.Append(id, @event);
-
-// Use Task<T> for async results
-public async Task<Author?> FindById(AuthorId id) => ...
-
-// Use await — never .Result or .Wait()
-var result = await _eventLog.Append(id, @event);
-```
-
----
-
-## Immutability
-
-Prefer immutable designs. Use `with` expressions to create modified copies of records:
-
-```csharp
-var updated = existing with { Name = newName };
-```
-
-Avoid returning mutable objects that callers could mutate. The owner of state is responsible for mutations.
-
----
-
-## Pattern matching
-
-Use pattern matching and switch expressions wherever possible:
-
-```csharp
-// ✅ Pattern matching
-if (result is Result<Author, Error>.Success success)
-    return success.Value;
-
-// ✅ Switch expression
-var description = status switch
-{
-    Status.Active => "Active",
-    Status.Inactive => "Inactive",
-    _ => "Unknown"
-};
-```
-
----
-
-## String interpolation
-
-```csharp
-// ✅
-var message = $"Author '{name}' already exists";
-
-// ❌
-var message = string.Format("Author '{0}' already exists", name);
-var message = "Author '" + name + "' already exists";
-```
-
----
-
-## Interface bodies
-
-For interfaces with no members, omit the body:
-
-```csharp
-// ✅
-public interface IMyMarker;
-
-// ❌
-public interface IMyMarker { }
-```
diff --git a/.github/skills/cratis-csharp-standards/references/domain-philosophy.md b/.github/skills/cratis-csharp-standards/references/domain-philosophy.md
deleted file mode 100644
index 46db1e1..0000000
--- a/.github/skills/cratis-csharp-standards/references/domain-philosophy.md
+++ /dev/null
@@ -1,99 +0,0 @@
-# Domain Philosophy — Reference
-
-## CUPID characteristics
-
-Rather than adhering purely to SOLID principles, Cratis favors the **CUPID** characteristics (Dan North):
-
-| Letter | Characteristic | What it means |
-| --- | --- | --- |
-| **C** | Composable | Things play nicely together, minimal coupling, components can be assembled freely |
-| **U** | Unix philosophy | Do one thing and do it well — focused, single-purpose components |
-| **P** | Predictable | Deterministic behavior, consistent output, no surprises |
-| **I** | Idiomatic | Code feels natural for the language and ecosystem |
-| **D** | Domain-based | Use domain vocabulary and structure, not technical vocabulary |
-
----
-
-## Cohesion over layers
-
-**Do not** split code by technical role (MVC-style):
-
-```
-❌ Layered (avoid)
-Models/
-  Author.cs
-Controllers/
-  AuthorsController.cs
-Services/
-  AuthorService.cs
-Events/
-  AuthorRegistered.cs
-```
-
-**Do** group by feature — everything that changes together lives together:
-
-```
-✅ Feature-cohesive (Cratis style)
-Features/
-  Authors/
-    Registration/
-      Registration.cs   ← command + event + validator
-      AddAuthor.tsx
-    Listing/
-      Listing.cs        ← read model + projection + query
-      Listing.tsx
-```
-
-For different technical concerns (frontend vs backend), naturally separate into different projects, but maintain the cohesive feature structure within each project.
-
----
-
-## Domain language (Ubiquitous Language)
-
-Name things after the domain concept they represent, not after the technical pattern:
-
-| ✅ Domain-named | ❌ Tech-named |
-| --- | --- |
-| `Authors` | `AuthorController`, `AuthorManager` |
-| `Registration` | `RegisterAuthorHandler`, `RegisterAuthorCommand` |
-| `AuthorNotFound` | `AuthorNotFoundException`, `NotFoundException` |
-| `AuthorId` | `Guid authorId` |
-| `Listing` | `GetAllAuthorsQuery` |
-
----
-
-## Pluralization
-
-Features are groupings — pluralize them:
-
-- Folder: `Authors/`, `Employees/`, `Orders/`
-- Route: `/api/Authors/{authorId}`, `/api/Orders`
-- Schema: `Authors`, `OrderItems`
-
----
-
-## 12-Factor
-
-Systems should follow [12factor.net](https://12factor.net) for scalability, maintainability, and operability:
-
-- Config from environment, not hardcoded
-- Stateless processes
-- Treat logs as event streams
-- Declarative setup for easy replication
-
----
-
-## Frictionless dependencies
-
-Healthy dependencies = fast, independent releases. If you must coordinate releases between two components, there is an unhealthy coupling that should be addressed through events, interfaces, or package versioning.
-
----
-
-## Immutability & side-effects
-
-Favor immutable designs to reduce side effects:
-
-- Records with `init`-only properties
-- Return new instances instead of mutating existing ones
-- Expose `IEnumerable<T>` and `IReadOnlyDictionary<K,V>` — never mutable collections from public APIs
-- The owner of state is responsible for its mutations — don't let consumers mutate your internal state
diff --git a/.github/skills/cratis-csharp-standards/references/exceptions-logging-di.md b/.github/skills/cratis-csharp-standards/references/exceptions-logging-di.md
deleted file mode 100644
index 53b2f9a..0000000
--- a/.github/skills/cratis-csharp-standards/references/exceptions-logging-di.md
+++ /dev/null
@@ -1,115 +0,0 @@
-# Exceptions, Logging, and Dependency Injection — Reference
-
-## Exceptions
-
-### Rules
-
-- Only throw for truly exceptional situations — never for control flow
-- Always create a **custom** exception type — never use built-in types (`InvalidOperationException`, `ArgumentException`, etc.)
-- **No `Exception` suffix** — `AuthorNotFound` reads better than `AuthorNotFoundException`
-- Provide a meaningful message
-- Add XML `<exception>` doc starting with "The exception that is thrown when ..."
-
-### Pattern
-
-```csharp
-/// <summary>
-/// The exception that is thrown when an author is not found.
-/// </summary>
-/// <param name="id">The <see cref="AuthorId"/> that was not found.</param>
-public class AuthorNotFound(AuthorId id) : Exception($"Author with id '{id}' was not found");
-```
-
-Usage:
-
-```csharp
-var author = await _authors.FindById(id)
-    ?? throw new AuthorNotFound(id);
-```
-
----
-
-## Logging
-
-### Rules
-
-- Structured logging with named parameters
-- `ILogger<T>` where `T` is the containing class
-- Log message definitions go in a separate `<ClassName>Logging.cs` file — partial static internal class
-- Use `[LoggerMessage]` attribute — do **not** include `eventId`
-- Appropriate log levels: `Information`, `Warning`, `Error`, `Debug`
-
-### Logging class pattern
-
-```csharp
-// AuthorServiceLogging.cs
-namespace MyApp.Authors;
-
-static partial class AuthorServiceLogging
-{
-    [LoggerMessage(LogLevel.Information, "Registering author '{Name}'")]
-    internal static partial void RegisteringAuthor(this ILogger<AuthorService> logger, AuthorName name);
-
-    [LoggerMessage(LogLevel.Warning, "Author with name '{Name}' already exists")]
-    internal static partial void AuthorAlreadyExists(this ILogger<AuthorService> logger, AuthorName name);
-}
-```
-
-Usage in the service:
-
-```csharp
-public class AuthorService(ILogger<AuthorService> logger)
-{
-    public Task Register(AuthorName name)
-    {
-        logger.RegisteringAuthor(name);
-        // ...
-    }
-}
-```
-
----
-
-## Dependency Injection
-
-### Rules
-
-- Prefer **constructor injection** — never use `IServiceProvider` directly (service locator anti-pattern)
-- For singletons, use the `[Singleton]` attribute — no explicit `services.AddSingleton<>()` needed
-- Convention-based systems (`IFoo → Foo`) are auto-discovered — don't register them explicitly
-- `Handle()` method parameters are automatically resolved from DI — no manual wiring needed
-
-### [Singleton] attribute
-
-```csharp
-[Singleton]
-public class AuthorService(IEventLog eventLog) : IAuthorService
-{
-    // registered as singleton automatically
-}
-```
-
-### DI in Handle()
-
-```csharp
-[Command]
-public record SendNotification(AuthorId AuthorId)
-{
-    // INotificationService resolved from DI automatically
-    public async Task Handle(INotificationService notifications) =>
-        await notifications.Notify(AuthorId, "Welcome!");
-}
-```
-
-### Avoiding service locator
-
-```csharp
-// ✅ Constructor injection
-public class MyService(IAuthorService authors) { ... }
-
-// ❌ Service locator
-public class MyService(IServiceProvider provider)
-{
-    void DoWork() => provider.GetService<IAuthorService>()!.DoSomething();
-}
-```
diff --git a/.github/skills/cratis-react-page/SKILL.md b/.github/skills/cratis-react-page/SKILL.md
deleted file mode 100644
index cd83df6..0000000
--- a/.github/skills/cratis-react-page/SKILL.md
+++ /dev/null
@@ -1,221 +0,0 @@
----
-name: cratis-react-page
-description: Step-by-step guidance for building a React page in a Cratis Arc application — listing data with DataPage, toolbar actions with CommandDialog, row selection, detail panels, observable queries, and MVVM. Use whenever building or modifying a React page that lists or displays data, adding a table or grid, wiring up Add/Edit/Delete actions, using DataPage, DataTableForQuery, CommandDialog, or any @cratis/components. Also trigger when connecting a React component to a proxy-generated query or observable query.
----
-
-## Workflow
-
-### Step 1 — Prerequisites
-
-- Backend query and command endpoints must already exist (see `cratis-readmodel` and `cratis-command` skills)
-- Run `dotnet build` on the backend to regenerate proxies
-- All proxy `.ts` files in `<CratisProxiesOutputPath>` must be committed/saved before importing
-
-Imports you will need:
-
-```tsx
-import { DataPage, MenuItemGroup, MenuItem } from '@cratis/components';
-import { CommandDialog } from '@cratis/components/CommandDialog';
-import { useDialog, DialogProps } from '@cratis/arc.react/dialogs';
-```
-
----
-
-### Step 2 — Basic DataPage setup
-
-`DataPage` combines a toolbar/menu, a data table, and an optional detail panel into one component.
-
-```tsx
-import { DataPage, Column } from '@cratis/components';
-import { AllAccounts } from './queries/AllAccounts';
-
-export const AccountsPage = () => {
-    return (
-        <DataPage
-            query={AllAccounts}
-            columns={[
-                { header: 'Name', field: 'name' },
-                { header: 'Balance', field: 'balance' },
-            ]}
-        />
-    );
-};
-```
-
----
-
-### Step 3 — Add menu actions
-
-Menu items appear in the toolbar above the table. Create a separate dialog component using `DialogProps`, then wire it up with `useDialog`.
-
-**Dialog component (`CreateAccountDialog.tsx`):**
-
-```tsx
-import { DialogProps } from '@cratis/arc.react/dialogs';
-import { CommandDialog } from '@cratis/components/CommandDialog';
-import { InputTextField } from '@cratis/components/CommandForm';
-import { CreateAccount } from './commands/CreateAccount';
-
-export const CreateAccountDialog = ({ closeDialog }: DialogProps) => {
-    return (
-        <CommandDialog<CreateAccount>
-            command={CreateAccount}
-            title="Create Account"
-            okLabel="Create"
-        >
-            <InputTextField<CreateAccount> value={c => c.name} label="Account Name" />
-        </CommandDialog>
-    );
-};
-```
-
-**Page component:**
-
-```tsx
-import { DataPage, MenuItemGroup, MenuItem } from '@cratis/components';
-import { useDialog } from '@cratis/arc.react/dialogs';
-import { CreateAccountDialog } from './CreateAccountDialog';
-
-export const AccountsPage = () => {
-    const [CreateAccountWrapper, showCreateAccount] = useDialog(CreateAccountDialog);
-
-    return (
-        <>
-            <DataPage
-                query={AllAccounts}
-                columns={[
-                    { header: 'Name', field: 'name' },
-                ]}
-                menuItems={
-                    <MenuItemGroup>
-                        <MenuItem label="Add Account" onClick={() => showCreateAccount()} />
-                    </MenuItemGroup>
-                }
-            />
-            <CreateAccountWrapper />
-        </>
-    );
-};
-```
-
-See `references/dialogs.md` for the full dialog pattern guide.
-
----
-
-### Step 4 — Row selection and edit dialog
-
-Pass a selected-row handler and an edit dialog. Supply the row data as props to the dialog component.
-
-**Edit dialog component (`EditAccountDialog.tsx`):**
-
-```tsx
-import { DialogProps, DialogResult } from '@cratis/arc.react/dialogs';
-import { CommandDialog } from '@cratis/components/CommandDialog';
-import { InputTextField } from '@cratis/components/CommandForm';
-import { EditAccount } from './commands/EditAccount';
-
-interface EditAccountDialogProps extends DialogProps {
-    accountId: string;
-    name: string;
-}
-
-export const EditAccountDialog = ({ closeDialog, accountId, name }: EditAccountDialogProps) => {
-    return (
-        <CommandDialog<EditAccount>
-            command={EditAccount}
-            title="Edit Account"
-            okLabel="Save"
-            initialValues={{ accountId }}
-            currentValues={{ name }}
-        >
-            <InputTextField<EditAccount> value={c => c.name} label="Account Name" />
-        </CommandDialog>
-    );
-};
-```
-
-**Page component:**
-
-```tsx
-const [EditAccountWrapper, showEditAccount] = useDialog(EditAccountDialog);
-
-<DataPage
-    ...
-    onRowSelected={(row) => showEditAccount({ accountId: row.id, name: row.name })}
-/>
-
-<EditAccountWrapper />
-```
-
----
-
-### Step 5 — Choose observable vs standard query
-
-Use an **observable query** when the data updates in real time without user-triggered refresh:
-
-```tsx
-<DataPage
-    observableQuery={AllAccountsLive}  // use observableQuery instead of query
-    ...
-/>
-```
-
-Observable query results push updates automatically. You cannot call `requery()` on observable queries.
-
-Use a **standard query** for data that only changes when the user takes an action (and use `useEffect`/`refresh` to invalidate after a command).
-
----
-
-### Step 6 — Detail panel (optional)
-
-A detail panel renders beside the table when a row is selected.
-
-```tsx
-<DataPage
-    query={AllAccounts}
-    detailPanel={(row) => <AccountDetail account={row} />}
-    ...
-/>
-```
-
----
-
-### Step 7 — MVVM view model (optional, for complex pages)
-
-For pages with complex state or coordination logic, wrap the page in a view model:
-
-```tsx
-import { withViewModel } from '@cratis/arc.react.mvvm';
-import { injectable } from 'tsyringe';
-
-@injectable()
-class AccountsViewModel {
-    selectedAccount?: AccountSummary;
-}
-
-export const AccountsPage = withViewModel(AccountsViewModel, ({ viewModel }) => (
-    <DataPage ... />
-));
-```
-
-See `references/mvvm.md` for full MVVM guidance.
-
----
-
-## Quick decision guide
-
-| Need | Use |
-| --- | --- |
-| Read-only list | `DataPage` with `query` |
-| Real-time updates | `DataPage` with `observableQuery` |
-| Add / create action | `MenuItem` + `CommandDialog` + `useDialog` |
-| Edit selected row | Row selection + `CommandDialog` + `currentValues`/`initialValues` |
-| Side detail panel | `detailPanel` prop on `DataPage` |
-| Complex page logic | `withViewModel` MVVM wrapper |
-
-## Reference files
-
-- `references/data-page.md` — DataPage props, MenuItems, Columns, detailPanel
-- `references/data-table.md` — Standalone DataTableForQuery / DataTableForObservableQuery
-- `references/dialogs.md` — `useDialog`, `DialogProps`, `CommandDialog` full API
-- `references/mvvm.md` — `withViewModel`, `@injectable`, `IHandleProps`, reactive props
diff --git a/.github/skills/cratis-react-page/evals/evals.json b/.github/skills/cratis-react-page/evals/evals.json
deleted file mode 100644
index f637c9c..0000000
--- a/.github/skills/cratis-react-page/evals/evals.json
+++ /dev/null
@@ -1,34 +0,0 @@
-{
-  "skill_name": "cratis-react-page",
-  "evals": [
-    {
-      "id": 1,
-      "prompt": "Build a React page for managing employees. It should list all employees in a table (name, department, start date columns). There should be an 'Add Employee' button in the toolbar that opens a dialog to create a new employee (fields: firstName, lastName, department). Assume the proxies AllEmployees (query) and RegisterEmployee (command) already exist.",
-      "expected_output": "A TSX file using DataPage with query prop, columns, MenuItemGroup/MenuItem for Add, and CommandDialog with InputTextField fields. useDialog for wiring the dialog. CommandForm fields for all employee properties.",
-      "files": [],
-      "assertions": [
-        "Output uses DataPage component (not a raw table)",
-        "Output uses useDialog for dialog state management",
-        "Output uses CommandDialog (not a manual Dialog/Button setup)",
-        "Output uses MenuItemGroup and MenuItem for the Add action",
-        "Output uses InputTextField or similar CommandForm fields inside the dialog",
-        "Output does NOT manually manage open/close boolean state for the dialog",
-        "Output imports from correct packages (@cratis/components, @cratis/arc.react/dialogs)"
-      ]
-    },
-    {
-      "id": 2,
-      "prompt": "I need a React page that shows a real-time list of active orders (observable query: AllOrdersLive). When clicking an order, show a detail panel on the side. There's a 'Cancel Order' action that should open a confirmation dialog. Use MVVM pattern since we have complex selection state.",
-      "expected_output": "TSX using DataPage with observableQuery, detailPanel prop, withViewModel from @cratis/arc.react.mvvm with @injectable() view model, and CommandDialog for cancel action.",
-      "files": [],
-      "assertions": [
-        "Output uses observableQuery prop (not query) for real-time data",
-        "Output uses withViewModel from @cratis/arc.react.mvvm",
-        "View model class has @injectable() decorator",
-        "View model class calls makeAutoObservable(this)",
-        "Output uses detailPanel prop on DataPage",
-        "Output uses CommandDialog for the cancel action"
-      ]
-    }
-  ]
-}
diff --git a/.github/skills/cratis-react-page/references/data-page.md b/.github/skills/cratis-react-page/references/data-page.md
deleted file mode 100644
index 74b13f3..0000000
--- a/.github/skills/cratis-react-page/references/data-page.md
+++ /dev/null
@@ -1,103 +0,0 @@
-# DataPage — Reference
-
-`DataPage` (from `@cratis/components`) is the standard full-page layout providing a menubar, data table, and optional detail panel in one component.
-
-## Import
-
-```tsx
-import { DataPage, MenuItemGroup, MenuItem, Column } from '@cratis/components';
-```
-
-## Core props
-
-| Prop | Type | Description |
-| --- | --- | --- |
-| `query` | Query class | Proxy-generated query class (standard) |
-| `observableQuery` | Observable query class | Proxy-generated observable query class (real-time) |
-| `columns` | `Column<T>[]` | Column definitions (see below) |
-| `menuItems` | `ReactNode` | Toolbar content (usually `<MenuItemGroup>`) |
-| `detailPanel` | `(row: T) => ReactNode` | Renders to the right when a row is selected |
-| `onRowSelected` | `(row: T) => void` | Callback when user clicks a row |
-| `noDataMessage` | `string` | Message when the query returns no rows |
-| `queryArgs` | `object` | Arguments forwarded to the query proxy |
-
-Use either `query` or `observableQuery` — not both.
-
-## Column definition
-
-```tsx
-type Column<T> = {
-    header: string;
-    field: keyof T | ((row: T) => string);
-    width?: number | string;
-    sortable?: boolean;
-};
-```
-
-Example with custom renderer:
-
-```tsx
-columns={[
-    { header: 'Name', field: 'name' },
-    { header: 'Balance', field: (row) => row.balance.toFixed(2) },
-]}
-```
-
-## MenuItemGroup / MenuItem
-
-```tsx
-<MenuItemGroup>
-    <MenuItem label="Add" onClick={openCreate} />
-    <MenuItem label="Delete" onClick={openDelete} disabled={!selectedRow} />
-</MenuItemGroup>
-```
-
-Multiple `MenuItemGroup` children create visual separators between groups.
-
-## Detail panel
-
-The detail panel receives the currently selected row. It is hidden when no row is selected.
-
-```tsx
-<DataPage
-    query={AllAccounts}
-    columns={...}
-    detailPanel={(account) => (
-        <AccountDetail accountId={account.id} />
-    )}
-/>
-```
-
-## Full example
-
-```tsx
-import { useDialog } from '@cratis/arc.react/dialogs';
-import { CreateAccountDialog } from './CreateAccountDialog';
-
-export const AccountsPage = () => {
-    const [CreateAccountWrapper, showCreateAccount] = useDialog(CreateAccountDialog);
-
-    return (
-        <>
-            <DataPage
-                query={AllAccounts}
-                columns={[
-                    { header: 'Account', field: 'name' },
-                    { header: 'Owner', field: 'ownerName' },
-                    { header: 'Balance', field: (r) => `$${r.balance.toFixed(2)}` },
-                ]}
-                menuItems={
-                    <MenuItemGroup>
-                        <MenuItem label="Create Account" onClick={() => showCreateAccount()} />
-                    </MenuItemGroup>
-                }
-                detailPanel={(row) => <AccountDetail account={row} />}
-                noDataMessage="No accounts found."
-            />
-            <CreateAccountWrapper />
-        </>
-    );
-};
-```
-
-`CreateAccountDialog` is a separate component that receives `closeDialog` via `DialogProps` and renders a `CommandDialog`. See `dialogs.md` for the full dialog pattern.
diff --git a/.github/skills/cratis-react-page/references/data-table.md b/.github/skills/cratis-react-page/references/data-table.md
deleted file mode 100644
index c2cce21..0000000
--- a/.github/skills/cratis-react-page/references/data-table.md
+++ /dev/null
@@ -1,62 +0,0 @@
-# Data Tables — Reference
-
-Use standalone data table components when you need a table without the built-in `DataPage` full-page chrome (e.g., embedded inside another panel or card).
-
-## DataTableForQuery
-
-```tsx
-import { DataTableForQuery } from '@cratis/components';
-import { AllAccounts } from './queries/AllAccounts';
-
-<DataTableForQuery
-    query={AllAccounts}
-    columns={[
-        { header: 'Name', field: 'name' },
-        { header: 'Balance', field: 'balance' },
-    ]}
-    onRowSelected={(row) => setSelected(row)}
-/>
-```
-
-## DataTableForObservableQuery
-
-```tsx
-import { DataTableForObservableQuery } from '@cratis/components';
-import { AllAccountsLive } from './queries/AllAccountsLive';
-
-<DataTableForObservableQuery
-    query={AllAccountsLive}
-    columns={...}
-    onRowSelected={(row) => setSelected(row)}
-/>
-```
-
-## Shared props
-
-| Prop | Type | Description |
-| --- | --- | --- |
-| `query` / `query` | Query class | Proxy query (use the appropriate component for type) |
-| `columns` | `Column<T>[]` | Column definitions (same shape as DataPage) |
-| `onRowSelected` | `(row: T) => void` | Row click callback |
-| `selectedRow` | `T \| undefined` | Externally controlled selected row |
-| `noDataMessage` | `string` | Message when no rows are returned |
-| `queryArgs` | `object` | Arguments forwarded to the query |
-
-## Column definition
-
-```ts
-type Column<T> = {
-    header: string;
-    field: keyof T | ((row: T) => string);
-    width?: number | string;
-};
-```
-
-## When to use each component
-
-| Situation | Component |
-| --- | --- |
-| Full page with toolbar | `DataPage` |
-| Embedded table, standard query | `DataTableForQuery` |
-| Embedded table, real-time push | `DataTableForObservableQuery` |
-| Inline data (no query) | Custom table (out of scope) |
diff --git a/.github/skills/cratis-react-page/references/dialogs.md b/.github/skills/cratis-react-page/references/dialogs.md
deleted file mode 100644
index 4569e6e..0000000
--- a/.github/skills/cratis-react-page/references/dialogs.md
+++ /dev/null
@@ -1,196 +0,0 @@
-# Dialogs — Reference
-
-## Core pattern
-
-Dialogs are **separate components** that receive `closeDialog` as a prop via `DialogProps`. The parent uses `useDialog(DialogComponent)` to get a wrapper and a `show` function.
-
-```tsx
-import { DialogProps } from '@cratis/arc.react/dialogs';
-import { CommandDialog } from '@cratis/components/CommandDialog';
-import { InputTextField } from '@cratis/components/CommandForm';
-import { CreateAccount } from './commands/CreateAccount';
-
-// 1. Define the dialog component
-export const CreateAccountDialog = ({ closeDialog }: DialogProps) => {
-    return (
-        <CommandDialog<CreateAccount>
-            command={CreateAccount}
-            title="Create Account"
-            okLabel="Create"
-        >
-            <InputTextField<CreateAccount> value={c => c.name} label="Account Name" />
-        </CommandDialog>
-    );
-};
-```
-
-```tsx
-// 2. Wire it up in the parent
-import { useDialog } from '@cratis/arc.react/dialogs';
-import { CreateAccountDialog } from './CreateAccountDialog';
-
-export const AccountsPage = () => {
-    const [CreateAccountWrapper, showCreateAccount] = useDialog(CreateAccountDialog);
-
-    return (
-        <>
-            <DataPage ... />
-            <CreateAccountWrapper />
-        </>
-    );
-};
-```
-
-`showCreateAccount()` opens the dialog. `closeDialog` (injected into the dialog component by the framework) closes it.
-
----
-
-## `useDialog`
-
-```tsx
-import { useDialog } from '@cratis/arc.react/dialogs';
-
-const [DialogWrapper, showDialog] = useDialog(MyDialogComponent);
-```
-
-- `DialogWrapper` — render this once in the JSX tree; it controls visibility
-- `showDialog(props?)` — call to open; returns a `Promise<[DialogResult, TResponse?]>`
-
-```tsx
-const [result, response] = await showDialog({ someInitialProp: value });
-if (result === DialogResult.Ok) {
-    // handle confirmed result
-}
-```
-
-Pass props to `showDialog()` when the dialog needs context from the parent (e.g. a selected row to edit).
-
----
-
-## Passing props to a dialog
-
-Define the dialog's props interface extending `DialogProps`:
-
-```tsx
-interface EditAccountDialogProps extends DialogProps {
-    accountId: string;
-    name: string;
-}
-
-export const EditAccountDialog = ({ closeDialog, accountId, name }: EditAccountDialogProps) => {
-    return (
-        <CommandDialog<EditAccount>
-            command={EditAccount}
-            title="Edit Account"
-            initialValues={{ accountId }}
-            currentValues={{ name }}
-        >
-            <InputTextField<EditAccount> value={c => c.name} label="Account Name" />
-        </CommandDialog>
-    );
-};
-```
-
-Then in the parent:
-
-```tsx
-const [EditAccountWrapper, showEditAccount] = useDialog(EditAccountDialog);
-
-// Pass the selected row when opening
-<DataPage onRowSelected={(row) => showEditAccount({ accountId: row.id, name: row.name })} ... />
-<EditAccountWrapper />
-```
-
----
-
-## CommandDialog
-
-Use for dialogs that execute a command on confirm. Import from `@cratis/components/CommandDialog`.
-
-```tsx
-import { CommandDialog } from '@cratis/components/CommandDialog';
-import { InputTextField, NumberField } from '@cratis/components/CommandForm';
-```
-
-**Key props:**
-
-| Prop | Purpose |
-| --- | --- |
-| `command` | Command constructor (proxy-generated class) |
-| `title` | Dialog header text |
-| `okLabel` | Confirm button text (default: `"Ok"`) |
-| `cancelLabel` | Cancel button text (default: `"Cancel"`) |
-| `initialValues` | Values set as the change-tracking baseline (e.g. injected IDs) |
-| `currentValues` | Values to pre-populate the fields for editing |
-| `isValid` | Extra validity gate in addition to field-level validation |
-| `onBeforeExecute` | Transform command values just before `.execute()` |
-
-`CommandDialog` automatically disables the confirm button until all required fields are filled.
-
-Use `initialValues` for values that must be present but not visible (e.g. a parent entity ID). Do **not** set them in `onBeforeExecute` — they won't be visible to form validation.
-
----
-
-## CommandForm field components
-
-All field components come from `@cratis/components/CommandForm`. Pass the command type as the generic parameter so `value` is fully typed.
-
-```tsx
-import {
-    InputTextField,  // text input
-    NumberField,     // number input
-    CheckboxField,   // boolean toggle
-    DateField,       // date picker
-    DropdownField,   // select from options list
-    TextAreaField,   // multi-line text
-} from '@cratis/components/CommandForm';
-
-<InputTextField<MyCommand> value={c => c.title} label="Title" />
-<NumberField<MyCommand> value={c => c.quantity} label="Qty" min={1} />
-<CheckboxField<MyCommand> value={c => c.isActive} label="Active" />
-<DateField<MyCommand> value={c => c.dueDate} label="Due date" />
-<DropdownField<MyCommand>
-    value={c => c.status}
-    label="Status"
-    options={statusOptions}
-    optionLabel="label"
-    optionValue="value"
-/>
-<TextAreaField<MyCommand> value={c => c.notes} label="Notes" rows={3} />
-```
-
-The `value` prop takes a function `(commandInstance) => property`. This drives both reading the value and writing it back on change.
-
----
-
-## Dialog (data-only, no command)
-
-Use when the dialog collects data and returns it without executing a command.
-
-```tsx
-import { DialogProps, DialogResult } from '@cratis/arc.react/dialogs';
-import { Dialog } from '@cratis/components/Dialogs';
-import { InputText } from 'primereact/inputtext';
-import { useState } from 'react';
-
-export const RenameDialog = ({ closeDialog }: DialogProps<{ name: string }>) => {
-    const [name, setName] = useState('');
-
-    return (
-        <Dialog
-            title="Rename"
-            isValid={name.trim().length > 0}
-            onConfirm={() => closeDialog(DialogResult.Ok, { name })}
-            onCancel={() => closeDialog(DialogResult.Cancelled)}
-        >
-            <InputText
-                value={name}
-                onChange={event => setName(event.target.value)}
-                autoFocus
-            />
-        </Dialog>
-    );
-};
-```
-
-Never import `Dialog` from `primereact/dialog` — always use `@cratis/components/Dialogs`.
diff --git a/.github/skills/cratis-react-page/references/mvvm.md b/.github/skills/cratis-react-page/references/mvvm.md
deleted file mode 100644
index 3ba11ca..0000000
--- a/.github/skills/cratis-react-page/references/mvvm.md
+++ /dev/null
@@ -1,133 +0,0 @@
-# MVVM — Reference
-
-The Arc MVVM pattern keeps page logic in plain TypeScript classes (view models) and keeps components purely declarative.
-
-## When to use MVVM
-
-- Page has complex coordinated state (selected item, filters, multiple dialogs)
-- Logic needs unit-testing independent of React
-- You want to share state across child components via injection
-
-For simple pages, MVVM is optional — use regular hooks directly in the component instead.
-
-## Setup
-
-Install packages if not already present:
-
-```
-npm install @cratis/arc.react.mvvm tsyringe reflect-metadata
-```
-
-Ensure `tsconfig.json` enables decorators:
-
-```json
-{
-    "compilerOptions": {
-        "experimentalDecorators": true,
-        "emitDecoratorMetadata": true
-    }
-}
-```
-
-Import `reflect-metadata` once, at the entry point of your app:
-
-```tsx
-import 'reflect-metadata';
-```
-
-## View model class
-
-```ts
-import { injectable } from 'tsyringe';
-import { makeAutoObservable } from 'mobx';
-
-@injectable()
-export class AccountsViewModel {
-    selectedAccount?: AccountSummary = undefined;
-
-    constructor() {
-        makeAutoObservable(this);
-    }
-
-    selectAccount(account: AccountSummary) {
-        this.selectedAccount = account;
-    }
-}
-```
-
-- `@injectable()` — registers the class with tsyringe for DI
-- `makeAutoObservable(this)` — makes all fields reactive (MobX)
-
-## withViewModel
-
-```tsx
-import { withViewModel } from '@cratis/arc.react.mvvm';
-
-export const AccountsPage = withViewModel(AccountsViewModel, ({ viewModel }) => {
-    return (
-        <DataPage
-            query={AllAccounts}
-            columns={...}
-            onRowSelected={(row) => viewModel.selectAccount(row)}
-            detailPanel={() => viewModel.selectedAccount
-                ? <AccountDetail account={viewModel.selectedAccount} />
-                : null
-            }
-        />
-    );
-});
-```
-
-The view model instance is created once per mount and disposed on unmount. It is the same instance for the whole component tree under `withViewModel`.
-
-## IHandleProps — reactive props
-
-When a child component needs to receive a prop and react to its changes, implement `IHandleProps<T>`:
-
-```ts
-import { IHandleProps } from '@cratis/arc.react.mvvm';
-
-interface DetailProps {
-    account: AccountSummary;
-}
-
-@injectable()
-export class AccountDetailViewModel implements IHandleProps<DetailProps> {
-    account!: AccountSummary;
-
-    propsChanged(props: DetailProps): void {
-        this.account = props.account;
-    }
-}
-```
-
-`propsChanged` is called whenever the parent passes new props, allowing the view model to react.
-
-## Dependency injection in view models
-
-Use tsyringe constructor injection. Cratis registers common singletons (e.g., `IEventStore`, query/command types):
-
-```ts
-@injectable()
-export class AccountsViewModel {
-    constructor(
-        private readonly _eventLog: IEventLog,
-    ) {
-        makeAutoObservable(this);
-    }
-}
-```
-
-## MVVM context
-
-Wrap the app (or route root) in `<MVVM>` to enable the DI container:
-
-```tsx
-import { MVVM } from '@cratis/arc.react.mvvm';
-
-<MVVM>
-    <App />
-</MVVM>
-```
-
-If you are using `<Arc>`, it already includes `<MVVM>` internally — do not double-wrap.
diff --git a/.github/skills/cratis-readmodel/SKILL.md b/.github/skills/cratis-readmodel/SKILL.md
deleted file mode 100644
index a0b37f3..0000000
--- a/.github/skills/cratis-readmodel/SKILL.md
+++ /dev/null
@@ -1,249 +0,0 @@
----
-name: cratis-readmodel
-description: Step-by-step guidance for creating a Cratis Chronicle read model from scratch — defining events, choosing between projection and reducer, [ReadModel] record with static query methods, and the generated TypeScript proxy in React. Use when creating a read model, working with [EventType], [ReadModel], IProjectionFor, IReducerFor, observable queries, or deriving state from events. For adding a projection or reactor to an existing read model, use add-projection instead.
----
-
-# Creating a Cratis Read Model
-
-A read model is derived state built from events. The path is:
-
-```
-[EventType] records  →  [ReadModel] record + static query methods  →  projection or reducer  →  TypeScript proxy  →  React
-```
-
----
-
-## Step 1 — Define your events
-
-Events are the source of truth. Define each as a `record` decorated with `[EventType]`. Name them in **past tense**.
-
-```csharp
-// Domain/Events/AccountEvents.cs
-using Cratis.Chronicle.Events;
-
-[EventType]
-public record DebitAccountOpened(string Name, Guid OwnerId);
-
-[EventType]
-public record DebitAccountClosed();
-
-[EventType]
-public record FundsDeposited(decimal Amount);
-
-[EventType]
-public record FundsWithdrawn(decimal Amount);
-```
-
-Good event design:
-- One clear purpose per event — do not mix concerns
-- No nullables unless genuinely optional
-- Properties represent facts, not intent
-
----
-
-## Step 2 — Define the read model record
-
-Decorate the record with `[ReadModel]` and add **static query methods** directly on it. The proxy generator turns each static method into a TypeScript query class.
-
-```csharp
-// Domain/ReadModels/AccountSummary.cs
-using Cratis.Arc.Queries.ModelBound;
-using MongoDB.Driver;
-
-[ReadModel]
-public record AccountSummary(AccountId Id, string Name, OwnerId OwnerId, decimal Balance, bool IsClosed)
-{
-    // Snapshot query — returns current data once
-    public static async Task<IEnumerable<AccountSummary>> AllAccounts(
-        IMongoCollection<AccountSummary> collection)
-        => await collection.Find(Builders<AccountSummary>.Filter.Empty).ToListAsync();
-
-    public static async Task<AccountSummary?> GetAccount(
-        AccountId id,
-        IMongoCollection<AccountSummary> collection)
-        => await collection.Find(a => a.Id == id).FirstOrDefaultAsync();
-
-    // Observable query — pushes updates in real time
-    public static ISubject<IEnumerable<AccountSummary>> ObserveAllAccounts(
-        IMongoCollection<AccountSummary> collection)
-        => collection.Observe();
-}
-```
-
-**Rules:**
-- `[ReadModel]` attribute is **required** for proxy generation and runtime routing
-- Static methods must be `public static` and return the record type, a collection of it, or `ISubject<T>` for real-time push
-- Do **not** return `Task<ISubject<T>>` — observable methods must return `ISubject<T>` directly
-- Use `ConceptAs<T>` wrappers for all identity fields — never raw `Guid`
-- One read model per use case — do not reuse them
-
----
-
-## Step 3 — Choose: projection or reducer?
-
-| | Projection | Reducer |
-|-| ---------- | ------- |
-| **Best for** | Shaped read models with mapping logic, joins, children | Running aggregates: balances, counts, sums |
-| **How it works** | Declarative mapping: each event updates specific fields | Receives events one by one and returns the new full state |
-| **When to pick** | The read model shape comes mostly from mapping event fields | The state is a function of *accumulating* multiple events |
-
-For the `AccountSummary` above: use a **projection** for name/owner fields and a **reducer** for balance (a running total). In practice, reducers cover both when the aggregate combines both concerns.
-
----
-
-## Step 4A — Implement a projection
-
-```csharp
-// Domain/Projections/AccountSummaryProjection.cs
-using Cratis.Chronicle.Projections;
-
-public class AccountSummaryProjection : IProjectionFor<AccountSummary>
-{
-    public void Define(IProjectionBuilderFor<AccountSummary> builder) => builder
-        .From<DebitAccountOpened>(from => from
-            .Set(m => m.Name).To(e => e.Name)
-            .Set(m => m.OwnerId).To(e => e.OwnerId)
-            .Set(m => m.Balance).WithValue(0m))
-        .From<FundsDeposited>(from => from
-            .Add(m => m.Balance).With(e => e.Amount))
-        .From<FundsWithdrawn>(from => from
-            .Subtract(m => m.Balance).With(e => e.Amount))
-        .From<DebitAccountClosed>(from => from
-            .Set(m => m.IsClosed).WithValue(true));
-}
-```
-
-- Discovered automatically — no registration needed
-- `IProjectionFor<T>` is keyed by **event source ID** by default (the `Id` passed to `IEventLog.Append`)
-- Appended `tags`, `eventSourceType`, and `eventStreamType` do not filter projections directly; use reducers or reactors alongside the projection when you need metadata-based filtering
-- See `references/projections.md` for joins, auto-mapping, children, composite keys
-
-### Model-bound shorthand (for simple cases)
-
-```csharp
-using Cratis.Chronicle.Keys;
-using Cratis.Chronicle.Projections.ModelBound;
-
-public record AccountInfo(
-    [Key] Guid Id,
-    [FromEvent<DebitAccountOpened>] string Name,   // auto-maps matching property
-    [SetFrom<DebitAccountOpened>(nameof(DebitAccountOpened.OwnerId))] Guid OwnerId
-);
-```
-
----
-
-## Step 4B — Implement a reducer
-
-Use a reducer when the state is built by accumulating values across events:
-
-```csharp
-// Domain/Reducers/AccountBalanceReducer.cs
-using Cratis.Chronicle.Events;
-using Cratis.Chronicle.Reducers;
-
-public class AccountBalanceReducer : IReducerFor<AccountBalance>
-{
-    public AccountBalance Opened(DebitAccountOpened @event, AccountBalance? current, EventContext context)
-        => new(0m, context.Occurred);
-
-    public AccountBalance Deposited(FundsDeposited @event, AccountBalance? current, EventContext context)
-        => (current ?? new(0m, context.Occurred)) with { Balance = (current?.Balance ?? 0m) + @event.Amount };
-
-    public AccountBalance Withdrawn(FundsWithdrawn @event, AccountBalance? current, EventContext context)
-        => (current ?? new(0m, context.Occurred)) with { Balance = (current?.Balance ?? 0m) - @event.Amount };
-}
-
-public record AccountBalance(decimal Balance, DateTimeOffset LastUpdated);
-```
-
-- Return the **complete new state** — do not mutate `current`
-- `current` is `null` on the first event for a given event source
-- `EventContext` provides `Occurred`, `EventSourceId`, `SequenceNumber`, `CorrelationId`
-- Discovered automatically — no registration needed
-- Add `[FilterEventsByTag]`, `[EventSourceType]`, and `[EventStreamType]` when the reducer should only observe events appended with matching metadata
-
----
-
-## Step 5 — Expose read model queries
-
-Query methods live **directly on the `[ReadModel]` record** as static methods (see Step 2). You do **not** need a separate controller or `IReadModels` injection.
-
-The method name becomes the TypeScript proxy class name — use descriptive names like `AllAccounts`, `GetAccount`, `ObserveAllAccounts`.
-
-### Snapshot (one-time) queries
-
-```csharp
-[ReadModel]
-public record AccountSummary(AccountId Id, string Name, decimal Balance)
-{
-    public static async Task<IEnumerable<AccountSummary>> AllAccounts(
-        IMongoCollection<AccountSummary> collection)
-        => await collection.Find(_ => true).ToListAsync();
-
-    public static async Task<AccountSummary?> GetAccount(
-        AccountId id,
-        IMongoCollection<AccountSummary> collection)
-        => await collection.Find(a => a.Id == id).FirstOrDefaultAsync();
-}
-```
-
-### Observable (real-time push) queries
-
-Return `ISubject<T>` to push updates as projection changes land:
-
-```csharp
-[ReadModel]
-public record AccountSummary(AccountId Id, string Name, decimal Balance)
-{
-    public static ISubject<IEnumerable<AccountSummary>> ObserveAllAccounts(
-        IMongoCollection<AccountSummary> collection)
-        => collection.Observe();
-
-    public static ISubject<AccountSummary> ObserveAccount(
-        AccountId id,
-        IMongoCollection<AccountSummary> collection)
-        => collection.Observe(a => a.Id == id);
-}
-```
-
-When the frontend uses an observable query, the query proxy type changes from `QueryFor` to `ObservableQueryFor`, and `DataPage` uses the `observableQuery` prop instead of `query`.
-
----
-
-## Step 6 — Build and use in React
-
-```bash
-dotnet build   # generates TypeScript proxies
-```
-
-```tsx
-import { AllAccounts } from '../api/Accounts/AllAccounts';
-
-export const AccountList = () => {
-    const [accounts] = AllAccounts.use();
-
-    if (accounts.isPerforming) return <Spinner />;
-
-    return (
-        <ul>
-            {accounts.data.map(a => (
-                <li key={a.id}>{a.name} — ${a.balance}</li>
-            ))}
-        </ul>
-    );
-};
-```
-
-For building full pages with filtering, sorting, and command actions — see the `cratis-react-page` skill.
-
----
-
-## Reference files
-
-| File | What's in it |
-| ---- | ------------ |
-| `references/projections.md` | Full builder API: `Set`, `Add`, `Join`, `Children`, `AutoMap`, composite keys |
-| `references/reducers.md` | Reducer signatures, async, passive, snapshot behaviour |
-| `references/events.md` | `[EventType]`, appending, `AppendResult`, tags, constraints |
-| `references/queries.md` | Query result shape, observable queries, paging |
diff --git a/.github/skills/cratis-readmodel/evals/evals.json b/.github/skills/cratis-readmodel/evals/evals.json
deleted file mode 100644
index 3493423..0000000
--- a/.github/skills/cratis-readmodel/evals/evals.json
+++ /dev/null
@@ -1,33 +0,0 @@
-{
-  "skill_name": "cratis-readmodel",
-  "evals": [
-    {
-      "id": 1,
-      "prompt": "I need a read model that shows all employees with their department name and start date. The read model should be updated whenever an employee is registered (EmployeeRegistered event has EmployeeName, DepartmentId) or when a department name changes (DepartmentNameChanged has DepartmentId, NewName). Use the model-bound/attribute approach if possible.",
-      "expected_output": "A C# file with [ReadModel] record using [FromEvent<>] and possibly [Join<>] attributes, with a static AllEmployees query method returning ISubject<IEnumerable<Employee>>. Includes [EventType] records if not already defined.",
-      "files": [],
-      "assertions": [
-        "Output contains [ReadModel] attribute on a record",
-        "Output includes a static query method on the read model record",
-        "Query method returns ISubject<T> for real-time push (observable query)",
-        "Output uses [FromEvent<T>] or IProjectionFor<T> for projection",
-        "Output uses ConceptAs<T> for id types",
-        "Output does NOT use IProjectionFor with joins reading from other read models (events only)"
-      ]
-    },
-    {
-      "id": 2,
-      "prompt": "Create a read model to track the current stock level for each product. Events are: ProductAdded(ProductId, InitialStock), StockIncreased(ProductId, Amount), StockDecreased(ProductId, Amount). I need a reducer since the logic involves arithmetic accumulation.",
-      "expected_output": "A C# file with [ReadModel] record + IReducerFor<T> class that handles all three events. Includes [EventType] records and a static query method.",
-      "files": [],
-      "assertions": [
-        "Output contains IReducerFor<T> implementation",
-        "Reducer handles null current state gracefully (first event initialization)",
-        "Reducer methods are pure (no side effects, no I/O calls)",
-        "Output contains [EventType] records for all three events",
-        "Output includes a static query method on the read model",
-        "Uses 'with' expressions for record state updates"
-      ]
-    }
-  ]
-}
diff --git a/.github/skills/cratis-readmodel/references/events.md b/.github/skills/cratis-readmodel/references/events.md
deleted file mode 100644
index 34d1e48..0000000
--- a/.github/skills/cratis-readmodel/references/events.md
+++ /dev/null
@@ -1,91 +0,0 @@
-# Events — Reference
-
-## [EventType] attribute
-
-```csharp
-using Cratis.Chronicle.Events;
-
-[EventType]
-public record OrderPlaced(string CustomerId, decimal Total);
-```
-
-Every event must be decorated with `[EventType]` — this makes it discoverable and registers its schema with Chronicle.
-
----
-
-## Good event design
-
-- **Past tense**: `OrderPlaced`, `UserOnboarded`, `BookReturned` ✓ — not `PlaceOrder`, `OnboardUser`
-- **One purpose**: an `AddressChanged` event should only carry address fields, not payment info
-- **No nullables** unless the field is genuinely optional (e.g. `string? MiddleName`)
-- **Immutable facts**: events represent something that *has happened* — do not use them to encode intent or possibility
-
----
-
-## Appending an event
-
-Inject `IEventLog` and call `Append(eventSourceId, eventInstance)`:
-
-```csharp
-public class OrdersController(IEventLog eventLog) : ControllerBase
-{
-    [HttpPost]
-    public async Task PlaceOrder([FromBody] PlaceOrder command)
-    {
-        var result = await eventLog.Append(
-            command.OrderId,
-            new OrderPlaced(command.CustomerId, command.Total));
-
-        if (!result.IsSuccess)
-        {
-            // result.HasConcurrencyViolation — two requests raced
-            // result.HasConstraintViolations — uniqueness constraint failed
-        }
-    }
-}
-```
-
-The first argument is the **event source ID** — the identity the event belongs to (analogous to an aggregate root ID). Projections and reducers are keyed by this ID by default.
-
-You can also append metadata that downstream reducers and reactors can filter on:
-
-```csharp
-await eventLog.Append(
-    command.OrderId,
-    new OrderPlaced(command.CustomerId, command.Total),
-    eventStreamType: "fulfillment",
-    eventSourceType: "order",
-    tags: ["priority"]);
-```
-
----
-
-## Constraints (uniqueness)
-
-Enforce uniqueness at append time without application-level checks:
-
-```csharp
-public class UniqueOrderNumberConstraint : IUniqueConstraintFor<OrderPlaced>
-{
-    public string GetConstraintValue(OrderPlaced @event) => @event.OrderNumber;
-}
-```
-
-Violation surfaces in `AppendResult.HasConstraintViolations`.
-
----
-
-## Tags
-
-```csharp
-[EventType]
-[Tag("high-value")]
-public record LargeOrderPlaced(decimal Total);
-
-// Or apply at append time:
-await eventLog.Append(orderId, new LargeOrderPlaced(2500m), tags: ["priority"]);
-```
-
-Tags allow reducers and reactors to filter which appended events they handle when you use `[FilterEventsByTag]`. `[Tag]` and `[Tags]` on projections, reducers, and reactors label the observer or event type; they do not filter the observer by themselves.
-
-See `Documentation/events/filtering/` for tag, event source type, and event stream type filtering examples.
diff --git a/.github/skills/cratis-readmodel/references/projections.md b/.github/skills/cratis-readmodel/references/projections.md
deleted file mode 100644
index 4b8b524..0000000
--- a/.github/skills/cratis-readmodel/references/projections.md
+++ /dev/null
@@ -1,155 +0,0 @@
-# Projections — Reference
-
-## Model-bound projections (preferred)
-
-Put projection metadata on the read model first. This is the default choice for Cratis projects because it keeps the read model and its projection behavior together.
-
-```csharp
-using Cratis.Chronicle.Keys;
-using Cratis.Chronicle.Projections.ModelBound;
-
-public record InvoiceInfo(
-    [Key] Guid Id,
-    [FromEvent<InvoiceCreated>] string Number,
-    [AddFrom<LineItemAdded>(nameof(LineItemAdded.Price))] decimal RunningTotal,
-    [SetFromContext<InvoicePaid>(nameof(EventContext.Occurred))] DateTimeOffset? PaidAt);
-```
-
-For child relationships where later child events arrive on the child event source, set `parentKey` on the child type's class-level `FromEvent`:
-
-```csharp
-public record Invoice(
-    [Key] Guid Id,
-    [ChildrenFrom<LineItemAdded>(
-        key: nameof(LineItemAdded.LineItemId),
-        identifiedBy: nameof(LineItem.Id),
-        parentKey: nameof(LineItemAdded.InvoiceId))]
-    IEnumerable<LineItem> Lines);
-
-[FromEvent<LineItemRenamed>(parentKey: nameof(LineItemRenamed.InvoiceId))]
-public record LineItem(
-    [Key] Guid Id,
-    string Description);
-```
-
-### Attribute reference
-
-| Attribute | Equivalent builder |
-| --------- | ------------------ |
-| `[Key]` | Default key (event source ID) |
-| `[FromEvent<T>]` | `.AutoMap()` for event T |
-| `[FromEvent<T>(key: nameof(T.Prop))]` | `.UsingKey(e => e.Prop)` |
-| `[FromEvent<T>(parentKey: nameof(T.Prop))]` | `.UsingParentKey(e => e.Prop)` on child projections |
-| `[SetFrom<T>(nameof(...))]` | `.Set(...).To(...)` |
-| `[AddFrom<T>(nameof(...))]` | `.Add(...).With(...)` |
-| `[SubtractFrom<T>(nameof(...))]` | `.Subtract(...).With(...)` |
-| `[SetFromContext<T>(nameof(...))]` | `.Set(...).ToEventContextProperty(...)` |
-| `[Increment<T>]` | counter increment |
-| `[Decrement<T>]` | counter decrement |
-| `[RemovedWith<T>]` | `.RemovedWith<T>()` |
-| `[Passive]` | `.Passive()` |
-| `[NotRewindable]` | `.NotRewindable()` |
-
----
-
-## Declarative projection builder (`IProjectionFor<T>`) — fallback
-
-```csharp
-public class InvoiceProjection : IProjectionFor<Invoice>
-{
-    public void Define(IProjectionBuilderFor<Invoice> builder) => builder
-        .From<InvoiceCreated>(from => from.AutoMap())        // map all matching property names
-        .From<InvoicePaid>(from => from
-            .Set(m => m.PaidAt).ToEventContextProperty(c => c.Occurred)
-            .Set(m => m.Status).WithValue(InvoiceStatus.Paid))
-        .From<LineItemAdded>(from => from
-            .Add(m => m.TotalAmount).With(e => e.Price))
-        .Join<CustomerRegistered>(j => j
-            .On(m => m.CustomerId)
-            .Set(m => m.CustomerName).To(e => e.Name))
-        .RemovedWith<InvoiceDeleted>()
-        .Children<LineItem>(m => m.Lines, cb => cb
-            .IdentifiedBy(li => li.LineItemId)
-            .From<LineItemAdded>(from => from.AutoMap())
-            .RemovedWith<LineItemRemoved>());
-}
-```
-
-### Builder method reference
-
-| Method | Purpose |
-| ------ | ------- |
-| `.From<TEvent>(cb)` | Handle an event type |
-| `.AutoMap()` | Map all event properties with matching names to the read model |
-| `.Set(m => m.Prop).To(e => e.Prop)` | Explicit property mapping |
-| `.Set(m => m.Prop).WithValue(val)` | Set a constant |
-| `.Set(m => m.Prop).ToEventContextProperty(c => c.X)` | Map from event metadata |
-| `.Add(m => m.Prop).With(e => e.X)` | Add (numeric) |
-| `.Subtract(m => m.Prop).With(e => e.X)` | Subtract |
-| `.Count(m => m.Prop)` | Increment a counter |
-| `.Join<TEvent>(j => j.On(key).Set(...))` | Cross-stream join another event |
-| `.RemovedWith<TEvent>()` | Delete the read model on this event |
-| `.Children<TChild>(m => m.Coll, cb)` | Manage a child collection |
-| `.FromEvery(cb)` | Apply mapping to every event type |
-| `.UsingKey(e => e.Prop)` | Override the key (default: event source ID) |
-| `.Passive()` | On-demand only, no active observer |
-| `.NotRewindable()` | Forward-only, no replay |
-
-### Event context properties
-
-```csharp
-.ToEventContextProperty(c => c.Occurred)       // DateTimeOffset
-.ToEventContextProperty(c => c.EventSourceId)  // string
-.ToEventContextProperty(c => c.SequenceNumber) // long
-.ToEventContextProperty(c => c.CorrelationId)  // Guid
-```
-
-### Composite keys
-
-```csharp
-builder.UsingCompositeKey<MonthlyReport>(key => key
-    .Set(k => k.Year).ToEventContextProperty(c => c.Occurred.Year)
-    .Set(k => k.Month).ToEventContextProperty(c => c.Occurred.Month));
-```
-
-## Reading projected read models
-
-```csharp
-// Inject IReadModels in a controller
-[HttpGet("{id}")]
-public async Task<AccountSummary?> Get(Guid id)
-    => await readModels.GetOne<AccountSummary>(id);
-
-// Strong consistency (replay events synchronously before returning)
-var account = await readModels.GetOneWithImmediateProjection<AccountSummary>(id);
-
-// All instances
-var all = await readModels.GetAll<AccountSummary>();
-```
-
----
-
-## Appended event metadata and projections
-
-When you append events, you can set tags, event source type, and event stream type:
-
-```csharp
-await eventLog.Append(
-    EventSourceId.New(),
-    new OrderPlaced(42m),
-    eventStreamType: "fulfillment",
-    eventSourceType: "order",
-    tags: ["priority"]);
-```
-
-Projection definitions do not use `[FilterEventsByTag]`, `[EventSourceType]`, or `[EventStreamType]` as observer filters. Projections choose input through event types, joins, and event sequence selection.
-
-Use appended metadata when you need:
-
-- A reducer or reactor alongside the projection to observe only matching events
-- Event context values inside projection mappings
-- Consistent metadata across downstream observers that react to the same append operation
-
-`[Tag]` and `[Tags]` on a projection label the projection definition; they do not filter appended events.
-
-For reducer and reactor filtering examples, see `Documentation/events/filtering/`.
diff --git a/.github/skills/cratis-readmodel/references/queries.md b/.github/skills/cratis-readmodel/references/queries.md
deleted file mode 100644
index cf78bc3..0000000
--- a/.github/skills/cratis-readmodel/references/queries.md
+++ /dev/null
@@ -1,104 +0,0 @@
-# Queries — Reference
-
-## Query endpoint patterns
-
-### Collection query
-
-```csharp
-[HttpGet]
-public IEnumerable<AccountSummary> AllAccounts()
-    => collection.Find(_ => true).ToList();
-```
-
-### Single item query
-
-```csharp
-[HttpGet("{id}")]
-public AccountSummary? GetAccount(Guid id)
-    => collection.Find(a => a.Id == id).FirstOrDefault();
-```
-
-### Filtered query (with proxy parameter)
-
-```csharp
-[HttpGet("search")]
-public IEnumerable<AccountSummary> Search([FromQuery] string? filter)
-    => collection.Find(a => a.Name.StartsWith(filter ?? string.Empty)).ToList();
-```
-
-The `[FromQuery]` parameter is included in the generated TypeScript proxy.
-
-### Observable (real-time) query
-
-Return `ISubject<T>` to push data to clients over WebSocket:
-
-```csharp
-[HttpGet("live")]
-public ISubject<IEnumerable<AccountSummary>> AllAccountsLive()
-{
-    var observable = new ClientObservable<IEnumerable<AccountSummary>>();
-    observable.OnNext(collection.Find(_ => true).ToList());
-
-    var changeStream = collection.Watch();
-    observable.ClientDisconnected += () => changeStream.Dispose();
-    Task.Run(async () =>
-    {
-        await foreach (var _ in changeStream.ToAsyncEnumerable())
-            observable.OnNext(collection.Find(_ => true).ToList());
-    });
-
-    return observable;
-}
-```
-
-The proxy generator produces an `ObservableQuery` TypeScript class for `ISubject<T>` return types. The React hook `useObservableQuery()` is used automatically.
-
----
-
-## QueryResult shape (frontend)
-
-```ts
-interface QueryResultWithState<T> {
-    data: T;
-    isSuccess: boolean;
-    isAuthorized: boolean;
-    isValid: boolean;
-    validationResults: ValidationResult[];
-    hasExceptions: boolean;
-    exceptionMessages: string[];
-    paging: { page: number; pageSize: number; totalItems: number; totalPages: number };
-
-    // React-specific:
-    hasData: boolean;       // non-null and non-empty
-    isPerforming: boolean;  // request in flight
-}
-```
-
----
-
-## React usage
-
-```tsx
-// Standard query — returns [result, requery]
-const [accounts, refresh] = AllAccounts.use();
-
-// With parameters
-const [results] = Search.use({ filter: searchText });
-
-// Observable query — returns [result] only (no manual refresh)
-const [liveAccounts] = AllAccountsLive.use();
-```
-
-For full page layouts with tables and menu actions, see the `cratis-react-page` skill.
-
----
-
-## Naming conventions
-
-The **method name** on the controller becomes the TypeScript proxy class name. Make it descriptive.
-
-| ✅ Good | ❌ Avoid |
-| ------- | ------- |
-| `AllAccounts` | `Get`, `GetAll`, `List` |
-| `AccountsByOwner` | `Query`, `Fetch` |
-| `AllAccountsLive` | `Observable`, `Live` |
diff --git a/.github/skills/cratis-readmodel/references/reducers.md b/.github/skills/cratis-readmodel/references/reducers.md
deleted file mode 100644
index 22e5237..0000000
--- a/.github/skills/cratis-readmodel/references/reducers.md
+++ /dev/null
@@ -1,136 +0,0 @@
-# Reducers — Reference
-
-## Signatures
-
-All public methods with a `[EventType]` record as the first parameter are treated as event handlers. All the following signatures are valid:
-
-```csharp
-public TState Handle(TEvent @event, TState? current, EventContext context)
-public TState Handle(TEvent @event, TState? current)
-public Task<TState> Handle(TEvent @event, TState? current, EventContext context)
-public Task<TState> Handle(TEvent @event, TState? current)
-```
-
-Method names are yours to choose — Chronicle matches by the event type parameter.
-
----
-
-## EventContext properties
-
-| Property | Type | Description |
-| -------- | ---- | ----------- |
-| `context.Occurred` | `DateTimeOffset` | When the event was appended |
-| `context.EventSourceId` | `EventSourceId` | The aggregate root identifier |
-| `context.SequenceNumber` | `EventSequenceNumber` | Position in the event sequence |
-| `context.CorrelationId` | `CorrelationId` | Correlation ID for causality tracking |
-
----
-
-## Full example: shopping cart
-
-```csharp
-public record CartItem(string Sku, int Quantity, decimal UnitPrice);
-public record CartState(IReadOnlyList<CartItem> Items, decimal Total, bool IsCheckedOut);
-
-public class CartReducer : IReducerFor<CartState>
-{
-    public CartState Created(CartCreated @event, CartState? current, EventContext context)
-        => new([], 0m, false);
-
-    public CartState ItemAdded(CartItemAdded @event, CartState? current, EventContext context)
-    {
-        var items = (current?.Items ?? []).ToList();
-        var existing = items.FirstOrDefault(i => i.Sku == @event.Sku);
-        if (existing is not null)
-        {
-            items.Remove(existing);
-            items.Add(existing with { Quantity = existing.Quantity + @event.Quantity });
-        }
-        else
-        {
-            items.Add(new CartItem(@event.Sku, @event.Quantity, @event.UnitPrice));
-        }
-        var total = items.Sum(i => i.Quantity * i.UnitPrice);
-        return new CartState(items, total, false);
-    }
-
-    public CartState ItemRemoved(CartItemRemoved @event, CartState? current, EventContext context)
-    {
-        var items = (current?.Items ?? []).Where(i => i.Sku != @event.Sku).ToList();
-        return new CartState(items, items.Sum(i => i.Quantity * i.UnitPrice), false);
-    }
-
-    public CartState CheckedOut(CartCheckedOut @event, CartState? current, EventContext context)
-        => (current ?? new([], 0m, false)) with { IsCheckedOut = true };
-}
-```
-
----
-
-## Passive reducers
-
-A passive reducer is not an active observer — it computes state on demand. Useful for previews or draft calculations:
-
-```csharp
-[Passive]
-public class DraftOrderReducer : IReducerFor<DraftOrder> { ... }
-```
-
-Call explicitly rather than subscribing automatically:
-
-```csharp
-var state = await readModels.GetOne<DraftOrder>(orderId);
-```
-
----
-
-## Reading reducer state
-
-```csharp
-// Single instance
-var cart = await readModels.GetOne<CartState>(cartId);
-
-// All instances
-var allCarts = await readModels.GetAll<CartState>();
-```
-
----
-
-## Filtering reducers by appended event metadata
-
-Use reducer filters when the reducer should only observe a subset of appended events:
-
-```csharp
-using Cratis.Chronicle;
-using Cratis.Chronicle.Events;
-using Cratis.Chronicle.Reducers;
-
-[FilterEventsByTag("priority")]
-[EventSourceType("order")]
-[EventStreamType("fulfillment")]
-public class PriorityOrderReducer : IReducerFor<PriorityOrderState>
-{
-    public PriorityOrderState Ordered(OrderPlaced @event, PriorityOrderState? current, EventContext context) =>
-        new((current?.Count ?? 0) + 1);
-}
-
-public record PriorityOrderState(int Count);
-```
-
-Match the reducer filters when you append:
-
-```csharp
-await eventLog.Append(
-    EventSourceId.New(),
-    new OrderPlaced(42m),
-    eventStreamType: "fulfillment",
-    eventSourceType: "order",
-    tags: ["priority"]);
-```
-
-- `[FilterEventsByTag]` matches any appended or static event tag
-- `[EventSourceType]` matches the appended `eventSourceType`
-- `[EventStreamType]` matches the appended `eventStreamType`
-- `[Tag]` and `[Tags]` label the reducer; they do not filter events
-
-For fuller guidance, see `Documentation/events/filtering/`.
diff --git a/.github/skills/cratis-specs-csharp/SKILL.md b/.github/skills/cratis-specs-csharp/SKILL.md
deleted file mode 100644
index c6c81ba..0000000
--- a/.github/skills/cratis-specs-csharp/SKILL.md
+++ /dev/null
@@ -1,174 +0,0 @@
----
-name: cratis-specs-csharp
-description: Step-by-step guidance for writing C# specs in Cratis using BDD-style Specification by Example — the Establish/Because/should_ pattern, for_/when_/and_ folder hierarchy, reusable given/ contexts, NSubstitute mocking, and Chronicle integration specs. Use when writing C# unit or integration specs, creating spec files or folders, structuring the for_<Type>/when_<behavior>/and_<condition> hierarchy, mocking with NSubstitute, writing given/ reusable contexts, or using ShouldEqual/ShouldBeTrue assertions. For integration specs specifically tied to a vertical slice command, write-specs offers a more focused workflow.
----
-
-## Core philosophy
-
-Specs are **executable documentation** — the folder tree reads like a spec sheet. Favor readability over DRY. Each spec file has:
-- One action under test (`Because`)
-- One setup (`Establish`)
-- One or more focused assertions (`should_*`)
-
----
-
-## Step 1 — Choose the spec type
-
-| Scenario | Spec type |
-| --- | --- |
-| Isolated unit logic (no I/O) | Unit spec in `for_<ClassName>/` |
-| A full vertical slice (command → event) | Chronicle integration spec in slice folder |
-| Complex setup shared across many specs | Reusable context in `given/` |
-
----
-
-## Step 2 — Create the folder structure
-
-```
-for_<ClassName>/
-├── given/
-│   ├── all_dependencies.cs       ← mocks all deps, inherits Specification
-│   └── a_<system_under_test>.cs  ← creates SUT, inherits all_dependencies
-├── when_<behavior>/              ← behavior with multiple outcomes
-│   ├── and_<condition>.cs
-│   └── with_<state>.cs
-└── when_<simple_behavior>.cs     ← single outcome = single file
-```
-
-Folder/file names read as English sentences:
-- `for_Changeset / when_adding_changes / and_there_are_differences`
-- `for_AuthorService / when_registering / and_name_already_exists`
-
----
-
-## Step 3 — Write a spec
-
-```csharp
-// for_KeyHelper/when_combining_parts.cs
-namespace MyApp.for_KeyHelper;
-
-public class when_combining_parts : Specification
-{
-    object[] _parts;
-    string _result;
-
-    void Establish() => _parts = ["First", "Second", "Third"];
-
-    void Because() => _result = KeyHelper.Combine(_parts);
-
-    [Fact] void should_combine_all_parts() => _result.ShouldEqual("First+Second+Third");
-    [Fact] void should_not_be_empty() => _result.ShouldNotBeEmpty();
-}
-```
-
-Rules:
-- Inherit `Specification` (from `Cratis.Specifications`)
-- `void Establish()` — setup before the action
-- `void Because()` — **one** action under test (the thing being specified)
-- `[Fact] void should_*()` — one assertion per fact, no blank lines between them
-- All fields: `private` (or `protected` in `given/` contexts), `_camelCase`
-- All methods can be `async Task` when needed
-
----
-
-## Step 4 — Add a reusable context (`given/`)
-
-When multiple specs share the same setup, extract it into a `given/` class:
-
-```csharp
-// for_AuthorService/given/all_dependencies.cs
-namespace MyApp.Authors.for_AuthorService.given;
-
-public class all_dependencies : Specification
-{
-    protected IEventLog _eventLog;
-    protected ILogger<AuthorService> _logger;
-
-    void Establish()
-    {
-        _eventLog = Substitute.For<IEventLog>();
-        _logger = Substitute.For<ILogger<AuthorService>>();
-    }
-}
-```
-
-```csharp
-// for_AuthorService/given/an_author_service.cs
-namespace MyApp.Authors.for_AuthorService.given;
-
-public class an_author_service : all_dependencies
-{
-    protected AuthorService _service;
-
-    void Establish() => _service = new(_eventLog, _logger);
-}
-```
-
-```csharp
-// for_AuthorService/when_registering/and_name_is_valid.cs
-namespace MyApp.Authors.for_AuthorService.when_registering;
-
-public class and_name_is_valid : given.an_author_service
-{
-    void Because() => _service.Register(new AuthorName("John"));
-
-    [Fact] void should_append_event() =>
-        _eventLog.Received(1).Append(Arg.Any<AuthorId>(), Arg.Any<AuthorRegistered>());
-}
-```
-
-See `references/csharp-patterns.md` for full NSubstitute patterns and assertion methods.
-
----
-
-## Step 5 — Write a Chronicle integration spec
-
-Integration specs live directly inside the slice's `when_<behavior>/` folder and test the full stack against a real Chronicle event store.
-
-```csharp
-// Authors/Registration/when_registering/and_there_are_no_authors.cs
-using context = MyApp.Authors.Registration.when_registering.and_there_are_no_authors.context;
-
-namespace MyApp.Authors.Registration.when_registering;
-
-[Collection(ChronicleCollection.Name)]
-public class and_there_are_no_authors(context context) : Given<context>(context)
-{
-    public class context(ChronicleOutOfProcessFixture fixture) : given.an_http_client(fixture)
-    {
-        public CommandResult<AuthorId>? Result;
-
-        async Task Because() =>
-            Result = await Client.ExecuteCommand<RegisterAuthor, AuthorId>(
-                "/api/authors/register",
-                new RegisterAuthor(new AuthorName("John Doe")));
-    }
-
-    [Fact] void should_be_successful() => Context.Result!.IsSuccess.ShouldBeTrue();
-    [Fact] void should_have_appended_one_event() =>
-        Context.ShouldHaveTailSequenceNumber(EventSequenceNumber.First);
-    [Fact] void should_append_author_registered_event() =>
-        Context.ShouldHaveAppendedEvent<AuthorRegistered>(
-            EventSequenceNumber.First, Context.Result!.Response,
-            evt => evt.Name.Value.ShouldEqual("John Doe"));
-}
-```
-
-See `references/integration-specs.md` for the full integration spec guide.
-
----
-
-## What NOT to spec
-
-- Simple auto-properties (`public AuthorId Id { get; }`)
-- Properties returning constructor parameters
-- Simple delegation (`public IEnumerable<Author> All => _list;`)
-- Logging calls
-- Trivial null checks
-
----
-
-## Reference files
-
-- `references/csharp-patterns.md` — BDD pattern detail, NSubstitute, assertions, exception catching
-- `references/integration-specs.md` — Chronicle integration spec structure, helpers, Given<context>
diff --git a/.github/skills/cratis-specs-csharp/evals/evals.json b/.github/skills/cratis-specs-csharp/evals/evals.json
deleted file mode 100644
index 6bbbd80..0000000
--- a/.github/skills/cratis-specs-csharp/evals/evals.json
+++ /dev/null
@@ -1,37 +0,0 @@
-{
-  "skill_name": "cratis-specs-csharp",
-  "evals": [
-    {
-      "id": 1,
-      "prompt": "Write C# specs for a class called EmployeeService that has a Register(EmployeeName name) method. The method appends an EmployeeRegistered event via IEventLog. Write specs for two scenarios: (1) when the name is valid - it should append the event, (2) when the employee name already exists - it should throw EmployeeAlreadyRegistered. Follow Cratis BDD spec conventions.",
-      "expected_output": "Folder structure: for_EmployeeService/given/all_dependencies.cs + given/an_employee_service.cs + when_registering/and_name_is_valid.cs + when_registering/and_name_already_exists.cs. Uses Specification base class, Establish/Because/should_ pattern, NSubstitute for IEventLog, ShouldXxx assertions, Catch.Exception.",
-      "files": [],
-      "assertions": [
-        "Uses Specification base class (not xUnit test class directly)",
-        "States are separated: Establish (setup), Because (single action), [Fact] should_* (assertions)",
-        "Folder structure follows for_<Type>/when_<behavior>/and_<condition> naming",
-        "Uses NSubstitute: Substitute.For<IEventLog>()",
-        "Uses ShouldXxx assertion methods (not Assert.Equal or similar)",
-        "Exception test uses Catch.Exception pattern",
-        "Reusable context in given/ folder with protected fields",
-        "No [Fact] attribute on Establish or Because methods"
-      ]
-    },
-    {
-      "id": 2,
-      "prompt": "Write a Chronicle integration spec for a command RegisterAuthor(AuthorName name) at route /api/authors/register. Write two scenarios: one where registration succeeds (no prior authors), and one where it fails because the author name already exists.",
-      "expected_output": "Two spec files under Authors/Registration/when_registering/ using [Collection(ChronicleCollection.Name)], Given<context>, context inner class inheriting given.an_http_client(fixture), async Task Because() using Client.ExecuteCommand. Assertions use ShouldHaveTailSequenceNumber and ShouldHaveAppendedEvent.",
-      "files": [],
-      "assertions": [
-        "Uses [Collection(ChronicleCollection.Name)] attribute",
-        "context is an inner public class",
-        "Uses Given<context> as base class for outer spec class",
-        "Has 'using context = ...' alias at the top of each file",
-        "Because() is async Task and calls Client.ExecuteCommand",
-        "Uses ShouldHaveTailSequenceNumber for event count verification",
-        "Precondition spec uses async Task Establish() to seed events",
-        "Uses ShouldHaveAppendedEvent for verifying event contents"
-      ]
-    }
-  ]
-}
diff --git a/.github/skills/cratis-specs-csharp/references/csharp-patterns.md b/.github/skills/cratis-specs-csharp/references/csharp-patterns.md
deleted file mode 100644
index db9a574..0000000
--- a/.github/skills/cratis-specs-csharp/references/csharp-patterns.md
+++ /dev/null
@@ -1,166 +0,0 @@
-# C# Spec Patterns — Reference
-
-## BDD phases
-
-| Method | Purpose | Notes |
-| --- | --- | --- |
-| `void Establish()` | Setup — runs before `Because()` | Base-class `Establish` runs first, then derived class |
-| `void Because()` | The single action under test | Only in concrete spec files — never in `given/` contexts |
-| `[Fact] void should_*()` | One assertion per fact | No blank lines between `should_` methods |
-| `void Destroy()` | Teardown after each test | Optional |
-
-All phases can be `async Task`.
-
----
-
-## Minimal spec
-
-```csharp
-namespace MyApp.for_KeyHelper;
-
-public class when_combining_parts : Specification
-{
-    object[] _parts;
-    string _result;
-
-    void Establish() => _parts = ["First", "Second", "Third"];
-    void Because() => _result = KeyHelper.Combine(_parts);
-
-    [Fact] void should_combine_all_parts() => _result.ShouldEqual("First+Second+Third");
-    [Fact] void should_not_be_empty() => _result.ShouldNotBeEmpty();
-}
-```
-
----
-
-## Reusable context (layered given)
-
-```csharp
-// given/all_dependencies.cs — mock all external deps
-public class all_dependencies : Specification
-{
-    protected IEventStore _eventStore;
-    protected IReactorInvoker _reactorInvoker;
-
-    void Establish()
-    {
-        _eventStore = Substitute.For<IEventStore>();
-        _reactorInvoker = Substitute.For<IReactorInvoker>();
-    }
-}
-
-// given/a_reactor_handler.cs — build system under test
-public class a_reactor_handler : all_dependencies
-{
-    protected ReactorHandler _handler;
-
-    void Establish() => _handler = new(_eventStore, _reactorInvoker);
-}
-
-// when_handling/and_event_is_received.cs — concrete spec
-public class and_event_is_received : given.a_reactor_handler
-{
-    void Because() => _handler.Handle(new SomeEvent());
-
-    [Fact] void should_invoke_reactor() =>
-        _reactorInvoker.Received(1).Invoke(Arg.Any<SomeEvent>());
-}
-```
-
----
-
-## NSubstitute patterns
-
-```csharp
-// Create substitutes
-_service = Substitute.For<IMyService>();
-
-// Return values
-_service.GetValue(Arg.Any<string>()).Returns("result");
-_service.GetAsync(Arg.Any<int>()).Returns(Task.FromResult(42));
-
-// Argument matchers
-Arg.Is<Request>(r => r.Id == expectedId && r.Name == expectedName)
-
-// Verify calls
-_service.Received(1).DoSomething(Arg.Any<string>());
-_service.DidNotReceive().DoSomethingElse();
-
-// Capture arguments
-_service.When(s => s.Process(Arg.Any<IDictionary<string, string>>()))
-    .Do(info => _captured = info.Arg<IDictionary<string, string>>());
-
-// Throw from substitute
-_handler.Handle(Arg.Any<CommandContext>()).Throws(new MyException("fail"));
-```
-
----
-
-## Assertion extension methods (Cratis.Specifications)
-
-| Method | Example |
-| --- | --- |
-| `.ShouldEqual(expected)` | `_result.ShouldEqual(42)` |
-| `.ShouldBeTrue()` | `_flag.ShouldBeTrue()` |
-| `.ShouldBeFalse()` | `_flag.ShouldBeFalse()` |
-| `.ShouldBeNull()` | `_error.ShouldBeNull()` |
-| `.ShouldNotBeNull()` | `_value.ShouldNotBeNull()` |
-| `.ShouldBeEmpty()` | `_list.ShouldBeEmpty()` |
-| `.ShouldNotBeEmpty()` | `_list.ShouldNotBeEmpty()` |
-| `.ShouldContain(item)` | `_list.ShouldContain(expected)` |
-| `.ShouldNotContain(item)` | `_list.ShouldNotContain(excluded)` |
-| `.ShouldContainOnly(items)` | `_list.ShouldContainOnly(expectedItems)` |
-| `.ShouldBeOfExactType<T>()` | `_obj.ShouldBeOfExactType<PropertiesChanged>()` |
-| `.ShouldBeGreaterThan(n)` | `_count.ShouldBeGreaterThan(0)` |
-| `.ShouldBeLessThan(n)` | `_count.ShouldBeLessThan(100)` |
-
----
-
-## Catching exceptions
-
-```csharp
-Exception? _error;
-
-async Task Because() => _error = await Catch.Exception(_sut.DoSomethingThatThrows);
-
-[Fact] void should_throw() => _error.ShouldNotBeNull();
-[Fact] void should_not_throw() => _error.ShouldBeNull();
-[Fact] void should_throw_author_not_found() => _error.ShouldBeOfExactType<AuthorNotFound>();
-```
-
----
-
-## Using statements
-
-Common usings are provided globally in `GlobalUsings.Specs.cs` — do **not** add them manually:
-- `Xunit`
-- `NSubstitute`
-- `Cratis.Specifications`
-
-Do **not** add a `using` for the namespace of the system under test.
-
----
-
-## Multiple outcomes → folder
-
-```
-// Single outcome → single file
-for_MyService/when_processing.cs
-
-// Multiple outcomes → folder + files
-for_MyService/when_processing/
-    and_input_is_valid.cs
-    and_input_is_null.cs
-    with_empty_collection.cs
-    without_required_field.cs
-```
-
-Allowed file name prefixes: `and_*`, `with_*`, `without_*`, `having_*`, `given_*`
-
----
-
-## Folder naming read as sentences
-
-`for_AuthorService / when_registering / and_name_already_exists`
-
-→ "For AuthorService, when registering, and name already exists, it should..."
diff --git a/.github/skills/cratis-specs-csharp/references/integration-specs.md b/.github/skills/cratis-specs-csharp/references/integration-specs.md
deleted file mode 100644
index 0f6ce1a..0000000
--- a/.github/skills/cratis-specs-csharp/references/integration-specs.md
+++ /dev/null
@@ -1,101 +0,0 @@
-# Chronicle Integration Specs — Reference
-
-Integration specs test a complete vertical slice end-to-end — from HTTP request through command handling, event appending, constraint checking, and projection — against a real Chronicle event store. If one passes, the entire stack works.
-
-They live under `when_<behavior>/` **inside the slice folder** (not in a `for_<Type>/` unit folder — there's no isolated unit, the entire slice is under test).
-
----
-
-## Structure
-
-```csharp
-// Authors/Registration/when_registering/and_there_are_no_authors.cs
-
-using context = MyApp.Authors.Registration.when_registering.and_there_are_no_authors.context;
-
-namespace MyApp.Authors.Registration.when_registering;
-
-[Collection(ChronicleCollection.Name)]
-public class and_there_are_no_authors(context context) : Given<context>(context)
-{
-    public class context(ChronicleOutOfProcessFixture fixture) : given.an_http_client(fixture)
-    {
-        public CommandResult<AuthorId>? Result;
-
-        async Task Because() =>
-            Result = await Client.ExecuteCommand<RegisterAuthor, AuthorId>(
-                "/api/authors/register",
-                new RegisterAuthor(new AuthorName("John Doe")));
-    }
-
-    [Fact] void should_be_successful() => Context.Result!.IsSuccess.ShouldBeTrue();
-    [Fact] void should_have_appended_one_event() =>
-        Context.ShouldHaveTailSequenceNumber(EventSequenceNumber.First);
-    [Fact] void should_append_author_registered_event() =>
-        Context.ShouldHaveAppendedEvent<AuthorRegistered>(
-            EventSequenceNumber.First, Context.Result!.Response,
-            evt => evt.Name.Value.ShouldEqual("John Doe"));
-}
-```
-
----
-
-## Spec with preconditions (seed the event store)
-
-Use `async Task Establish()` to append events before `Because()`:
-
-```csharp
-public class context(ChronicleOutOfProcessFixture fixture) : given.an_http_client(fixture)
-{
-    public const string ExistingName = "John Doe";
-    public CommandResult<object>? Result;
-
-    async Task Establish() =>
-        await EventStore.EventLog.Append(AuthorId.New(), new AuthorRegistered(ExistingName));
-
-    async Task Because() =>
-        Result = await Client.ExecuteCommand<RegisterAuthor>(
-            "/api/authors/register",
-            new RegisterAuthor(ExistingName));
-}
-
-[Fact] void should_not_be_successful() => Context.Result!.IsSuccess.ShouldBeFalse();
-[Fact] void should_not_have_appended_additional_events() =>
-    Context.ShouldHaveTailSequenceNumber(EventSequenceNumber.First);
-```
-
----
-
-## ExecuteCommand overloads
-
-```csharp
-// Command with no typed response (returns CommandResult<object>)
-Result = await Client.ExecuteCommand<RegisterAuthor>(url, command);
-
-// Command with typed response (returns CommandResult<TResult>)
-Result = await Client.ExecuteCommand<RegisterAuthor, AuthorId>(url, command);
-```
-
----
-
-## Integration assertion helpers
-
-| Helper | What it verifies |
-| --- | --- |
-| `Context.Result!.IsSuccess.ShouldBeTrue()` | Command succeeded |
-| `Context.Result!.IsSuccess.ShouldBeFalse()` | Command failed (validation, constraint, etc.) |
-| `Context.ShouldHaveTailSequenceNumber(EventSequenceNumber.First)` | Event log has exactly one event (sequence 0) |
-| `Context.ShouldHaveTailSequenceNumber(n)` | Event log tail is at sequence `n` |
-| `Context.ShouldHaveAppendedEvent<TEvent>(seq, eventSourceId, validator)` | Specific event was appended at sequence with correct values |
-
----
-
-## Key rules
-
-- `context` is an **inner public class** inheriting from `given.an_http_client(fixture)`
-- Always add `using context = <full.namespace>.context;` alias at the top
-- `[Collection(ChronicleCollection.Name)]` on the outer class — required for test isolation
-- `Establish` seeds preconditions; `Because` executes the command under test
-- Declare `Result` as nullable, initialized to `null!` if needed for nullable analysis
-- The outer class constructor receives `context` via xUnit constructor injection
-- Never mix unit specs and integration specs in the same folder
diff --git a/.github/skills/cratis-specs-typescript/SKILL.md b/.github/skills/cratis-specs-typescript/SKILL.md
deleted file mode 100644
index f153bb1..0000000
--- a/.github/skills/cratis-specs-typescript/SKILL.md
+++ /dev/null
@@ -1,135 +0,0 @@
----
-name: cratis-specs-typescript
-description: Step-by-step guidance for writing TypeScript specs in Cratis using BDD-style Specification by Example — the given()/describe/it pattern, for_/when_/ folder hierarchy, reusable context classes, Sinon mocking, and Chai assertions. Use whenever writing TypeScript specs or tests, creating spec files/folders, using the given() helper, mocking with sinon.createStubInstance or sinon.stub, asserting with Chai .should, or understanding how yarn test runs specs.
----
-
-## Core philosophy
-
-Same BDD philosophy as C# specs — specs describe behaviors, not implementations. The `given()` helper + context class mirrors the C# `Specification` base class: setup is separated from the action, each `it()` verifies a single outcome.
-
----
-
-## Step 1 — Create the folder structure
-
-```
-for_<ClassName>/
-├── given/
-│   └── a_<system>.ts             ← reusable context class
-├── when_<behavior>/              ← behavior with multiple outcomes
-│   ├── with_<condition>.ts
-│   ├── without_<condition>.ts
-│   └── and_<condition>.ts
-└── when_<simple_behavior>.ts     ← single outcome = single file
-```
-
-Example:
-```
-for_EventsCommandResponseValueHandler/
-├── given/
-│   └── an_events_command_response_value_handler.ts
-├── when_checking_can_handle/
-│   ├── with_valid_events_collection.ts
-│   ├── with_null_value.ts
-│   └── without_event_source_id.ts
-└── when_handling/
-    ├── empty_events_collection.ts
-    └── multiple_events_collection.ts
-```
-
----
-
-## Step 2 — Write a reusable context class
-
-```ts
-// for_AuthorService/given/an_author_service.ts
-import sinon from 'sinon';
-import { AuthorService } from '../../../AuthorService';
-
-export class an_author_service {
-    eventLog: sinon.StubbedInstance<IEventLog>;
-    service: AuthorService;
-
-    constructor() {
-        this.eventLog = sinon.createStubInstance(EventLog);
-        this.service = new AuthorService(this.eventLog);
-    }
-}
-```
-
-Properties are **public** (unlike C# protected fields) — tests access them via `context.propertyName`.
-
----
-
-## Step 3 — Write a spec using `given()`
-
-```ts
-// for_AuthorService/when_registering/with_valid_name.ts
-import { an_author_service } from '../given/an_author_service';
-import { given } from '../../given';   // import from package root
-
-describe('when registering with valid name', given(an_author_service, context => {
-    beforeEach(async () => {
-        await context.service.register('John Doe');
-    });
-
-    it('should append an event', () => {
-        context.eventLog.append.calledOnce.should.be.true;
-    });
-
-    it('should pass the author name', () => {
-        const call = context.eventLog.append.firstCall;
-        call.args[1].name.should.equal('John Doe');
-    });
-}));
-```
-
----
-
-## Step 4 — Simple spec (no shared context)
-
-For behaviors without shared setup:
-
-```ts
-describe('when replacing route parameters', () => {
-    let result: { route: string; unusedParameters: object };
-
-    beforeEach(() => {
-        result = UrlHelpers.replaceRouteParameters('/api/items/{id}', { id: '123' });
-    });
-
-    it('should replace the route parameter', () => {
-        result.route.should.equal('/api/items/123');
-    });
-
-    it('should remove used parameters', () => {
-        Object.keys(result.unusedParameters).should.have.lengthOf(0);
-    });
-});
-```
-
----
-
-## Naming conventions
-
-| Element | Convention | Example |
-| --- | --- | --- |
-| `describe()` text | Natural language sentence | `'when registering with valid name'` |
-| `it()` text | Starts with "should", uses **spaces** | `'should append an event'` |
-| Context class | `a_` or `an_` prefix | `an_author_service` |
-| Spec file | Descriptive, `with_` / `without_` / `and_` | `with_valid_name.ts` |
-
-**Always use spaces in `it()` descriptions** — never underscores.
-
----
-
-## Running specs
-
-```bash
-yarn test   # from the package root
-```
-
----
-
-## Reference files
-
-- `references/typescript-patterns.md` — Chai assertions, Sinon patterns, async specs, full examples
diff --git a/.github/skills/cratis-specs-typescript/evals/evals.json b/.github/skills/cratis-specs-typescript/evals/evals.json
deleted file mode 100644
index 1ca8aee..0000000
--- a/.github/skills/cratis-specs-typescript/evals/evals.json
+++ /dev/null
@@ -1,35 +0,0 @@
-{
-  "skill_name": "cratis-specs-typescript",
-  "evals": [
-    {
-      "id": 1,
-      "prompt": "Write TypeScript specs for a function formatCurrency(amount: number, currency: string): string. Test two behaviors: (1) when formatting a positive amount — it should include the currency symbol and two decimal places; (2) when formatting a negative amount — it should include a minus sign. Follow Cratis TypeScript spec conventions.",
-      "expected_output": "Folder structure: for_formatCurrency/when_formatting_positive_amount.ts + when_formatting_negative_amount.ts. Uses describe()/it() with Vitest/Mocha. Uses Chai .should fluent assertions (NOT expect()). it() descriptions use spaces not underscores.",
-      "files": [],
-      "assertions": [
-        "Uses Chai .should fluent interface (value.should.equal(), NOT expect(value).to.equal())",
-        "it() descriptions use spaces (not underscores)",
-        "it() descriptions start with 'should'",
-        "Folder structure follows for_<function>/when_<behavior>.ts naming",
-        "Uses describe() for the outer scenario block",
-        "Uses beforeEach() for setup",
-        "Does NOT use expect() from Chai"
-      ]
-    },
-    {
-      "id": 2,
-      "prompt": "Write TypeScript specs for a class CommandHandler that has a method handle(command: Command): Promise<CommandResult>. It depends on an ICommandExecutor interface. Spec two scenarios: (1) when the executor succeeds — result.isSuccess should be true; (2) when the executor throws — result.isSuccess should be false. Use the given() helper with a reusable context class.",
-      "expected_output": "for_CommandHandler/given/a_command_handler.ts context class using sinon.createStubInstance + for_CommandHandler/when_handling/with_successful_execution.ts and with_failed_execution.ts using given() helper. Chai .should assertions.",
-      "files": [],
-      "assertions": [
-        "Reusable context class defined in given/ folder",
-        "Context class uses sinon.createStubInstance for mocking",
-        "Context class properties are public (not protected)",
-        "Specs use given() helper from package root",
-        "Uses Chai .should assertions",
-        "Folder names follow for_<Class>/when_<behavior>/with_<condition> pattern",
-        "it() descriptions start with 'should' and use spaces"
-      ]
-    }
-  ]
-}
diff --git a/.github/skills/cratis-specs-typescript/references/typescript-patterns.md b/.github/skills/cratis-specs-typescript/references/typescript-patterns.md
deleted file mode 100644
index 74f0aae..0000000
--- a/.github/skills/cratis-specs-typescript/references/typescript-patterns.md
+++ /dev/null
@@ -1,168 +0,0 @@
-# TypeScript Spec Patterns — Reference
-
-## Frameworks
-
-| Framework | Role |
-| --- | --- |
-| [Vitest](https://vitest.dev/) | Test runner |
-| [Mocha](https://mochajs.org/) | Test structure (`describe`, `it`, `beforeEach`) |
-| [Chai](https://www.chaijs.com/) | Assertions — always `.should` fluent interface |
-| [SinonJS](https://sinonjs.org/) | Mocking and stubbing |
-
----
-
-## Chai assertions — always use `.should`
-
-Never use `expect()`. The `.should` style reads as a natural English sentence.
-
-```ts
-// Equality
-value.should.equal(expected);
-value.should.deep.equal({ id: 1, name: 'John' });
-
-// Booleans
-flag.should.be.true;
-flag.should.be.false;
-
-// Null / undefined
-value.should.be.null;
-value.should.not.be.null;
-value.should.be.undefined;
-value.should.not.be.undefined;
-
-// Arrays
-array.should.contain(item);
-array.should.have.lengthOf(3);
-array.should.be.empty;
-array.should.not.be.empty;
-
-// Types
-value.should.be.instanceOf(MyClass);
-
-// Throwing
-(() => throwingFn()).should.throw(ErrorType);
-```
-
----
-
-## Sinon mocking
-
-```ts
-import sinon from 'sinon';
-
-// Stub an entire class (all methods become stubs)
-const service = sinon.createStubInstance(ConcreteService);
-
-// Stub a global function
-const fetchStub = sinon.stub(globalThis, 'fetch');
-fetchStub.resolves({ ok: true, json: async () => ({ data: 'value' }) });
-
-// Configure return values
-service.getValue.returns('result');
-service.getAsync.resolves(42);
-
-// Access call details
-service.doSomething.calledOnce.should.be.true;
-service.doSomething.calledWith('expected-arg').should.be.true;
-service.doSomething.callCount.should.equal(2);
-
-const firstCall = service.doSomething.firstCall;
-firstCall.args[0].should.equal('expected');
-
-// Restore stubs after test
-afterEach(() => sinon.restore());
-```
-
----
-
-## given() helper — full pattern
-
-The `given()` function instantiates a context class, runs tests with it, and ensures setup is isolated per test.
-
-```ts
-import { given } from '../../given';  // import from package root
-import { a_my_service } from '../given/a_my_service';
-
-describe('when doing something', given(a_my_service, context => {
-    let result: string;
-
-    beforeEach(async () => {
-        result = await context.service.doSomething('input');
-    });
-
-    it('should return expected result', () => {
-        result.should.equal('expected');
-    });
-
-    it('should call dependency once', () => {
-        context.dependency.process.calledOnce.should.be.true;
-    });
-}));
-```
-
----
-
-## Reusable context class
-
-```ts
-// given/a_my_service.ts
-import sinon from 'sinon';
-import { MyService } from '../../../MyService';
-
-export class a_my_service {
-    dependency: sinon.StubbedInstance<DependencyClass>;
-    service: MyService;
-
-    constructor() {
-        this.dependency = sinon.createStubInstance(DependencyClass);
-        // configure defaults:
-        this.dependency.getValue.returns('default');
-        this.service = new MyService(this.dependency as unknown as IDependency);
-    }
-}
-```
-
-Properties are **public** — accessed via `context.propertyName` in specs.
-
----
-
-## Multiple outcomes — folder pattern
-
-```
-when_processing/
-├── with_valid_input.ts     → happy path
-├── with_empty_input.ts     → edge case
-└── without_required_field.ts  → failure path
-```
-
-Each file has its own `describe()` block, its own `beforeEach`, and its own `it()` assertions.
-
----
-
-## Async specs
-
-`beforeEach`, `afterEach`, and `it` can all be `async`:
-
-```ts
-describe('when loading data', given(a_loader, context => {
-    let result: Data[];
-
-    beforeEach(async () => {
-        result = await context.loader.load('source');
-    });
-
-    it('should return items', () => {
-        result.should.have.lengthOf(3);
-    });
-}));
-```
-
----
-
-## What NOT to spec
-
-Same as C#:
-- Simple property getters and setters
-- Properties that return constructor parameters directly
-- Trivial delegation
-- Don't write specs whose `describe` starts with "when getting" or "when returning" — these are almost always testing getters, not behavior
diff --git a/.github/skills/cratis-vertical-slice/SKILL.md b/.github/skills/cratis-vertical-slice/SKILL.md
deleted file mode 100644
index 8ca0fe1..0000000
--- a/.github/skills/cratis-vertical-slice/SKILL.md
+++ /dev/null
@@ -1,206 +0,0 @@
----
-name: cratis-vertical-slice
-description: Explains how vertical feature slices are structured in a Cratis Chronicle + Arc application — folder layout, what goes in the single backend .cs file, the four slice types (State Change/View/Automation/Translation), and how features compose slices. Use when asking how vertical slices work, where to put files, how to organize a feature folder, which slice type to choose, or how the workflow from C# to TypeScript proxies to React looks. For actively building a new slice end-to-end right now, use new-vertical-slice instead.
----
-
-## Core principle
-
-A vertical slice contains **everything for a single behavior**: the command or query, the events it produces, the projections that build read models, the React component, and the specs. Everything lives together because everything changes together.
-
-One feature folder → many slices.   
-One slice folder → one `.cs` file (all backend) + one `.tsx` file (frontend).
-
----
-
-## Step 1 — Identify the feature and slice type
-
-First, name the feature (a domain noun, pluralized) and identify the slice type:
-
-| Slice type | What it does | Key artifacts |
-| --- | --- | --- |
-| **State Change** | Mutates system state | Command + events + validators/constraints |
-| **State View** | Projects events into queryable data | Read model + projection + queries |
-| **Automation** | Reacts to events, makes decisions | Reactor + optional local read models |
-| **Translation** | Adapts events between slices | Reactor → triggers commands in own slice |
-
----
-
-## Step 2 — Create the folder structure
-
-```
-Features/
-└── <Feature>/                      ← feature root (pluralized domain noun)
-    ├── <Feature>.tsx               ← composition page
-    ├── <ConceptName>.cs            ← shared ConceptAs<T> types for the feature
-    └── <SliceName>/                ← slice (action or view name)
-        ├── <SliceName>.cs          ← ALL backend artifacts in ONE file
-        ├── <Component>.tsx         ← React component
-        └── when_<behavior>/        ← integration specs (state-change slices)
-            └── and_<scenario>.cs
-```
-
-✅ Correct:
-```
-Features/Authors/
-├── Authors.tsx
-├── AuthorId.cs
-├── AuthorName.cs
-├── Registration/
-│   ├── Registration.cs     ← command + event + constraint + validator
-│   ├── AddAuthor.tsx
-│   └── when_registering/
-│       └── and_there_are_no_authors.cs
-└── Listing/
-    ├── Listing.cs          ← read model + projection + query
-    └── Listing.tsx
-```
-
-❌ Wrong — never split by artifact type:
-```
-Features/Authors/
-├── Commands/RegisterAuthor.cs
-├── Handlers/RegisterAuthorHandler.cs
-├── Events/AuthorRegistered.cs
-```
-
-**Namespace rule**: Drop the `.Features.` segment.  
-`MyApp.Authors.Registration` — not `MyApp.Features.Authors.Registration`.
-
----
-
-## Step 3 — Write the backend slice file
-
-All backend artifacts for one slice go in a single `<SliceName>.cs` file. File header:
-
-```csharp
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-```
-
-For a **State Change** slice, the file contains:
-```csharp
-[EventType]
-public record AuthorRegistered(AuthorName Name);
-
-public class UniqueAuthorName : IConstraint { ... }
-
-public class RegisterAuthorValidator : CommandValidator<RegisterAuthor> { ... }
-
-[Command]
-public record RegisterAuthor(AuthorName Name)
-{
-    public (AuthorId, AuthorRegistered) Handle()
-    {
-        var authorId = AuthorId.New();
-        return (authorId, new(Name));
-    }
-}
-```
-
-For a **State View** slice, the file contains:
-```csharp
-[ReadModel]
-[FromEvent<Registration.AuthorRegistered>]
-public record Author(
-    [Key] AuthorId Id,
-    AuthorName Name)
-{
-    public static ISubject<IEnumerable<Author>> AllAuthors(IMongoCollection<Author> collection) =>
-        collection.Observe();
-}
-```
-
-See `references/slice-anatomy.md` for all artifact patterns.
-
----
-
-## Step 4 — Define domain concepts
-
-For every identity or domain value, create a `ConceptAs<T>` type — one file per concept, in the feature folder (or `Features/` root if shared across features).
-
-```csharp
-public record AuthorId(Guid Value) : ConceptAs<Guid>(Value)
-{
-    public static readonly AuthorId NotSet = new(Guid.Empty);
-    public static implicit operator Guid(AuthorId id) => id.Value;
-    public static implicit operator AuthorId(Guid value) => new(value);
-    public static implicit operator EventSourceId(AuthorId id) => new(id.Value.ToString());
-    public static AuthorId New() => new(Guid.NewGuid());
-}
-```
-
-See `references/concepts.md` for all concept patterns.
-
----
-
-## Step 5 — Build to generate TypeScript proxies
-
-```bash
-dotnet build
-```
-
-This generates `.ts` proxy files in the configured `<CratisProxiesOutputPath>`. The frontend cannot be written until this succeeds — the proxies are the contract.
-
----
-
-## Step 6 — Write the React component
-
-```tsx
-// Listing.tsx
-import { AllAuthors } from '../proxies/Listing';   // auto-generated
-
-export const Listing = () => {
-    const [result] = AllAuthors.use();
-    return (
-        <DataTable value={result.data} ...>
-            <Column field="name" header="Name" />
-        </DataTable>
-    );
-};
-```
-
----
-
-## Step 7 — Compose the feature page
-
-The feature's `<Feature>.tsx` assembles slices into a page:
-
-```tsx
-// Authors.tsx
-import { AddAuthor } from './Registration/AddAuthor';
-import { Listing } from './Listing/Listing';
-import { useDialog } from '@cratis/arc.react/dialogs';
-
-export const Authors = () => {
-    const [AddAuthorDialog, showAddAuthorDialog] = useDialog(AddAuthor);
-    const menuItems = [{ label: 'Add Author', command: () => showAddAuthorDialog() }];
-    return (
-        <Page title="Authors">
-            <Menubar model={menuItems} />
-            <Listing />
-            <AddAuthorDialog />
-        </Page>
-    );
-};
-```
-
----
-
-## Development workflow order
-
-Work in this exact sequence — TypeScript proxies are generated from C# during `dotnet build`:
-
-1. Implement the C# slice file (step 3)
-2. Write integration specs for state-change slices
-3. `dotnet build` — generates TypeScript proxies (step 5)
-4. Implement React component(s) (step 6)
-5. Register in the feature composition page (step 7)
-6. Add/update routes if needed
-
----
-
-## Reference files
-
-- `references/slice-anatomy.md` — complete patterns for every artifact type
-- `references/slice-types.md` — when to use each slice type with decision guide
-- `references/concepts.md` — ConceptAs<T> patterns for all primitive backing types
diff --git a/.github/skills/cratis-vertical-slice/evals/evals.json b/.github/skills/cratis-vertical-slice/evals/evals.json
deleted file mode 100644
index aadc3de..0000000
--- a/.github/skills/cratis-vertical-slice/evals/evals.json
+++ /dev/null
@@ -1,34 +0,0 @@
-{
-  "skill_name": "cratis-vertical-slice",
-  "evals": [
-    {
-      "id": 1,
-      "prompt": "I'm building an invoicing feature for my app (namespace: MyBillingApp). The feature needs: (1) creating an invoice, (2) listing all invoices, (3) marking an invoice as paid. Show me the complete folder structure and the content of each .cs and .tsx file, following vertical slice architecture.",
-      "expected_output": "Features/Invoices/ folder with: Invoices.tsx composition page, InvoiceId.cs and InvoiceNumber.cs concepts, and three slice subdirs: Creation/Creation.cs, Listing/Listing.cs+Listing.tsx, MarkingAsPaid/MarkingAsPaid.cs. All .cs files have all backend artifacts in one file. No Commands/ Events/ Handlers/ folders.",
-      "files": [],
-      "assertions": [
-        "Output shows a Features/Invoices/ root folder (pluralized feature name)",
-        "Output contains concept files at the feature root (InvoiceId.cs etc.)",
-        "Each slice has a subfolder with a single .cs file containing ALL backend artifacts",
-        "No Commands/, Events/, Handlers/, or Projections/ subfolders within the feature",
-        "Namespace does NOT include '.Features.' segment (e.g. MyBillingApp.Invoices.Creation)",
-        "All .cs files start with the Cratis copyright header",
-        "State change slices contain command + event in the same .cs file",
-        "State view slice contains read model + projection + query in the same .cs file"
-      ]
-    },
-    {
-      "id": 2,
-      "prompt": "Explain what slice type I should use and how to structure the code for this scenario: when a new order is placed (OrderPlaced event), I need to automatically send a confirmation email via our IEmailService. This is separate from the order creation command itself.",
-      "expected_output": "Identifies this as an Automation slice (IReactor). Shows IReactor class dispatching on OrderPlaced event, using IEmailService via constructor injection. Places it inside the Orders feature folder in its own slice subfolder.",
-      "files": [],
-      "assertions": [
-        "Correctly identifies the slice type as Automation (IReactor), not Translation",
-        "Output shows IReactor implementation with method dispatching on the event type",
-        "IEmailService injected via constructor (not service locator)",
-        "Slice is placed inside the feature folder (e.g. Orders/EmailConfirmation/)",
-        "All artifacts in a single .cs file"
-      ]
-    }
-  ]
-}
diff --git a/.github/skills/cratis-vertical-slice/references/concepts.md b/.github/skills/cratis-vertical-slice/references/concepts.md
deleted file mode 100644
index 0dcd399..0000000
--- a/.github/skills/cratis-vertical-slice/references/concepts.md
+++ /dev/null
@@ -1,102 +0,0 @@
-# Concepts — Reference
-
-## What is a Concept?
-
-A `ConceptAs<T>` wraps a primitive (`Guid`, `string`, `int`, etc.) in a named domain type. The compiler enforces that you cannot pass a `UserId` where an `AuthorId` was expected — both are `Guid` underneath, but they are distinct types.
-
-**Never use raw primitives in domain models, commands, events, or queries.**
-
----
-
-## Full canonical pattern — Guid identity
-
-```csharp
-public record AuthorId(Guid Value) : ConceptAs<Guid>(Value)
-{
-    public static readonly AuthorId NotSet = new(Guid.Empty);
-
-    public static implicit operator Guid(AuthorId id) => id.Value;
-    public static implicit operator AuthorId(Guid value) => new(value);
-    public static implicit operator EventSourceId(AuthorId id) => new(id.Value.ToString());
-
-    public static AuthorId New() => new(Guid.NewGuid());
-}
-```
-
-Include the `EventSourceId` implicit conversion on every **identity** concept that is used as a Chronicle event source key.
-
----
-
-## String value concept
-
-```csharp
-public record AuthorName(string Value) : ConceptAs<string>(Value)
-{
-    public static readonly AuthorName NotSet = new(string.Empty);
-
-    public static implicit operator string(AuthorName name) => name.Value;
-    public static implicit operator AuthorName(string value) => new(value);
-}
-```
-
----
-
-## Integer value concept
-
-```csharp
-public record PageNumber(int Value) : ConceptAs<int>(Value)
-{
-    public static readonly PageNumber NotSet = new(0);
-
-    public static implicit operator int(PageNumber p) => p.Value;
-    public static implicit operator PageNumber(int value) => new(value);
-}
-```
-
----
-
-## Rules
-
-| Rule | Detail |
-| --- | --- |
-| Inherit as `record` | Gives value equality and immutability for free |
-| `ConceptAs<T>` provides `T → Concept` implicitly | You only need to add the `Concept → T` direction |
-| Always add `NotSet` sentinel | Use `Guid.Empty`, `string.Empty`, or `0` — no `null` |
-| Add `New()` on Guid identity types | Reads better than `new AuthorId(Guid.NewGuid())` |
-| Add `EventSourceId` conversion on identity keys | Enables Chronicle to auto-resolve the event source |
-| One concept per file | Named after the concept, e.g. `AuthorId.cs` |
-
----
-
-## File placement
-
-| Scope | Location |
-| --- | --- |
-| Used only within one slice | Inside the slice folder |
-| Shared between slices of a feature | Feature root folder (`Features/Authors/AuthorId.cs`) |
-| Shared between features | `Features/` root (`Features/TenantId.cs`) |
-
-Never create a standalone `Concepts/` folder — concepts belong near the code that uses them.
-
----
-
-## In commands and events
-
-```csharp
-// Event uses concepts
-[EventType]
-public record AuthorRegistered(AuthorName Name);
-
-// Command uses concepts
-[Command]
-public record RegisterAuthor(AuthorName Name)
-{
-    public (AuthorId, AuthorRegistered) Handle() =>
-        (AuthorId.New(), new(Name));
-}
-
-// Read model uses concepts
-[ReadModel]
-[FromEvent<AuthorRegistered>]
-public record Author([Key] AuthorId Id, AuthorName Name);
-```
diff --git a/.github/skills/cratis-vertical-slice/references/slice-anatomy.md b/.github/skills/cratis-vertical-slice/references/slice-anatomy.md
deleted file mode 100644
index dc0f647..0000000
--- a/.github/skills/cratis-vertical-slice/references/slice-anatomy.md
+++ /dev/null
@@ -1,255 +0,0 @@
-# Slice Anatomy — Reference
-
-All backend artifacts for a slice go in a single `<SliceName>.cs` file. Below are complete patterns for every artifact type.
-
----
-
-## File header
-
-Every `.cs` file starts with:
-
-```csharp
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-```
-
-File-scoped namespace (no extra indentation):
-
-```csharp
-namespace MyApp.Authors.Registration;
-```
-
----
-
-## Events
-
-```csharp
-[EventType]
-public record AuthorRegistered(AuthorName Name);
-```
-
-Rules:
-- `[EventType]` takes **no arguments** — the type name is the identifier
-- Past tense: `ItemAddedToCart`, `UserRegistered`, `AddressChangedForPerson`
-- No nullable properties — ambiguous events need a second event
-- One purpose per event
-
----
-
-## Commands (model-bound)
-
-```csharp
-[Command]
-public record RegisterAuthor(AuthorName Name)
-{
-    public (AuthorId, AuthorRegistered) Handle()
-    {
-        var authorId = AuthorId.New();
-        return (authorId, new(Name));
-    }
-}
-```
-
-`Handle()` return types:
-
-| Return | When |
-| --- | --- |
-| `TEvent Handle()` | Single event, no client result needed |
-| `(TResult, TEvent) Handle()` | Return a value to the client + append one event |
-| `Result<TSuccess, TError> Handle()` | Business rule success/failure path |
-| `void Handle()` | Side-effect only, no event |
-
-Event source resolution (in priority order):
-1. Parameter marked with `[Key]`
-2. Parameter whose type has implicit conversion to `EventSourceId`
-3. Implement `ICanProvideEventSourceId`
-
-DI in `Handle()`: extra parameters beyond command + read model are resolved from DI automatically.
-
----
-
-## Business rules via DCB (Dynamic Consistency Boundary)
-
-Accept a read model parameter in `Handle()` — the framework injects the current projected state:
-
-```csharp
-[Command]
-public record ReserveBook(ISBN Isbn, MemberId MemberId)
-{
-    public Result<BookReserved, ValidationResult> Handle(Book book)
-    {
-        if (book.Available <= 0)
-            return ValidationResult.Error($"No available copies for {Isbn}");
-        return new BookReserved(Isbn, MemberId);
-    }
-}
-```
-
----
-
-## Validators
-
-```csharp
-public class RegisterAuthorValidator : CommandValidator<RegisterAuthor>
-{
-    public RegisterAuthorValidator()
-    {
-        RuleFor(c => c.Name).NotEmpty().WithMessage("Author name is required");
-    }
-}
-```
-
-With DI dependencies:
-
-```csharp
-public class MyValidator : CommandValidator<MyCommand>
-{
-    public MyValidator(IMyService service)
-    {
-        RuleFor(x => x).MustAsync(async (cmd, ct) => await service.IsValid(cmd));
-    }
-}
-```
-
----
-
-## Constraints
-
-Unique property across events:
-
-```csharp
-public class UniqueAuthorName : IConstraint
-{
-    public void Define(IConstraintBuilder builder) => builder
-        .Unique(_ => _
-            .On<AuthorRegistered>(e => e.Name)
-            .On<AuthorNameChanged>(e => e.Name)
-            .RemovedWith<AuthorRemoved>()
-            .WithMessage("Author name must be unique"));
-}
-```
-
-Unique event type per event source (one-per-stream):
-
-```csharp
-public class UniqueUser : IConstraint
-{
-    public void Define(IConstraintBuilder builder) =>
-        builder.Unique<UserRegistered>();
-}
-```
-
----
-
-## Read models (model-bound, preferred)
-
-```csharp
-[ReadModel]
-[FromEvent<Registration.AuthorRegistered>]
-public record Author(
-    [Key] AuthorId Id,
-    AuthorName Name)
-{
-    public static ISubject<IEnumerable<Author>> AllAuthors(IMongoCollection<Author> collection) =>
-        collection.Observe();
-
-    public static Author? ById(IMongoCollection<Author> collection, AuthorId id) =>
-        collection.Find(a => a.Id == id).FirstOrDefault();
-}
-```
-
-Model-bound projection attributes:
-
-| Attribute | Purpose |
-| --- | --- |
-| `[Key]` | Read model primary key |
-| `[FromEvent<T>]` | Auto-map from event (class-level) |
-| `[SetFrom<T>]` | Explicit property mapping |
-| `[AddFrom<T>]` / `[SubtractFrom<T>]` | Arithmetic |
-| `[Increment<T>]` / `[Decrement<T>]` | ±1 counters |
-| `[Count<T>]` | Absolute count |
-| `[ChildrenFrom<T>]` | Child collection from event |
-| `[Join<T>]` | Join from another event stream |
-| `[RemovedWith<T>]` | Remove entry when event occurs |
-| `[Passive]` | On-demand only, not actively observed |
-
----
-
-## Fluent projections (for complex cases)
-
-```csharp
-public class BorrowedBooksProjection : IProjectionFor<BorrowedBook>
-{
-    public void Define(IProjectionBuilderFor<BorrowedBook> builder) => builder
-        .From<BookBorrowed>(from => from
-            .Set(m => m.UserId).To(e => e.UserId)
-            .Set(m => m.Borrowed).ToEventContextProperty(c => c.Occurred))
-        .Join<BookAddedToInventory>(j => j
-            .On(m => m.Id)
-            .Set(m => m.Title).To(e => e.Title))
-        .RemovedWith<BookReturned>();
-}
-```
-
-AutoMap is on by default — just call `.From<>()` directly. Use `.NoAutoMap()` then explicit `.Set()` calls when you need selective mapping.
-
-**Projections join EVENTS, never read models.**
-
----
-
-## Reducers
-
-```csharp
-public class AccountBalanceReducer : IReducerFor<AccountBalance>
-{
-    public AccountBalance OnDepositMade(DepositMade @event, AccountBalance? current, EventContext context)
-    {
-        var balance = current?.Balance ?? 0m;
-        return new AccountBalance(balance + @event.Amount, context.Occurred);
-    }
-}
-```
-
-- `current` is `null` for the first event — always handle initialization
-- Keep reducers pure — no side effects, no I/O
-- Use `with` expressions on records for state updates
-
----
-
-## Reactors
-
-```csharp
-public class StockKeeping(ICommandPipeline commandPipeline) : IReactor
-{
-    public async Task HandleBookReserved(BookReserved @event) =>
-        await commandPipeline.Execute(new DecreaseStock(@event.Isbn));
-}
-```
-
-- `IReactor` is a marker interface — method dispatch by first-parameter event type
-- Method name is descriptive; `EventContext` parameter is optional
-- `[OnceOnly]` — skips method during event replay
-
----
-
-## Integration specs
-
-Live under `when_<behavior>/` inside the slice folder:
-
-```csharp
-namespace MyApp.Authors.Registration.when_registering;
-
-[Collection(ChronicleCollection.Name)]
-public class and_there_are_no_authors(context context) : Given<context>(context)
-{
-    public class context(ChronicleOutOfProcessFixture fixture) : given.an_http_client(fixture)
-    {
-        public CommandResult<Guid>? Result;
-        async Task Because() =>
-            Result = await Client.ExecuteCommand<RegisterAuthor, Guid>(
-                "/api/authors/register", new RegisterAuthor("John Doe"));
-    }
-
-    [Fact] void should_be_successful() => Context.Result.IsSuccess.ShouldBeTrue();
-}
-```
diff --git a/.github/skills/cratis-vertical-slice/references/slice-types.md b/.github/skills/cratis-vertical-slice/references/slice-types.md
deleted file mode 100644
index 5e86cea..0000000
--- a/.github/skills/cratis-vertical-slice/references/slice-types.md
+++ /dev/null
@@ -1,105 +0,0 @@
-# Slice Types — Reference
-
-## The four slice types
-
-| Type | When to use | Key artifacts |
-| --- | --- | --- |
-| **State Change** | User action that mutates system state | `[Command]` + `[EventType]` + optional validator/constraint |
-| **State View** | Projecting events into queryable read models | `[ReadModel]` + projection/reducer + query methods |
-| **Automation** | Reacting to events to make decisions or call external APIs | `IReactor` + optional local read models |
-| **Translation** | Adapting events between slices or bounded contexts | `IReactor` → `ICommandPipeline.Execute()` |
-
-Most features are built from a **State Change slice** paired with a **State View slice**.
-
----
-
-## State Change slice
-
-**When**: An action happens (user submits a form, a timer fires, an API is called) that changes state.
-
-**Contains**:
-- `[EventType]` records — the facts that occurred
-- `[Command]` record with `Handle()` — validates intent and produces events
-- `CommandValidator<T>` — input validation (FluentValidation, exported to TypeScript)
-- `IConstraint` — server-side business rules enforced at event-append time
-
-**Example**: `Authors/Registration/Registration.cs`
-
-```
-Registration.cs
-├── AuthorRegistered (event)
-├── UniqueAuthorName (constraint)
-├── RegisterAuthorValidator (validator)
-└── RegisterAuthor (command with Handle())
-```
-
----
-
-## State View slice
-
-**When**: Data needs to be queried and displayed. Projects the event stream into a read model.
-
-**Prefer model-bound** (`[ReadModel]` + attribute-based projection) over fluent `IProjectionFor<T>` unless the mapping is too complex for attributes.
-
-**Contains**:
-- `[ReadModel]` record — the query-optimized data shape
-- Projection attributes (`[FromEvent<T>]`, `[SetFrom<T>]`, etc.) or `IProjectionFor<T>`
-- Static query methods on the record (DI parameters auto-resolved)
-
-**Example**: `Authors/Listing/Listing.cs`
-
-```
-Listing.cs
-├── Author (read model record)
-├── [FromEvent<AuthorRegistered>] (projection via attribute)
-└── AllAuthors() static query method
-```
-
----
-
-## Automation slice
-
-**When**: An event should trigger a side effect automatically — sending an email, calling an external API, making a decision.
-
-**Contains**:
-- `IReactor` implementation — dispatches on event type by first parameter
-- Optional local read models for decision state
-
-```csharp
-public class WelcomeEmailSender(IEmailService email) : IReactor
-{
-    public async Task OnAuthorRegistered(AuthorRegistered @event, EventContext ctx) =>
-        await email.SendWelcome(@event.Name);
-}
-```
-
----
-
-## Translation slice
-
-**When**: An event from one slice should trigger a command in another slice (event-driven integration). Keeps slices decoupled — neither knows about the other directly.
-
-**Contains**:
-- `IReactor` that listens to source events
-- `ICommandPipeline.Execute()` to trigger a command in another slice
-
-```csharp
-public class StockKeeping(ICommandPipeline commandPipeline) : IReactor
-{
-    public async Task HandleBookReserved(BookReserved @event) =>
-        await commandPipeline.Execute(new DecreaseStock(@event.Isbn));
-}
-```
-
----
-
-## Decision guide
-
-```
-User action → State Change slice
-Data display → State View slice
-Automatic side effect → Automation slice
-Cross-slice event reaction → Translation slice
-```
-
-A typical feature has at least one State Change + one State View. They are completely separate `.cs` files in separate sub-folders. They share events through the feature's namespace — the State View projection references the event type defined in the State Change file.
diff --git a/.github/skills/new-vertical-slice/SKILL.md b/.github/skills/new-vertical-slice/SKILL.md
deleted file mode 100644
index 647dda5..0000000
--- a/.github/skills/new-vertical-slice/SKILL.md
+++ /dev/null
@@ -1,98 +0,0 @@
----
-name: new-vertical-slice
-description: Use this skill when asked to implement a new feature, command, query, slice, or screen in a Cratis-based project. Guides the full end-to-end workflow: C# backend → specs → dotnet build → React frontend → quality gates.
----
-
-Implement a complete vertical slice following this EXACT order. Never skip steps or work on multiple slices in parallel.
-
-## Step 1 — Identify the slice type
-
-Choose **one** of:
-- **State Change** — a command that mutates state and records events (most common)
-- **State View** — a query that reads from a read model
-- **Automation** — a background reactor triggered by events
-- **Translation** — transforms events into other events
-
-## Step 2 — Determine the namespace root
-
-Read `global.json` and existing `.cs` files in `Features/` to find the namespace root (e.g. `Studio`, `Library`). Never hard-code it.
-
-## Step 3 — Create the C# slice file
-
-Place ALL backend artifacts in a single file: `Features/<Feature>/<Slice>.cs`
-
-File creation order within the slice:
-1. Concept types (if new strongly-typed IDs are needed — see `add-concept` skill)
-2. Command `record` with `Handle()` method and optional validation attributes
-   - If a business rule depends on Chronicle event-sourced state, add the relevant read model as a parameter to `Handle()` — see `add-business-rule` skill (DCB pattern)
-3. Constraint class `<Name>Constraint` (if needed)
-5. Event `record` with `[EventType]` (no arguments, no mutable properties)
-6. Read model `record` with `[ReadModel]` and model-bound projection attributes (`[FromEvent<T>]`, `[Key]`, etc.)
-   - Use fluent `IProjectionFor<T>` only when model-bound attributes don't fit
-
-**Critical rules:**
-- Commands are `record` types with a `Handle()` method directly on them — DO NOT create separate handler classes
-- Events use `[EventType]` with NO arguments — never pass a GUID or string
-- Projection: prefer model-bound attributes on the read model; if using `IProjectionFor<T>`, AutoMap is on by default — just call `.From<>()` directly
-- Namespace must be `<NamespaceRoot>.<Feature>.<Slice>` (drop `.Features.` segment)
-- Copyright header on every file: `// Copyright (c) Cratis. All rights reserved. // Licensed under the MIT license. See LICENSE file in the project root for full license information.`
-
-## Step 4 — Build
-
-Run `dotnet build`. Fix ALL errors and warnings before proceeding. This generates TypeScript proxies.
-
-## Step 5 — Write specs (State Change slices only)
-
-For each command, write specs covering:
-- Happy path — command succeeds, correct event appended
-- Each validation failure (one spec per rule)
-- Each business rule violation (one spec per DCB condition in `Handle()` that inspects a read model)
-- Each constraint violation
-
-See `write-specs` skill for the complete spec structure.
-
-Run `dotnet test`. Fix all failures before proceeding.
-
-## Step 6 — Implement React component(s)
-
-Place `.tsx` files in `Features/<Feature>/<Slice>/`.
-
-- Import the auto-generated command/query proxy from the same folder
-- Use `CommandDialog` from `@cratis/components/CommandDialog` for command dialogs
-- Use `Dialog` from `@cratis/components/Dialogs` for data-only dialogs — NEVER import from `primereact/dialog`
-- Use PrimeReact CSS variables for all colours — never hard-code hex values
-- Use full descriptive variable names — never abbreviations (`event` not `e`, `index` not `idx`)
-- No `any` types — use `unknown` with type guards
-
-**Command usage:**
-```tsx
-const [myCommand] = MyCommand.use();
-const handleSubmit = async () => {
-    myCommand.propertyName = value;
-    const result = await myCommand.execute();
-    if (result.isSuccess) closeDialog(DialogResult.Ok);
-};
-```
-
-**Query with paging:**
-```tsx
-const pageSize = 10;
-const [result, , setPage] = MyQuery.useWithPaging(pageSize);
-// Use result.data, result.paging.totalItems, result.paging.page
-```
-
-## Step 7 — Update the composition page
-
-Open `Features/<Feature>/<Feature>.tsx` and add the new component. If a new page is introduced, also update the router and navigation.
-
-## Step 8 — Quality gates
-
-All must pass before the slice is considered done:
-- `dotnet build` — zero errors/warnings
-- `dotnet test` — zero failures
-- `yarn lint` — zero errors
-- `npx tsc -b` — zero errors
-
----
-
-For complete code patterns for all 4 slice types and frontend examples, see [references/PATTERNS.md](references/PATTERNS.md).
diff --git a/.github/skills/new-vertical-slice/references/PATTERNS.md b/.github/skills/new-vertical-slice/references/PATTERNS.md
deleted file mode 100644
index 639fa8c..0000000
--- a/.github/skills/new-vertical-slice/references/PATTERNS.md
+++ /dev/null
@@ -1,351 +0,0 @@
-# Vertical Slice Patterns
-
-## State Change — full example
-
-```csharp
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-
-namespace MyApp.Projects.Registration;
-
-// ─── Concepts ───────────────────────────────────────────────────────────────
-
-/// <summary>
-/// Represents the unique identifier of a project.
-/// </summary>
-public record ProjectId(Guid Value) : ConceptAs<Guid>(Value)
-{
-    /// <summary>
-    /// Gets a sentinel value for an unset identifier.
-    /// </summary>
-    public static readonly ProjectId NotSet = new(Guid.Empty);
-
-    /// <summary>
-    /// Implicitly converts a <see cref="Guid"/> to a <see cref="ProjectId"/>.
-    /// </summary>
-    /// <param name="value">The <see cref="Guid"/> to convert.</param>
-    public static implicit operator ProjectId(Guid value) => new(value);
-
-    /// <summary>
-    /// Implicitly converts a <see cref="ProjectId"/> to an <see cref="EventSourceId"/>.
-    /// </summary>
-    /// <param name="id">The <see cref="ProjectId"/> to convert.</param>
-    public static implicit operator EventSourceId(ProjectId id) => new(id.Value.ToString());
-
-    /// <summary>
-    /// Creates a new <see cref="ProjectId"/> with a unique value.
-    /// </summary>
-    /// <returns>A new <see cref="ProjectId"/>.</returns>
-    public static ProjectId New() => new(Guid.NewGuid());
-}
-
-/// <summary>
-/// Represents the name of a project.
-/// </summary>
-public record ProjectName(string Value) : ConceptAs<string>(Value)
-{
-    /// <summary>
-    /// Gets a sentinel value for an unset name.
-    /// </summary>
-    public static readonly ProjectName NotSet = new(string.Empty);
-
-    /// <summary>
-    /// Implicitly converts a <see cref="string"/> to a <see cref="ProjectName"/>.
-    /// </summary>
-    /// <param name="value">The string to convert.</param>
-    public static implicit operator ProjectName(string value) => new(value);
-
-    /// <summary>
-    /// Implicitly extracts the underlying <see cref="string"/> value.
-    /// </summary>
-    /// <param name="name">The <see cref="ProjectName"/> to convert.</param>
-    public static implicit operator string(ProjectName name) => name.Value;
-}
-
-// ─── Command ────────────────────────────────────────────────────────────────
-
-/// <summary>
-/// Command to register a new project.
-/// </summary>
-[Command]
-public record RegisterProject(ProjectId ProjectId, ProjectName Name)
-{
-    /// <summary>
-    /// Produces a <see cref="ProjectRegistered"/> event.
-    /// </summary>
-    /// <returns>The <see cref="ProjectRegistered"/> event.</returns>
-    public ProjectRegistered Handle() => new(Name);
-}
-
-// ─── Constraint ─────────────────────────────────────────────────────────────
-
-/// <summary>
-/// Prevents two projects from being registered with the same name.
-/// </summary>
-public class UniqueProjectNameConstraint : IConstraint
-{
-    /// <inheritdoc/>
-    public void Define(IConstraintBuilder builder) => builder
-        .Unique(unique => unique.On<ProjectRegistered>(e => e.Name));
-}
-
-// ─── Event ──────────────────────────────────────────────────────────────────
-
-/// <summary>
-/// Raised when a new project has been successfully registered.
-/// </summary>
-[EventType]
-public record ProjectRegistered(ProjectName Name);
-```
-
-**Key rules for State Change slices:**
-- `[Command]` attribute marks the command record — there is no `ICommand` interface
-- `Handle()` RETURNS the event — it never calls `IEventLog` or `eventLog.Append()`
-- `[EventType]` has **no arguments** — the event type name is resolved automatically
-- The event source ID is resolved in order: `ICanProvideEventSourceId` > `EventSourceId` property > `[Key]` attribute
-- Constraints (`IConstraint`) live in the same slice file as the command
-- The read model and projection live in the **State View** slice, not here
-
----
-
-## State View — full example
-
-```csharp
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-
-namespace MyApp.Projects.Listing;
-
-// ─── Read Model (model-bound projection — no separate class needed) ──────────
-
-/// <summary>
-/// Represents a project in the listing read model.
-/// </summary>
-[ReadModel]
-[FromEvent<Registration.ProjectRegistered>]
-public record Project(
-    [Key] ProjectId Id,
-    ProjectName Name)
-{
-    /// <summary>
-    /// Observes all projects in the collection.
-    /// </summary>
-    /// <param name="collection">The <see cref="IMongoCollection{Project}"/> to observe.</param>
-    /// <returns>An observable subject of all projects.</returns>
-    public static ISubject<IEnumerable<Project>> AllProjects(IMongoCollection<Project> collection) =>
-        collection.Observe();
-}
-```
-
-**Key rules for State View slices:**
-- The read model is decorated with `[ReadModel]` — needed for the static observable query API
-- **Model-bound projection (preferred):** Add `[FromEvent<TEvent>]` at class level for auto-mapping — no separate `IProjectionFor<T>` class needed
-- Mark the primary key property with `[Key]` from `Cratis.Chronicle.Keys`
-- Use `[SetFrom<TEvent>]` / `[AddFrom<TEvent>]` / `[SubtractFrom<TEvent>]` for explicit property-level mapping
-- Use `[ChildrenFrom<TEvent>]` for nested child collections, `[Join<TEvent>]` for cross-event enrichment
-- Query methods are **static methods** on the read model record itself
-- `Observe()` returns an `ISubject<IEnumerable<T>>` for live updates; use `ObserveWithPaging(...)` for paged results
-
-**When to use fluent `IProjectionFor<T>` instead:**
-- Projection logic is too complex for attributes (e.g. conditional branching)
-- You prefer to separate projection definition from the read model type
-
-```csharp
-// Fluent alternative — still correct, use for complex cases
-public class ProjectProjection : IProjectionFor<Project>
-{
-    public void Define(IProjectionBuilderFor<Project> builder) => builder
-        .From<Registration.ProjectRegistered>();     // AutoMap is on by default
-}
-```
-
----
-
-## Automation (Reactor) — full example
-
-```csharp
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-
-namespace MyApp.Projects.Notifications;
-
-/// <summary>
-/// Sends a notification when a project is registered.
-/// </summary>
-/// <param name="notifications">The notification service.</param>
-public class ProjectRegisteredNotifier(INotificationService notifications) : IReactor
-{
-    /// <summary>
-    /// Reacts to <see cref="Registration.ProjectRegistered"/> events.
-    /// </summary>
-    /// <param name="event">The event.</param>
-    /// <param name="context">The event context.</param>
-    public async Task ProjectRegistered(Registration.ProjectRegistered @event, EventContext context) =>
-        await notifications.Notify($"Project '{@event.Name}' was registered.");
-}
-```
-
-**Key rules:**
-- Reactors implement `IReactor` — a marker interface with **no methods**
-- The method name MUST match the event type name exactly (by convention)
-- Reactors MUST be idempotent — they can be called more than once per event
-- Use the event data directly — do not query the read model inside the handler
-- To trigger further commands, inject and call `ICommandPipeline` — never use `IEventLog` directly
-
----
-
-## Translation — full example
-
-```csharp
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-
-namespace MyApp.Projects.StockKeeping;
-
-/// <summary>
-/// Reacts to reservation events and decreases stock accordingly.
-/// </summary>
-/// <param name="stockKeeper">The stock keeper service.</param>
-/// <param name="commandPipeline">The command pipeline for executing commands.</param>
-public class StockKeeping(IStockKeeper stockKeeper, ICommandPipeline commandPipeline) : IReactor
-{
-    /// <summary>
-    /// Handles a <see cref="BookReserved"/> event.
-    /// </summary>
-    /// <param name="event">The event.</param>
-    /// <param name="context">The event context.</param>
-    public async Task BookReserved(BookReserved @event, EventContext context) =>
-        await commandPipeline.Execute(new DecreaseStock(@event.Isbn, await stockKeeper.GetStock(@event.Isbn)));
-}
-
-/// <summary>
-/// Command to decrease available stock of a book.
-/// </summary>
-[Command]
-public record DecreaseStock(ISBN Isbn, BookStock StockBeforeDecrease)
-{
-    /// <summary>
-    /// Produces a <see cref="StockDecreased"/> event.
-    /// </summary>
-    /// <returns>The <see cref="StockDecreased"/> event.</returns>
-    public StockDecreased Handle() => new(Isbn, StockBeforeDecrease);
-}
-
-/// <summary>
-/// Raised when the available stock of a book has decreased.
-/// </summary>
-[EventType]
-public record StockDecreased(ISBN Isbn, BookStock StockBeforeDecrease);
-```
-
----
-
-## Frontend — complete component examples
-
-### Listing component with paging
-
-```tsx
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-
-import { Column } from 'primereact/column';
-import { DataTable } from 'primereact/datatable';
-import { AllProjects } from './AllProjects';
-
-const pageSize = 10;
-
-export const Listing = () => {
-    const [result, , setPage] = AllProjects.useWithPaging(pageSize);
-
-    return (
-        <DataTable
-            lazy paginator
-            value={result.data}
-            rows={pageSize}
-            totalRecords={result.paging.totalItems}
-            alwaysShowPaginator={false}
-            first={result.paging.page * pageSize}
-            onPage={event => setPage(event.page ?? 0)}
-            scrollable scrollHeight="flex"
-            emptyMessage="No projects found.">
-            <Column field="name" header="Name" />
-        </DataTable>
-    );
-};
-```
-
-### CommandDialog for state-change commands
-
-```tsx
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-
-import { useState } from 'react';
-import { DialogProps, DialogResult } from '@cratis/arc.react/dialogs';
-import { CommandDialog } from '@cratis/components/CommandDialog';
-import { InputText } from 'primereact/inputtext';
-import { RegisterProject } from './RegisterProject';
-
-export const AddProject = ({ closeDialog }: DialogProps) => {
-    const [name, setName] = useState('');
-
-    return (
-        <CommandDialog
-            command={RegisterProject}
-            visible
-            header="Add Project"
-            width="32rem"
-            confirmLabel="Add"
-            cancelLabel="Cancel"
-            onBeforeExecute={(values) => {
-                values.name = name;
-                return values;
-            }}
-            onConfirm={() => closeDialog(DialogResult.Ok)}
-            onCancel={() => closeDialog(DialogResult.Cancelled)}>
-            <CommandDialog.Fields>
-                <InputText
-                    value={name}
-                    onChange={event => setName(event.target.value)}
-                    autoFocus
-                />
-            </CommandDialog.Fields>
-        </CommandDialog>
-    );
-};
-```
-
-### Composition page
-
-```tsx
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-
-import { DialogResult, useDialog } from '@cratis/arc.react/dialogs';
-import { Menubar } from 'primereact/menubar';
-import { MenuItem } from 'primereact/menuitem';
-import * as mdIcons from 'react-icons/md';
-import { Page } from '../../Core/Page';
-import { AddProject } from './Registration/AddProject';
-import { Listing } from './Listing/Listing';
-
-export const Projects = () => {
-    const [AddProjectDialog, showAddProjectDialog] = useDialog(AddProject);
-
-    const menuItems: MenuItem[] = [
-        {
-            label: 'Add Project',
-            icon: mdIcons.MdAdd,
-            command: async () => { await showAddProjectDialog(); }
-        }
-    ];
-
-    return (
-        <Page title="Projects">
-            <Menubar model={menuItems} />
-            <Listing />
-            <AddProjectDialog />
-        </Page>
-    );
-};
-```
diff --git a/.github/skills/observable-query-curl/SKILL.md b/.github/skills/observable-query-curl/SKILL.md
deleted file mode 100644
index a2ec1fc..0000000
--- a/.github/skills/observable-query-curl/SKILL.md
+++ /dev/null
@@ -1,156 +0,0 @@
----
-name: observable-query-curl
-description: Practical guidance for debugging and exploring Cratis Arc observable queries with cURL or other plain HTTP clients. Use this whenever the user mentions observable queries together with cURL, curl, HTTP GET debugging, waiting for the first payload, Server-Sent Events, SSE, streaming JSON, or long polling. Also use it when the user wants to inspect an observable query without writing frontend code.
----
-
-# Debug Observable Queries with cURL
-
-Use this skill to help the user work with observable query endpoints from a terminal.
-
-The main choice is:
-
-1. **Snapshot GET** — return the current observable value once
-2. **Wait for first payload** — block until the first payload exists
-3. **SSE stream** — keep the connection open and follow updates
-4. **Long polling** — repeat blocking HTTP requests instead of keeping one stream open
-
-## Step 1 — Identify the endpoint style
-
-Figure out whether the user has:
-
-- a **model-bound observable query** route
-- a **controller-based observable query** route
-- the **observable query demultiplexer SSE endpoint**
-
-If the user does not know the route yet, help them find it before suggesting commands.
-
-## Step 2 — Choose the correct cURL workflow
-
-### A. Get the current snapshot
-
-Use a plain `GET` when the user wants the current observable value right now.
-
-```bash
-curl "https://localhost:5001/api/orders/observe-all"
-```
-
-Explain that this returns the current `QueryResult` snapshot once and then closes.
-
-### B. Wait for the first payload
-
-Use `waitForFirstResult=true` when the observable might not have produced its first value yet.
-
-```bash
-curl "https://localhost:5001/api/orders/observe-all?waitForFirstResult=true"
-```
-
-If the user wants a custom timeout, add `waitForFirstResultTimeout=<seconds>`.
-
-```bash
-curl "https://localhost:5001/api/orders/observe-all?waitForFirstResult=true&waitForFirstResultTimeout=10"
-```
-
-Explain that:
-
-- the timeout value is in **seconds**
-- the response is still a normal JSON `QueryResult`
-- if the observable is not ready and waiting is **not** enabled, the endpoint can return a not-ready response instead of blocking
-
-### C. Stream JSON continuously with SSE
-
-Use SSE when the user wants the observable to keep pushing updates over one connection.
-
-```bash
-curl --no-buffer \
-  -H "Accept: text/event-stream" \
-  "https://localhost:5001/api/orders/observe-all"
-```
-
-Explain that:
-
-- Arc sends SSE frames such as `data: {...}`
-- each frame contains a serialized `QueryResult`
-- `--no-buffer` helps cURL print each event as it arrives
-
-### D. Emulate long polling
-
-Use long polling when the user wants repeated blocking HTTP responses instead of one continuous SSE stream.
-
-```bash
-while true; do
-  curl --silent \
-    "https://localhost:5001/api/orders/observe-all?waitForFirstResult=true&waitForFirstResultTimeout=15"
-  echo
-done
-```
-
-Explain that long polling means:
-
-- each request waits for data or timeout
-- the server returns one JSON payload
-- the client immediately opens the next request
-
-## Step 3 — Tailor the answer to the user's goal
-
-If the user says:
-
-- **"I want to see the first payload"** → prefer `waitForFirstResult=true`
-- **"I want live streaming JSON"** → prefer SSE with `Accept: text/event-stream`
-- **"I want long polling"** → prefer a loop that repeatedly calls the HTTP endpoint with `waitForFirstResult=true`
-- **"I just need the latest value"** → prefer a plain `GET`
-
-## Step 4 — Mention what the user will receive
-
-Always explain the payload shape:
-
-- snapshot and long-poll requests return a normal JSON `QueryResult`
-- SSE returns `data:` frames, each containing a serialized `QueryResult`
-
-If relevant, mention that `QueryResult.Data` holds the full snapshot and `QueryResult.ChangeSet` may also be present for collection updates.
-
-## Response format
-
-When helping the user, prefer this structure:
-
-1. Short recommendation for the correct transport
-2. One or two ready-to-run `curl` commands
-3. One short note explaining what the response looks like
-4. One short note about timeout or retry behavior when relevant
-
-## Examples
-
-**Example 1 — wait for first payload**
-
-User request:
-
-> I need to debug an observable query with curl and the first item is not ready immediately.
-
-Good response shape:
-
-- recommend `waitForFirstResult=true`
-- provide the command
-- mention `waitForFirstResultTimeout`
-
-**Example 2 — streaming JSON**
-
-User request:
-
-> How do I keep watching an observable query from the terminal?
-
-Good response shape:
-
-- recommend SSE
-- provide `curl --no-buffer -H "Accept: text/event-stream" ...`
-- explain the `data:` frames
-
-**Example 3 — long polling**
-
-User request:
-
-> I do not want SSE, I want long polling from curl.
-
-Good response shape:
-
-- provide a looped `curl` example
-- explain that each response is a normal JSON snapshot
-- explain that the client opens the next request after each response
diff --git a/.github/skills/review-code/SKILL.md b/.github/skills/review-code/SKILL.md
deleted file mode 100644
index 4e75a23..0000000
--- a/.github/skills/review-code/SKILL.md
+++ /dev/null
@@ -1,69 +0,0 @@
----
-name: review-code
-description: Use this skill when asked to review, check, or validate code in a Cratis-based project. Produces a structured review report with blocking issues and suggestions, checked against all project architecture and style standards.
----
-
-Review changed code against all Cratis project standards and produce a structured report.
-
-## C# Architecture
-
-- [ ] Each slice: single file `Features/<Feature>/<Slice>.cs` with all backend artifacts
-- [ ] Commands: `record` type with `Handle()` directly on them — no separate handler classes
-- [ ] Events: `record` type, no mutable properties
-- [ ] Projections: AutoMap is on by default — `.AutoMap()` only needed after `.NoAutoMap()`
-- [ ] No service locator (`IServiceProvider` not injected)
-- [ ] Namespace matches folder: `<NamespaceRoot>.<Feature>.<Slice>`
-
-## C# Code Style
-
-- [ ] File-scoped namespaces; `using` directives alphabetically sorted
-- [ ] No unused `using` directives
-- [ ] `is null` / `is not null` — never `== null` / `!= null`
-- [ ] `var` preferred over explicit types
-- [ ] No postfixes: `Async`, `Impl`, `Service` on class names
-- [ ] No regions
-- [ ] All public types, methods, and properties have multiline XML doc comments
-- [ ] `<summary>` tags always multiline — never `/// <summary>Text</summary>` on one line
-- [ ] Methods with parameters include `<param name="...">` for each
-- [ ] Non-void methods include `<returns>`
-- [ ] Custom exception types only — never `InvalidOperationException`, `ArgumentException`, etc.
-- [ ] Exception XML docs start with "The exception that is thrown when …"
-- [ ] Copyright header on every file
-- [ ] Strongly-typed Concepts for all domain IDs/values (no raw `Guid`/`string` in domain models)
-
-## TypeScript Code Style
-
-- [ ] `const` over `let` over `var`
-- [ ] Full descriptive names — never `e`, `idx`, `prev`, `dir`, `pos`
-- [ ] No `any` type — `unknown` with type guards
-- [ ] No `(x as any)` — use `value as unknown as TargetType`
-- [ ] No unused imports
-- [ ] Copyright header on every file
-
-## Component Rules
-
-- [ ] `CommandDialog` from `@cratis/components/CommandDialog` for command dialogs
-- [ ] `Dialog` from `@cratis/components/Dialogs` for data-only dialogs
-- [ ] Never imports `Dialog` from `primereact/dialog` directly
-- [ ] No hard-coded hex/rgb colours — PrimeReact CSS variables only
-- [ ] README.md present for complex component folders with multiple sub-components
-
-## Output format
-
-Start with: **Review result: ✅ Approved / ⚠️ Approved with comments / ❌ Changes requested**
-
-Then list issues:
-```
-### <file path>
-
-**[BLOCKING]** Line N: `problematic code`
-Because: explanation
-Fix:
-```corrected code```
-```
-
-End with a concise summary of what passed and what must change.
-
----
-
-For the full expanded checklists across all categories, see [references/CHECKLISTS.md](references/CHECKLISTS.md).
diff --git a/.github/skills/review-code/references/CHECKLISTS.md b/.github/skills/review-code/references/CHECKLISTS.md
deleted file mode 100644
index 57babb4..0000000
--- a/.github/skills/review-code/references/CHECKLISTS.md
+++ /dev/null
@@ -1,109 +0,0 @@
-# Code Review Checklists
-
-## C# Architecture
-
-- [ ] Each slice lives in a single file at `Features/<Feature>/<Slice>.cs`
-- [ ] ALL backend artifacts in one file: command, validator, business rules, event, read model, projection, slice class
-- [ ] No separate handler classes — `Handle()` is on the command `record` directly
-- [ ] No shared mutable state between commands
-- [ ] No service locator (`IServiceProvider` not injected as a dependency)
-- [ ] No explicit singleton registration when `[Singleton]` attribute suffices
-- [ ] Logging in a separate `*Logging.cs` partial file using `[LoggerMessage]`
-- [ ] Namespace: `<NamespaceRoot>.<Feature>.<Slice>` (no `.Features.` segment)
-
-## C# Commands
-
-- [ ] `record` type, not `class`
-- [ ] All properties use `init` (immutable)
-- [ ] `Handle()` is the single public entry point
-- [ ] No unnecessary constructors — primary constructor only
-- [ ] Validation attributes on properties (`[Required]`, `[MaxLength]`, etc.) for simple rules
-- [ ] Business rules that depend on Chronicle state use a read model parameter in `Handle()` (DCB pattern)
-
-## C# Events
-
-- [ ] `record` type with no mutable properties
-- [ ] Decorated with `[EventType("stable-guid-string")]`
-- [ ] No behaviour — data only
-- [ ] Properties are domain types (Concepts), not raw primitives like `Guid` or `string`
-
-## C# Read Models & Projections
-
-- [ ] Read model is a `record` type
-- [ ] Projection: AutoMap is on by default — `.AutoMap()` only needed after `.NoAutoMap()`
-- [ ] No joins on the read model — joins are on Chronicle events only
-- [ ] `ProjectionId` is a stable GUID string — never changes after first deployment
-- [ ] No `ToList()`, `ToArray()`, or mutable collection exposed from public API
-
-## C# Concepts
-
-- [ ] Domain IDs/values use `ConceptAs<T>` (see `add-concept` skill)
-- [ ] No raw `Guid`, `string`, `int` used where a concept should wrap it
-- [ ] Concept has `static readonly NotSet`/`Empty` sentinel
-- [ ] Concept has implicit conversion from primitive
-- [ ] Concept has `New()` factory if Guid-backed
-
-## C# Code Style
-
-- [ ] File-scoped namespace declaration (`namespace Foo.Bar;`)
-- [ ] `using` directives alphabetically sorted, no unused ones
-- [ ] `is null` / `is not null` — never `== null` / `!= null`
-- [ ] `var` preferred over explicit type declarations
-- [ ] No postfixes on class names: `Async`, `Impl`, `Service`, `Manager`, `Helper`
-- [ ] No regions (`#region`)
-- [ ] No built-in exception types: `InvalidOperationException`, `ArgumentException`, etc.
-- [ ] All public types, methods, and properties have multiline XML doc comments
-- [ ] `<summary>` tags always multiline — never `/// <summary>Text</summary>` on one line
-- [ ] Methods with parameters include `<param name="...">` for each parameter
-- [ ] Non-void methods include `<returns>` documentation
-- [ ] Methods that throw include `<exception cref="...">` documentation
-- [ ] Custom exceptions derive from `Exception`, XML doc starts with "The exception that is thrown when …"
-- [ ] Copyright header on every file
-- [ ] No trailing whitespace or missing newlines at end of file
-
-## TypeScript Architecture
-
-- [ ] Components placed in the slice folder (`Features/<Feature>/<Slice>/`)
-- [ ] No `index.ts` barrel just to re-export a single component
-- [ ] No technical folder groupings (`hooks/`, `utils/`, `types/`) at feature level
-- [ ] Feature folder structure is functional, not technical
-
-## TypeScript Type Safety
-
-- [ ] No `any` types — `unknown` with type guards
-- [ ] No `(x as any)` — use `value as unknown as TargetType`
-- [ ] React synthetic events and DOM events not confused (`React.MouseEvent` vs `MouseEvent`)
-- [ ] Generic defaults use `unknown`, not `any` (e.g. `<T = unknown>`)
-- [ ] No `@ts-ignore` or `@ts-expect-error` without a comment explaining why
-
-## TypeScript Styling
-
-- [ ] No hard-coded hex/rgb values — PrimeReact CSS variables (`var(--...)`) only
-- [ ] CSS co-located with component (`.css` file in same folder)
-- [ ] No `!important` unless justified with a comment
-
-## TypeScript Code Style
-
-- [ ] `const` over `let`, `let` over `var`
-- [ ] Full descriptive names — never `e`, `evt`, `idx`, `i`, `prev`, `dir`, `pos`, `ctx`
-- [ ] No async functions that don't `await` anything
-- [ ] No unused variables or imports
-- [ ] String enums for all enumerations (not numeric)
-- [ ] Copyright header on every file
-
-## Component Conventions
-
-- [ ] `CommandDialog` from `@cratis/components/CommandDialog` for command-based dialogs
-- [ ] `Dialog` from `@cratis/components/Dialogs` for data-only dialogs
-- [ ] Never imports `Dialog` from `primereact/dialog` directly
-- [ ] No monolithic components — decomposed into focused sub-components
-- [ ] `README.md` exists for component folders with ≥2 sub-components or non-trivial architecture
-
-## Spec Coverage
-
-- [ ] Every State Change command has specs
-- [ ] Happy path covered
-- [ ] Every validation rule has a failure spec
-- [ ] Every business rule violation has a spec
-- [ ] Every constraint violation has a spec
-- [ ] No spec for trivial property getters or constructor passthrough
diff --git a/.github/skills/review-performance/SKILL.md b/.github/skills/review-performance/SKILL.md
deleted file mode 100644
index 44d465f..0000000
--- a/.github/skills/review-performance/SKILL.md
+++ /dev/null
@@ -1,55 +0,0 @@
----
-name: review-performance
-description: Use this skill when asked to check for performance issues, inefficiencies, or scalability problems in a Cratis-based project. Covers Chronicle projections, MongoDB query patterns, .NET allocations, and React render overhead.
----
-
-Perform a focused performance review of changed code.
-
-## Chronicle / Event Sourcing
-
-- [ ] Projections use AutoMap (on by default) — avoids manual mapping cost
-- [ ] Projections do NOT join on the read model (forces full re-read)
-- [ ] Reactors do NOT re-query the event log inside `On()` — use event data directly
-- [ ] No eager loading of entire event sequences without paging/filtering
-- [ ] New projections can replay all historical events without crashing
-- [ ] Events are small — no large blobs or base64-encoded content embedded
-
-## MongoDB / Read Models
-
-- [ ] Queries filter on indexed fields — no unintentional full-collection scans
-- [ ] Paged queries use `.Skip()` + `.Take()` — never load all rows
-- [ ] No N+1 pattern — single query returns all needed data
-- [ ] Read-model records do not embed large nested collections that are never fully iterated
-
-## ASP.NET Core / Commands & Queries
-
-- [ ] Query endpoints do not hydrate the full collection when only a count is needed
-- [ ] Command validators are synchronous and in-memory — no I/O in validation
-- [ ] No `await Task.Run(() => syncWork)` wrapping for naturally async work
-- [ ] Response payloads include only fields the client uses — no over-fetching
-
-## React / TypeScript
-
-- [ ] `DataTable` uses `lazy` + `paginator` for collections larger than ~20 rows
-- [ ] No inline object/array literals passed as props (causes identity change every render)
-- [ ] `useEffect` dependencies are correct — no missing deps, no over-broad deps
-- [ ] Large-collection components wrapped in `React.memo` or use stable references
-- [ ] No `JSON.parse(JSON.stringify(x))` deep cloning
-
-## General .NET
-
-- [ ] No LINQ `.ToList()` before `.Where()` — filter before materialising
-- [ ] `IEnumerable<T>` not enumerated multiple times — materialise once if needed
-- [ ] Large object logging uses `{@obj}` only at `Debug` level
-
-## Risk classification
-
-- 🔴 High — measurable degradation at moderate load — must fix before merge
-- 🟡 Medium — could degrade under load or at scale
-- 🟢 Low — minor inefficiency or style issue
-
-## Output format
-
-Start with: **Performance Review: ✅ No issues / ⚠️ Minor findings / ❌ Blocking issues found**
-
-Group findings by category. End with a summary table showing ✅/⚠️/❌ per category.
diff --git a/.github/skills/review-security/SKILL.md b/.github/skills/review-security/SKILL.md
deleted file mode 100644
index e06f17c..0000000
--- a/.github/skills/review-security/SKILL.md
+++ /dev/null
@@ -1,57 +0,0 @@
----
-name: review-security
-description: Use this skill when asked to perform a security review or security audit of code in a Cratis-based project. Checks for injection, auth/authz, data exposure, secrets, and event-sourcing-specific vulnerabilities.
----
-
-Perform a structured security review of changed code.
-
-## Input Validation & Injection
-
-- [ ] All command properties validated before use (null, empty, range, format)
-- [ ] No raw SQL concatenation — parameterised queries or EF Core only
-- [ ] No user-supplied values passed to `Path.Combine`, `File.*`, shell commands, or process args
-- [ ] No user-supplied values used as event-store keys without sanitisation
-
-## Authentication & Authorisation
-
-- [ ] All HTTP endpoints decorated with `[Authorize]` or explicitly `[AllowAnonymous]` with justification
-- [ ] Tenant isolation enforced — no cross-tenant data accessible without authorisation
-- [ ] Claims verified before acting on identity-dependent command data
-
-## Sensitive Data Exposure
-
-- [ ] No passwords, secrets, API keys, or tokens stored in event properties or read models
-- [ ] No PII returned to clients that did not provide it
-- [ ] Query results scoped to requesting tenant/user — never return all-tenant data
-
-## Secrets & Configuration
-
-- [ ] No secrets in source code, config files, or test fixtures
-- [ ] Secrets loaded from environment variables or a secrets manager
-- [ ] No hard-coded connection strings in non-test code
-
-## Event Sourcing Specifics
-
-- [ ] Events are immutable records — no mutable state in the event store
-- [ ] Aggregate/event-store IDs generated server-side, never accepted from untrusted clients
-- [ ] Event upcasting logic does not allow injection of unexpected properties
-- [ ] Uniqueness constraints cannot be bypassed by concurrent multi-tenant writes
-
-## Frontend
-
-- [ ] No user-supplied values in `dangerouslySetInnerHTML`
-- [ ] No tokens or secrets in `localStorage` — use `httpOnly` cookies or in-memory state
-- [ ] Command DTOs contain only the minimum required fields
-- [ ] No client-side access control not also enforced server-side
-
-## Risk classification
-
-- 🔴 Critical — must fix before merge
-- 🟡 Medium — should fix soon
-- 🟢 Low — fix when convenient
-
-## Output format
-
-Start with: **Security Review: ✅ No issues / ⚠️ Low-risk findings / ❌ Blocking issues found**
-
-Group findings by category. End with a summary table showing ✅/⚠️/❌ per category.
diff --git a/.github/skills/scaffold-feature/SKILL.md b/.github/skills/scaffold-feature/SKILL.md
deleted file mode 100644
index bd83ee6..0000000
--- a/.github/skills/scaffold-feature/SKILL.md
+++ /dev/null
@@ -1,82 +0,0 @@
----
-name: scaffold-feature
-description: Use this skill when asked to create a new feature, section, or page that does not yet exist in a Cratis-based project. Sets up the folder, composition page, routing, and navigation entry before any slices are added.
----
-
-Scaffold a brand-new feature folder with routing and navigation — ready for slices.
-
-## What to produce
-
-### 1 — Feature folder
-
-```
-Features/<Feature>/
-├── <Feature>.tsx    ← composition page
-├── <Feature>.css    ← feature-level styles (can be empty initially)
-└── index.ts         ← re-exports the composition page
-```
-
-### 2 — Composition page (`<Feature>.tsx`)
-
-```tsx
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-
-import { Page } from '../../Core/Page';
-
-export const <Feature> = () => {
-    return (
-        <Page title="<NavigationLabel>">
-            {/* Slices will be composed here */}
-        </Page>
-    );
-};
-```
-
-### 3 — `index.ts`
-
-```typescript
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-
-export { <Feature> } from './<Feature>';
-```
-
-### 4 — Update routing
-
-Locate the app router (typically `App.tsx` or a `routes.ts` file) and add:
-
-```tsx
-import { <Feature> } from './Features/<Feature>';
-
-{ path: '<route-path>', element: <<Feature> /> }
-```
-
-### 5 — Update navigation
-
-Locate the sidebar/navigation configuration and add:
-
-```tsx
-import * as mdIcons from 'react-icons/md';
-
-{
-    label: '<NavigationLabel>',
-    icon: mdIcons.<NavigationIcon>,
-    url: '<route-path>'
-}
-```
-
-## Rules
-
-- Feature name: PascalCase (e.g. `Projects`, `Invoices`)
-- Route path: kebab-case (e.g. `/projects`, `/user-management`)
-- Navigation icon: from `react-icons/md` — e.g. `MdFolderOpen`, `MdPeople`
-- Copyright header on every file
-
-## Validation
-
-Run `yarn lint` and `npx tsc -b`. Fix all errors. Confirm the blank page renders without runtime errors.
-
-## Next step
-
-Add slices using the `new-vertical-slice` skill.
diff --git a/.github/skills/ship-changes/SKILL.md b/.github/skills/ship-changes/SKILL.md
deleted file mode 100644
index ef61599..0000000
--- a/.github/skills/ship-changes/SKILL.md
+++ /dev/null
@@ -1,233 +0,0 @@
----
-name: ship-changes
-description: >
-  Ship staged or unstaged local changes: create a branch, make logical commits,
-  push to origin, open a PR with the correct description and label, merge it,
-  and delete the branch locally and on origin. Use whenever the user asks to
-  commit, push, create a PR, ship, or land changes.
----
-
-# Ship Changes
-
-This skill takes local modifications from the working tree and lands them on
-`main` via a properly structured branch, commits, and PR — following all
-project conventions for commits, PR descriptions, and labels.
-
-## Inputs
-
-Collect the following before starting:
-
-- **Semantic label** — `patch`, `minor`, `major`, or none. Skip labeling entirely if the user says no label or omits the label. Do not ask — infer from the nature of the change or follow the user's explicit instruction.
-- **Branch name suffix** — short kebab-case description of the work, e.g. `fix/testing-orleans-runtime-assemblies`. Determine from the nature of the changes if not provided.
-- **Related GitHub issue** — search GitHub issues if the change likely relates to one; use the real number or omit the reference when none exists. Never invent or reuse example numbers.
-
-## Step 1 — Review the working tree
-
-```
-git status
-git diff
-```
-
-Read the full diff. Understand every changed file before touching git.
-Do **not** start staging until you know exactly how the commits will be split.
-
-## Step 2 — Create the branch
-
-Branch off of current `main`. Always use a prefix:
-
-| Prefix | When to use |
-|--------|-------------|
-| `fix/` | bug fixes, runtime errors, incorrect behavior |
-| `feat/` | new features, new slices, new capabilities |
-| `chore/` | build infra, tooling, docs, refactoring |
-
-```
-git checkout -b <prefix>/<short-description>
-```
-
-## Step 3 — Make logical commits
-
-### Commit splitting rules
-
-Split commits so that each one is a single logical unit of work:
-
-1. **Infrastructure / plumbing first** — new types, interfaces, MSBuild targets, shared build props — anything that later commits build on.
-2. **Core behavior second** — the actual fix or feature that uses the infrastructure.
-3. **Specs / tests third** — only when specs are clearly separate from the behavior change (e.g. new integration spec added after the source fix). Combine with behavior commit when tightly coupled.
-4. **Integration or wiring last** — DI registration, routing, UI hookup.
-
-Never mix unrelated changes in a single commit.
-
-### Staging discipline
-
-Stage files explicitly — never `git add .` or `git add -A`:
-
-```
-git add <file1> <file2>
-git diff --cached          # verify staged content before committing
-git commit -m "<message>"
-```
-
-### Commit message format
-
-```
-<imperative summary — 72-char max, no trailing period>
-
-<optional body: WHY the change was made, context, trade-offs>
-<use bullets for multi-part changes>
-```
-
-- Subject starts with a verb: `Add`, `Fix`, `Remove`, `Rename`, `Extract`, `Update`, `Support`.
-- Body separated from subject by a blank line.
-- Body explains *why*, not *what* — the diff shows the what.
-
-**Good examples:**
-
-```
-Add _PackPrivateAssemblyGlobs target for runtime-only NuGet package embedding
-
-Extend the shared client build infrastructure with a new MSBuild target.
-The new PrivatePackageAssemblyGlob item type globs $(OutputPath) at pack
-time and embeds matching DLLs into lib/{tfm}/ without a nuspec dependency.
-```
-
-```
-Fix duplicate key crash in IdentityStorage.Populate
-
-The upsert used InsertOne which threw on existing identities.
-Replace with ReplaceOne using upsert: true.
-```
-
-**Bad examples** (never do these):
-
-- `Fix stuff`
-- `WIP`
-- `Added files`
-- `Fix bug and add feature and update docs`
-
-## Step 4 — Push the branch
-
-```
-git push -u origin <branch-name>
-```
-
-## Step 5 — Create the PR
-
-Use `mcp_github_github_create_pull_request` with:
-
-- `owner`: `Cratis`
-- `repo`: `Chronicle`
-- `head`: the branch name
-- `base`: `main`
-- `title`: short imperative sentence describing the overall change
-- `body`: PR description (see below)
-
-### PR description format
-
-Follow `.github/pull_request_template.md`. Include only non-empty sections.
-
-```markdown
-## Added
-- <release-note bullet> (#<actual-issue-number>)
-
-## Changed
-- <release-note bullet> (#<actual-issue-number>)
-
-## Fixed
-- <release-note bullet> (#<actual-issue-number>)
-```
-
-Rules:
-- Bullets are short, release-note ready, written for a user reading the changelog.
-- End every bullet with `(#<N>)` using the **real** GitHub issue number. Search issues first. If there is no issue, omit the reference entirely — never write `(#issue)` or reuse an example number.
-- Remove any empty sections — no blank headings.
-- Include a `# Summary` paragraph only when the bullets alone don't tell the full story.
-- Never include any Copilot prompt transcript or "Original prompt" block.
-
-### Searching for a related issue
-
-```
-mcp_github_github_search_issues  query="<keywords> repo:Cratis/Chronicle"
-```
-
-If nothing relevant is found, omit the issue reference from affected bullets.
-
-## Step 6 — Add the label (optional)
-
-Skip this step entirely if the user said no label or did not provide one.
-
-Otherwise use `gh pr edit <number> --add-label "<label>"` with one of:
-
-| Label | When |
-|-------|------|
-| `patch` | bug fixes, docs, refactoring with identical behavior |
-| `minor` | new features, new slices, non-breaking additions |
-| `major` | breaking changes to public APIs |
-
-## Step 7 — Merge the PR
-
-Use `mcp_github_github_merge_pull_request` with:
-
-- `merge_method`: `merge`
-- `owner`: `Cratis`
-- `repo`: `Chronicle`
-- `pullNumber`: the PR number returned in step 5
-
-## Step 8 — Clean up the branch
-
-```
-git checkout main && git pull && git branch -d <branch-name> && git push origin --delete <branch-name>
-```
-
-Run all four commands in one shell invocation to avoid partial state.
-After this completes, `main` is fully up to date locally.
-
-## Full example sequence
-
-Below is the exact sequence used when embedding Orleans assemblies in the
-Testing package — use this as a reference for what a correct execution looks like:
-
-```
-# 1. Review
-git status
-git diff
-
-# 2. Branch
-git checkout -b fix/testing-package-orleans-runtime-assemblies
-
-# 3a. First commit — infrastructure
-git add Source/Clients/Directory.Build.props
-git diff --cached
-git commit -m "Add _PackPrivateAssemblyGlobs target for runtime-only NuGet package embedding
-
-Extend the shared client build infrastructure with a new MSBuild target
-_PackPrivateAssemblyGlobs registered via TargetsForTfmSpecificContentInPackage.
-..."
-
-# 3b. Second commit — consumer usage
-git add Source/Clients/Testing/Testing.csproj
-git commit -m "Embed Orleans assemblies in Testing package as runtime-only
-..."
-
-# 4. Push
-git push -u origin fix/testing-package-orleans-runtime-assemblies
-
-# 5. PR (via mcp_github_github_create_pull_request)
-
-# 6. Label
-gh pr edit 2994 --add-label "patch"
-
-# 7. Merge (via mcp_github_github_merge_pull_request)
-
-# 8. Clean up (pulls main as part of the same command)
-git checkout main && git pull && git branch -d fix/testing-package-orleans-runtime-assemblies && git push origin --delete fix/testing-package-orleans-runtime-assemblies
-```
-
-## Common mistakes to avoid
-
-- **Never `git add .`** — always stage specific files and verify with `git diff --cached`.
-- **Never invent issue numbers** — search first; omit the reference if nothing matches.
-- **Never leave placeholder text** in PR bodies (`(#issue)`, `(#123)`).
-- **Never commit code that does not compile** — every commit must be a working state.
-- **Never push directly to `main`** — always go through the branch + PR flow.
-- **Never skip branch cleanup** — delete both the local and remote branch after merging.
diff --git a/.github/skills/skill-creator/LICENSE.txt b/.github/skills/skill-creator/LICENSE.txt
deleted file mode 100644
index 7a4a3ea..0000000
--- a/.github/skills/skill-creator/LICENSE.txt
+++ /dev/null
@@ -1,202 +0,0 @@
-
-                                 Apache License
-                           Version 2.0, January 2004
-                        http://www.apache.org/licenses/
-
-   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
-
-   1. Definitions.
-
-      "License" shall mean the terms and conditions for use, reproduction,
-      and distribution as defined by Sections 1 through 9 of this document.
-
-      "Licensor" shall mean the copyright owner or entity authorized by
-      the copyright owner that is granting the License.
-
-      "Legal Entity" shall mean the union of the acting entity and all
-      other entities that control, are controlled by, or are under common
-      control with that entity. For the purposes of this definition,
-      "control" means (i) the power, direct or indirect, to cause the
-      direction or management of such entity, whether by contract or
-      otherwise, or (ii) ownership of fifty percent (50%) or more of the
-      outstanding shares, or (iii) beneficial ownership of such entity.
-
-      "You" (or "Your") shall mean an individual or Legal Entity
-      exercising permissions granted by this License.
-
-      "Source" form shall mean the preferred form for making modifications,
-      including but not limited to software source code, documentation
-      source, and configuration files.
-
-      "Object" form shall mean any form resulting from mechanical
-      transformation or translation of a Source form, including but
-      not limited to compiled object code, generated documentation,
-      and conversions to other media types.
-
-      "Work" shall mean the work of authorship, whether in Source or
-      Object form, made available under the License, as indicated by a
-      copyright notice that is included in or attached to the work
-      (an example is provided in the Appendix below).
-
-      "Derivative Works" shall mean any work, whether in Source or Object
-      form, that is based on (or derived from) the Work and for which the
-      editorial revisions, annotations, elaborations, or other modifications
-      represent, as a whole, an original work of authorship. For the purposes
-      of this License, Derivative Works shall not include works that remain
-      separable from, or merely link (or bind by name) to the interfaces of,
-      the Work and Derivative Works thereof.
-
-      "Contribution" shall mean any work of authorship, including
-      the original version of the Work and any modifications or additions
-      to that Work or Derivative Works thereof, that is intentionally
-      submitted to Licensor for inclusion in the Work by the copyright owner
-      or by an individual or Legal Entity authorized to submit on behalf of
-      the copyright owner. For the purposes of this definition, "submitted"
-      means any form of electronic, verbal, or written communication sent
-      to the Licensor or its representatives, including but not limited to
-      communication on electronic mailing lists, source code control systems,
-      and issue tracking systems that are managed by, or on behalf of, the
-      Licensor for the purpose of discussing and improving the Work, but
-      excluding communication that is conspicuously marked or otherwise
-      designated in writing by the copyright owner as "Not a Contribution."
-
-      "Contributor" shall mean Licensor and any individual or Legal Entity
-      on behalf of whom a Contribution has been received by Licensor and
-      subsequently incorporated within the Work.
-
-   2. Grant of Copyright License. Subject to the terms and conditions of
-      this License, each Contributor hereby grants to You a perpetual,
-      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
-      copyright license to reproduce, prepare Derivative Works of,
-      publicly display, publicly perform, sublicense, and distribute the
-      Work and such Derivative Works in Source or Object form.
-
-   3. Grant of Patent License. Subject to the terms and conditions of
-      this License, each Contributor hereby grants to You a perpetual,
-      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
-      (except as stated in this section) patent license to make, have made,
-      use, offer to sell, sell, import, and otherwise transfer the Work,
-      where such license applies only to those patent claims licensable
-      by such Contributor that are necessarily infringed by their
-      Contribution(s) alone or by combination of their Contribution(s)
-      with the Work to which such Contribution(s) was submitted. If You
-      institute patent litigation against any entity (including a
-      cross-claim or counterclaim in a lawsuit) alleging that the Work
-      or a Contribution incorporated within the Work constitutes direct
-      or contributory patent infringement, then any patent licenses
-      granted to You under this License for that Work shall terminate
-      as of the date such litigation is filed.
-
-   4. Redistribution. You may reproduce and distribute copies of the
-      Work or Derivative Works thereof in any medium, with or without
-      modifications, and in Source or Object form, provided that You
-      meet the following conditions:
-
-      (a) You must give any other recipients of the Work or
-          Derivative Works a copy of this License; and
-
-      (b) You must cause any modified files to carry prominent notices
-          stating that You changed the files; and
-
-      (c) You must retain, in the Source form of any Derivative Works
-          that You distribute, all copyright, patent, trademark, and
-          attribution notices from the Source form of the Work,
-          excluding those notices that do not pertain to any part of
-          the Derivative Works; and
-
-      (d) If the Work includes a "NOTICE" text file as part of its
-          distribution, then any Derivative Works that You distribute must
-          include a readable copy of the attribution notices contained
-          within such NOTICE file, excluding those notices that do not
-          pertain to any part of the Derivative Works, in at least one
-          of the following places: within a NOTICE text file distributed
-          as part of the Derivative Works; within the Source form or
-          documentation, if provided along with the Derivative Works; or,
-          within a display generated by the Derivative Works, if and
-          wherever such third-party notices normally appear. The contents
-          of the NOTICE file are for informational purposes only and
-          do not modify the License. You may add Your own attribution
-          notices within Derivative Works that You distribute, alongside
-          or as an addendum to the NOTICE text from the Work, provided
-          that such additional attribution notices cannot be construed
-          as modifying the License.
-
-      You may add Your own copyright statement to Your modifications and
-      may provide additional or different license terms and conditions
-      for use, reproduction, or distribution of Your modifications, or
-      for any such Derivative Works as a whole, provided Your use,
-      reproduction, and distribution of the Work otherwise complies with
-      the conditions stated in this License.
-
-   5. Submission of Contributions. Unless You explicitly state otherwise,
-      any Contribution intentionally submitted for inclusion in the Work
-      by You to the Licensor shall be under the terms and conditions of
-      this License, without any additional terms or conditions.
-      Notwithstanding the above, nothing herein shall supersede or modify
-      the terms of any separate license agreement you may have executed
-      with Licensor regarding such Contributions.
-
-   6. Trademarks. This License does not grant permission to use the trade
-      names, trademarks, service marks, or product names of the Licensor,
-      except as required for reasonable and customary use in describing the
-      origin of the Work and reproducing the content of the NOTICE file.
-
-   7. Disclaimer of Warranty. Unless required by applicable law or
-      agreed to in writing, Licensor provides the Work (and each
-      Contributor provides its Contributions) on an "AS IS" BASIS,
-      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
-      implied, including, without limitation, any warranties or conditions
-      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
-      PARTICULAR PURPOSE. You are solely responsible for determining the
-      appropriateness of using or redistributing the Work and assume any
-      risks associated with Your exercise of permissions under this License.
-
-   8. Limitation of Liability. In no event and under no legal theory,
-      whether in tort (including negligence), contract, or otherwise,
-      unless required by applicable law (such as deliberate and grossly
-      negligent acts) or agreed to in writing, shall any Contributor be
-      liable to You for damages, including any direct, indirect, special,
-      incidental, or consequential damages of any character arising as a
-      result of this License or out of the use or inability to use the
-      Work (including but not limited to damages for loss of goodwill,
-      work stoppage, computer failure or malfunction, or any and all
-      other commercial damages or losses), even if such Contributor
-      has been advised of the possibility of such damages.
-
-   9. Accepting Warranty or Additional Liability. While redistributing
-      the Work or Derivative Works thereof, You may choose to offer,
-      and charge a fee for, acceptance of support, warranty, indemnity,
-      or other liability obligations and/or rights consistent with this
-      License. However, in accepting such obligations, You may act only
-      on Your own behalf and on Your sole responsibility, not on behalf
-      of any other Contributor, and only if You agree to indemnify,
-      defend, and hold each Contributor harmless for any liability
-      incurred by, or claims asserted against, such Contributor by reason
-      of your accepting any such warranty or additional liability.
-
-   END OF TERMS AND CONDITIONS
-
-   APPENDIX: How to apply the Apache License to your work.
-
-      To apply the Apache License to your work, attach the following
-      boilerplate notice, with the fields enclosed by brackets "[]"
-      replaced with your own identifying information. (Don't include
-      the brackets!)  The text should be enclosed in the appropriate
-      comment syntax for the file format. We also recommend that a
-      file or class name and description of purpose be included on the
-      same "printed page" as the copyright notice for easier
-      identification within third-party archives.
-
-   Copyright [yyyy] [name of copyright owner]
-
-   Licensed under the Apache License, Version 2.0 (the "License");
-   you may not use this file except in compliance with the License.
-   You may obtain a copy of the License at
-
-       http://www.apache.org/licenses/LICENSE-2.0
-
-   Unless required by applicable law or agreed to in writing, software
-   distributed under the License is distributed on an "AS IS" BASIS,
-   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-   See the License for the specific language governing permissions and
-   limitations under the License.
\ No newline at end of file
diff --git a/.github/skills/skill-creator/SKILL.md b/.github/skills/skill-creator/SKILL.md
deleted file mode 100644
index 5bb069c..0000000
--- a/.github/skills/skill-creator/SKILL.md
+++ /dev/null
@@ -1,486 +0,0 @@
----
-name: skill-creator
-description: Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, update or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.
----
-
-# Skill Creator
-
-A skill for creating new skills and iteratively improving them.
-
-At a high level, the process of creating a skill goes like this:
-
-- Decide what you want the skill to do and roughly how it should do it
-- Write a draft of the skill
-- Create a few test prompts and run claude-with-access-to-the-skill on them
-- Help the user evaluate the results both qualitatively and quantitatively
-  - While the runs happen in the background, draft some quantitative evals if there aren't any (if there are some, you can either use as is or modify if you feel something needs to change about them). Then explain them to the user (or if they already existed, explain the ones that already exist)
-  - Use the `eval-viewer/generate_review.py` script to show the user the results for them to look at, and also let them look at the quantitative metrics
-- Rewrite the skill based on feedback from the user's evaluation of the results (and also if there are any glaring flaws that become apparent from the quantitative benchmarks)
-- Repeat until you're satisfied
-- Expand the test set and try again at larger scale
-
-Your job when using this skill is to figure out where the user is in this process and then jump in and help them progress through these stages. So for instance, maybe they're like "I want to make a skill for X". You can help narrow down what they mean, write a draft, write the test cases, figure out how they want to evaluate, run all the prompts, and repeat.
-
-On the other hand, maybe they already have a draft of the skill. In this case you can go straight to the eval/iterate part of the loop.
-
-Of course, you should always be flexible and if the user is like "I don't need to run a bunch of evaluations, just vibe with me", you can do that instead.
-
-Then after the skill is done (but again, the order is flexible), you can also run the skill description improver, which we have a whole separate script for, to optimize the triggering of the skill.
-
-Cool? Cool.
-
-## Communicating with the user
-
-The skill creator is liable to be used by people across a wide range of familiarity with coding jargon. If you haven't heard (and how could you, it's only very recently that it started), there's a trend now where the power of Claude is inspiring plumbers to open up their terminals, parents and grandparents to google "how to install npm". On the other hand, the bulk of users are probably fairly computer-literate.
-
-So please pay attention to context cues to understand how to phrase your communication! In the default case, just to give you some idea:
-
-- "evaluation" and "benchmark" are borderline, but OK
-- for "JSON" and "assertion" you want to see serious cues from the user that they know what those things are before using them without explaining them
-
-It's OK to briefly explain terms if you're in doubt, and feel free to clarify terms with a short definition if you're unsure if the user will get it.
-
----
-
-## Creating a skill
-
-### Capture Intent
-
-Start by understanding the user's intent. The current conversation might already contain a workflow the user wants to capture (e.g., they say "turn this into a skill"). If so, extract answers from the conversation history first — the tools used, the sequence of steps, corrections the user made, input/output formats observed. The user may need to fill the gaps, and should confirm before proceeding to the next step.
-
-1. What should this skill enable Claude to do?
-2. When should this skill trigger? (what user phrases/contexts)
-3. What's the expected output format?
-4. Should we set up test cases to verify the skill works? Skills with objectively verifiable outputs (file transforms, data extraction, code generation, fixed workflow steps) benefit from test cases. Skills with subjective outputs (writing style, art) often don't need them. Suggest the appropriate default based on the skill type, but let the user decide.
-
-### Interview and Research
-
-Proactively ask questions about edge cases, input/output formats, example files, success criteria, and dependencies. Wait to write test prompts until you've got this part ironed out.
-
-Check available MCPs - if useful for research (searching docs, finding similar skills, looking up best practices), research in parallel via subagents if available, otherwise inline. Come prepared with context to reduce burden on the user.
-
-### Write the SKILL.md
-
-Based on the user interview, fill in these components:
-
-- **name**: Skill identifier
-- **description**: When to trigger, what it does. This is the primary triggering mechanism - include both what the skill does AND specific contexts for when to use it. All "when to use" info goes here, not in the body. Note: currently Claude has a tendency to "undertrigger" skills -- to not use them when they'd be useful. To combat this, please make the skill descriptions a little bit "pushy". So for instance, instead of "How to build a simple fast dashboard to display internal Anthropic data.", you might write "How to build a simple fast dashboard to display internal Anthropic data. Make sure to use this skill whenever the user mentions dashboards, data visualization, internal metrics, or wants to display any kind of company data, even if they don't explicitly ask for a 'dashboard.'"
-- **compatibility**: Required tools, dependencies (optional, rarely needed)
-- **the rest of the skill :)**
-
-### Skill Writing Guide
-
-#### Anatomy of a Skill
-
-```
-skill-name/
-├── SKILL.md (required)
-│   ├── YAML frontmatter (name, description required)
-│   └── Markdown instructions
-└── Bundled Resources (optional)
-    ├── scripts/    - Executable code for deterministic/repetitive tasks
-    ├── references/ - Docs loaded into context as needed
-    └── assets/     - Files used in output (templates, icons, fonts)
-```
-
-#### Progressive Disclosure
-
-Skills use a three-level loading system:
-1. **Metadata** (name + description) - Always in context (~100 words)
-2. **SKILL.md body** - In context whenever skill triggers (<500 lines ideal)
-3. **Bundled resources** - As needed (unlimited, scripts can execute without loading)
-
-These word counts are approximate and you can feel free to go longer if needed.
-
-**Key patterns:**
-- Keep SKILL.md under 500 lines; if you're approaching this limit, add an additional layer of hierarchy along with clear pointers about where the model using the skill should go next to follow up.
-- Reference files clearly from SKILL.md with guidance on when to read them
-- For large reference files (>300 lines), include a table of contents
-
-**Domain organization**: When a skill supports multiple domains/frameworks, organize by variant:
-```
-cloud-deploy/
-├── SKILL.md (workflow + selection)
-└── references/
-    ├── aws.md
-    ├── gcp.md
-    └── azure.md
-```
-Claude reads only the relevant reference file.
-
-#### Principle of Lack of Surprise
-
-This goes without saying, but skills must not contain malware, exploit code, or any content that could compromise system security. A skill's contents should not surprise the user in their intent if described. Don't go along with requests to create misleading skills or skills designed to facilitate unauthorized access, data exfiltration, or other malicious activities. Things like a "roleplay as an XYZ" are OK though.
-
-#### Writing Patterns
-
-Prefer using the imperative form in instructions.
-
-**Defining output formats** - You can do it like this:
-```markdown
-## Report structure
-ALWAYS use this exact template:
-# [Title]
-## Executive summary
-## Key findings
-## Recommendations
-```
-
-**Examples pattern** - It's useful to include examples. You can format them like this (but if "Input" and "Output" are in the examples you might want to deviate a little):
-```markdown
-## Commit message format
-**Example 1:**
-Input: Added user authentication with JWT tokens
-Output: feat(auth): implement JWT-based authentication
-```
-
-### Writing Style
-
-Try to explain to the model why things are important in lieu of heavy-handed musty MUSTs. Use theory of mind and try to make the skill general and not super-narrow to specific examples. Start by writing a draft and then look at it with fresh eyes and improve it.
-
-### Quality Gate Before Finalizing
-
-Before saving any SKILL.md, do a quick self-check:
-
-- **American English only** — this project enforces US spelling throughout all code, comments, and documentation. Scan for common UK variants and replace: `behaviour→behavior`, `colour→color`, `customisation→customization`, `customising→customizing`, `organisation→organization`, `recognise→recognize`, `favour→favor`, `neighbour→neighbor`, `analyse→analyze`, `initialise→initialize`, `finalise→finalize`.
-- All examples follow the project's coding conventions (naming, formatting, structure).
-
-### Test Cases
-
-After writing the skill draft, come up with 2-3 realistic test prompts — the kind of thing a real user would actually say. Share them with the user: [you don't have to use this exact language] "Here are a few test cases I'd like to try. Do these look right, or do you want to add more?" Then run them.
-
-Save test cases to `evals/evals.json`. Don't write assertions yet — just the prompts. You'll draft assertions in the next step while the runs are in progress.
-
-```json
-{
-  "skill_name": "example-skill",
-  "evals": [
-    {
-      "id": 1,
-      "prompt": "User's task prompt",
-      "expected_output": "Description of expected result",
-      "files": []
-    }
-  ]
-}
-```
-
-See `references/schemas.md` for the full schema (including the `assertions` field, which you'll add later).
-
-## Running and evaluating test cases
-
-This section is one continuous sequence — don't stop partway through. Do NOT use `/skill-test` or any other testing skill.
-
-Put results in `<skill-name>-workspace/` as a sibling to the skill directory. Within the workspace, organize results by iteration (`iteration-1/`, `iteration-2/`, etc.) and within that, each test case gets a directory (`eval-0/`, `eval-1/`, etc.). Don't create all of this upfront — just create directories as you go.
-
-### Step 1: Spawn all runs (with-skill AND baseline) in the same turn
-
-For each test case, spawn two subagents in the same turn — one with the skill, one without. This is important: don't spawn the with-skill runs first and then come back for baselines later. Launch everything at once so it all finishes around the same time.
-
-**With-skill run:**
-
-```
-Execute this task:
-- Skill path: <path-to-skill>
-- Task: <eval prompt>
-- Input files: <eval files if any, or "none">
-- Save outputs to: <workspace>/iteration-<N>/eval-<ID>/with_skill/outputs/
-- Outputs to save: <what the user cares about — e.g., "the .docx file", "the final CSV">
-```
-
-**Baseline run** (same prompt, but the baseline depends on context):
-- **Creating a new skill**: no skill at all. Same prompt, no skill path, save to `without_skill/outputs/`.
-- **Improving an existing skill**: the old version. Before editing, snapshot the skill (`cp -r <skill-path> <workspace>/skill-snapshot/`), then point the baseline subagent at the snapshot. Save to `old_skill/outputs/`.
-
-Write an `eval_metadata.json` for each test case (assertions can be empty for now). Give each eval a descriptive name based on what it's testing — not just "eval-0". Use this name for the directory too. If this iteration uses new or modified eval prompts, create these files for each new eval directory — don't assume they carry over from previous iterations.
-
-```json
-{
-  "eval_id": 0,
-  "eval_name": "descriptive-name-here",
-  "prompt": "The user's task prompt",
-  "assertions": []
-}
-```
-
-### Step 2: While runs are in progress, draft assertions
-
-Don't just wait for the runs to finish — you can use this time productively. Draft quantitative assertions for each test case and explain them to the user. If assertions already exist in `evals/evals.json`, review them and explain what they check.
-
-Good assertions are objectively verifiable and have descriptive names — they should read clearly in the benchmark viewer so someone glancing at the results immediately understands what each one checks. Subjective skills (writing style, design quality) are better evaluated qualitatively — don't force assertions onto things that need human judgment.
-
-Update the `eval_metadata.json` files and `evals/evals.json` with the assertions once drafted. Also explain to the user what they'll see in the viewer — both the qualitative outputs and the quantitative benchmark.
-
-### Step 3: As runs complete, capture timing data
-
-When each subagent task completes, you receive a notification containing `total_tokens` and `duration_ms`. Save this data immediately to `timing.json` in the run directory:
-
-```json
-{
-  "total_tokens": 84852,
-  "duration_ms": 23332,
-  "total_duration_seconds": 23.3
-}
-```
-
-This is the only opportunity to capture this data — it comes through the task notification and isn't persisted elsewhere. Process each notification as it arrives rather than trying to batch them.
-
-### Step 4: Grade, aggregate, and launch the viewer
-
-Once all runs are done:
-
-1. **Grade each run** — spawn a grader subagent (or grade inline) that reads `agents/grader.md` and evaluates each assertion against the outputs. Save results to `grading.json` in each run directory. The grading.json expectations array must use the fields `text`, `passed`, and `evidence` (not `name`/`met`/`details` or other variants) — the viewer depends on these exact field names. For assertions that can be checked programmatically, write and run a script rather than eyeballing it — scripts are faster, more reliable, and can be reused across iterations.
-
-2. **Aggregate into benchmark** — run the aggregation script from the skill-creator directory:
-   ```bash
-   python -m scripts.aggregate_benchmark <workspace>/iteration-N --skill-name <name>
-   ```
-   This produces `benchmark.json` and `benchmark.md` with pass_rate, time, and tokens for each configuration, with mean ± stddev and the delta. If generating benchmark.json manually, see `references/schemas.md` for the exact schema the viewer expects.
-Put each with_skill version before its baseline counterpart.
-
-3. **Do an analyst pass** — read the benchmark data and surface patterns the aggregate stats might hide. See `agents/analyzer.md` (the "Analyzing Benchmark Results" section) for what to look for — things like assertions that always pass regardless of skill (non-discriminating), high-variance evals (possibly flaky), and time/token tradeoffs.
-
-4. **Launch the viewer** with both qualitative outputs and quantitative data:
-   ```bash
-   nohup python <skill-creator-path>/eval-viewer/generate_review.py \
-     <workspace>/iteration-N \
-     --skill-name "my-skill" \
-     --benchmark <workspace>/iteration-N/benchmark.json \
-     > /dev/null 2>&1 &
-   VIEWER_PID=$!
-   ```
-   For iteration 2+, also pass `--previous-workspace <workspace>/iteration-<N-1>`.
-
-   **Cowork / headless environments:** If `webbrowser.open()` is not available or the environment has no display, use `--static <output_path>` to write a standalone HTML file instead of starting a server. Feedback will be downloaded as a `feedback.json` file when the user clicks "Submit All Reviews". After download, copy `feedback.json` into the workspace directory for the next iteration to pick up.
-
-Note: please use generate_review.py to create the viewer; there's no need to write custom HTML.
-
-5. **Tell the user** something like: "I've opened the results in your browser. There are two tabs — 'Outputs' lets you click through each test case and leave feedback, 'Benchmark' shows the quantitative comparison. When you're done, come back here and let me know."
-
-### What the user sees in the viewer
-
-The "Outputs" tab shows one test case at a time:
-- **Prompt**: the task that was given
-- **Output**: the files the skill produced, rendered inline where possible
-- **Previous Output** (iteration 2+): collapsed section showing last iteration's output
-- **Formal Grades** (if grading was run): collapsed section showing assertion pass/fail
-- **Feedback**: a textbox that auto-saves as they type
-- **Previous Feedback** (iteration 2+): their comments from last time, shown below the textbox
-
-The "Benchmark" tab shows the stats summary: pass rates, timing, and token usage for each configuration, with per-eval breakdowns and analyst observations.
-
-Navigation is via prev/next buttons or arrow keys. When done, they click "Submit All Reviews" which saves all feedback to `feedback.json`.
-
-### Step 5: Read the feedback
-
-When the user tells you they're done, read `feedback.json`:
-
-```json
-{
-  "reviews": [
-    {"run_id": "eval-0-with_skill", "feedback": "the chart is missing axis labels", "timestamp": "..."},
-    {"run_id": "eval-1-with_skill", "feedback": "", "timestamp": "..."},
-    {"run_id": "eval-2-with_skill", "feedback": "perfect, love this", "timestamp": "..."}
-  ],
-  "status": "complete"
-}
-```
-
-Empty feedback means the user thought it was fine. Focus your improvements on the test cases where the user had specific complaints.
-
-Kill the viewer server when you're done with it:
-
-```bash
-kill $VIEWER_PID 2>/dev/null
-```
-
----
-
-## Improving the skill
-
-This is the heart of the loop. You've run the test cases, the user has reviewed the results, and now you need to make the skill better based on their feedback.
-
-### How to think about improvements
-
-1. **Generalize from the feedback.** The big picture thing that's happening here is that we're trying to create skills that can be used a million times (maybe literally, maybe even more who knows) across many different prompts. Here you and the user are iterating on only a few examples over and over again because it helps move faster. The user knows these examples in and out and it's quick for them to assess new outputs. But if the skill you and the user are codeveloping works only for those examples, it's useless. Rather than put in fiddly overfitty changes, or oppressively constrictive MUSTs, if there's some stubborn issue, you might try branching out and using different metaphors, or recommending different patterns of working. It's relatively cheap to try and maybe you'll land on something great.
-
-2. **Keep the prompt lean.** Remove things that aren't pulling their weight. Make sure to read the transcripts, not just the final outputs — if it looks like the skill is making the model waste a bunch of time doing things that are unproductive, you can try getting rid of the parts of the skill that are making it do that and seeing what happens.
-
-3. **Explain the why.** Try hard to explain the **why** behind everything you're asking the model to do. Today's LLMs are *smart*. They have good theory of mind and when given a good harness can go beyond rote instructions and really make things happen. Even if the feedback from the user is terse or frustrated, try to actually understand the task and why the user is writing what they wrote, and what they actually wrote, and then transmit this understanding into the instructions. If you find yourself writing ALWAYS or NEVER in all caps, or using super rigid structures, that's a yellow flag — if possible, reframe and explain the reasoning so that the model understands why the thing you're asking for is important. That's a more humane, powerful, and effective approach.
-
-4. **Look for repeated work across test cases.** Read the transcripts from the test runs and notice if the subagents all independently wrote similar helper scripts or took the same multi-step approach to something. If all 3 test cases resulted in the subagent writing a `create_docx.py` or a `build_chart.py`, that's a strong signal the skill should bundle that script. Write it once, put it in `scripts/`, and tell the skill to use it. This saves every future invocation from reinventing the wheel.
-
-This task is pretty important (we are trying to create billions a year in economic value here!) and your thinking time is not the blocker; take your time and really mull things over. I'd suggest writing a draft revision and then looking at it anew and making improvements. Really do your best to get into the head of the user and understand what they want and need.
-
-### The iteration loop
-
-After improving the skill:
-
-1. Apply your improvements to the skill
-2. Rerun all test cases into a new `iteration-<N+1>/` directory, including baseline runs. If you're creating a new skill, the baseline is always `without_skill` (no skill) — that stays the same across iterations. If you're improving an existing skill, use your judgment on what makes sense as the baseline: the original version the user came in with, or the previous iteration.
-3. Launch the reviewer with `--previous-workspace` pointing at the previous iteration
-4. Wait for the user to review and tell you they're done
-5. Read the new feedback, improve again, repeat
-
-Keep going until:
-- The user says they're happy
-- The feedback is all empty (everything looks good)
-- You're not making meaningful progress
-
----
-
-## Advanced: Blind comparison
-
-For situations where you want a more rigorous comparison between two versions of a skill (e.g., the user asks "is the new version actually better?"), there's a blind comparison system. Read `agents/comparator.md` and `agents/analyzer.md` for the details. The basic idea is: give two outputs to an independent agent without telling it which is which, and let it judge quality. Then analyze why the winner won.
-
-This is optional, requires subagents, and most users won't need it. The human review loop is usually sufficient.
-
----
-
-## Description Optimization
-
-The description field in SKILL.md frontmatter is the primary mechanism that determines whether Claude invokes a skill. After creating or improving a skill, offer to optimize the description for better triggering accuracy.
-
-### Step 1: Generate trigger eval queries
-
-Create 20 eval queries — a mix of should-trigger and should-not-trigger. Save as JSON:
-
-```json
-[
-  {"query": "the user prompt", "should_trigger": true},
-  {"query": "another prompt", "should_trigger": false}
-]
-```
-
-The queries must be realistic and something a Claude Code or Claude.ai user would actually type. Not abstract requests, but requests that are concrete and specific and have a good amount of detail. For instance, file paths, personal context about the user's job or situation, column names and values, company names, URLs. A little bit of backstory. Some might be in lowercase or contain abbreviations or typos or casual speech. Use a mix of different lengths, and focus on edge cases rather than making them clear-cut (the user will get a chance to sign off on them).
-
-Bad: `"Format this data"`, `"Extract text from PDF"`, `"Create a chart"`
-
-Good: `"ok so my boss just sent me this xlsx file (its in my downloads, called something like 'Q4 sales final FINAL v2.xlsx') and she wants me to add a column that shows the profit margin as a percentage. The revenue is in column C and costs are in column D i think"`
-
-For the **should-trigger** queries (8-10), think about coverage. You want different phrasings of the same intent — some formal, some casual. Include cases where the user doesn't explicitly name the skill or file type but clearly needs it. Throw in some uncommon use cases and cases where this skill competes with another but should win.
-
-For the **should-not-trigger** queries (8-10), the most valuable ones are the near-misses — queries that share keywords or concepts with the skill but actually need something different. Think adjacent domains, ambiguous phrasing where a naive keyword match would trigger but shouldn't, and cases where the query touches on something the skill does but in a context where another tool is more appropriate.
-
-The key thing to avoid: don't make should-not-trigger queries obviously irrelevant. "Write a fibonacci function" as a negative test for a PDF skill is too easy — it doesn't test anything. The negative cases should be genuinely tricky.
-
-### Step 2: Review with user
-
-Present the eval set to the user for review using the HTML template:
-
-1. Read the template from `assets/eval_review.html`
-2. Replace the placeholders:
-   - `__EVAL_DATA_PLACEHOLDER__` → the JSON array of eval items (no quotes around it — it's a JS variable assignment)
-   - `__SKILL_NAME_PLACEHOLDER__` → the skill's name
-   - `__SKILL_DESCRIPTION_PLACEHOLDER__` → the skill's current description
-3. Write to a temp file (e.g., `/tmp/eval_review_<skill-name>.html`) and open it: `open /tmp/eval_review_<skill-name>.html`
-4. The user can edit queries, toggle should-trigger, add/remove entries, then click "Export Eval Set"
-5. The file downloads to `~/Downloads/eval_set.json` — check the Downloads folder for the most recent version in case there are multiple (e.g., `eval_set (1).json`)
-
-This step matters — bad eval queries lead to bad descriptions.
-
-### Step 3: Run the optimization loop
-
-Tell the user: "This will take some time — I'll run the optimization loop in the background and check on it periodically."
-
-Save the eval set to the workspace, then run in the background:
-
-```bash
-python -m scripts.run_loop \
-  --eval-set <path-to-trigger-eval.json> \
-  --skill-path <path-to-skill> \
-  --model <model-id-powering-this-session> \
-  --max-iterations 5 \
-  --verbose
-```
-
-Use the model ID from your system prompt (the one powering the current session) so the triggering test matches what the user actually experiences.
-
-While it runs, periodically tail the output to give the user updates on which iteration it's on and what the scores look like.
-
-This handles the full optimization loop automatically. It splits the eval set into 60% train and 40% held-out test, evaluates the current description (running each query 3 times to get a reliable trigger rate), then calls Claude with extended thinking to propose improvements based on what failed. It re-evaluates each new description on both train and test, iterating up to 5 times. When it's done, it opens an HTML report in the browser showing the results per iteration and returns JSON with `best_description` — selected by test score rather than train score to avoid overfitting.
-
-### How skill triggering works
-
-Understanding the triggering mechanism helps design better eval queries. Skills appear in Claude's `available_skills` list with their name + description, and Claude decides whether to consult a skill based on that description. The important thing to know is that Claude only consults skills for tasks it can't easily handle on its own — simple, one-step queries like "read this PDF" may not trigger a skill even if the description matches perfectly, because Claude can handle them directly with basic tools. Complex, multi-step, or specialized queries reliably trigger skills when the description matches.
-
-This means your eval queries should be substantive enough that Claude would actually benefit from consulting a skill. Simple queries like "read file X" are poor test cases — they won't trigger skills regardless of description quality.
-
-### Step 4: Apply the result
-
-Take `best_description` from the JSON output and update the skill's SKILL.md frontmatter. Show the user before/after and report the scores.
-
----
-
-### Package and Present (only if `present_files` tool is available)
-
-Check whether you have access to the `present_files` tool. If you don't, skip this step. If you do, package the skill and present the .skill file to the user:
-
-```bash
-python -m scripts.package_skill <path/to/skill-folder>
-```
-
-After packaging, direct the user to the resulting `.skill` file path so they can install it.
-
----
-
-## Claude.ai-specific instructions
-
-In Claude.ai, the core workflow is the same (draft → test → review → improve → repeat), but because Claude.ai doesn't have subagents, some mechanics change. Here's what to adapt:
-
-**Running test cases**: No subagents means no parallel execution. For each test case, read the skill's SKILL.md, then follow its instructions to accomplish the test prompt yourself. Do them one at a time. This is less rigorous than independent subagents (you wrote the skill and you're also running it, so you have full context), but it's a useful sanity check — and the human review step compensates. Skip the baseline runs — just use the skill to complete the task as requested.
-
-**Reviewing results**: If you can't open a browser (e.g., Claude.ai's VM has no display, or you're on a remote server), skip the browser reviewer entirely. Instead, present results directly in the conversation. For each test case, show the prompt and the output. If the output is a file the user needs to see (like a .docx or .xlsx), save it to the filesystem and tell them where it is so they can download and inspect it. Ask for feedback inline: "How does this look? Anything you'd change?"
-
-**Benchmarking**: Skip the quantitative benchmarking — it relies on baseline comparisons which aren't meaningful without subagents. Focus on qualitative feedback from the user.
-
-**The iteration loop**: Same as before — improve the skill, rerun the test cases, ask for feedback — just without the browser reviewer in the middle. You can still organize results into iteration directories on the filesystem if you have one.
-
-**Description optimization**: This section requires the `claude` CLI tool (specifically `claude -p`) which is only available in Claude Code. Skip it if you're on Claude.ai.
-
-**Blind comparison**: Requires subagents. Skip it.
-
-**Packaging**: The `package_skill.py` script works anywhere with Python and a filesystem. On Claude.ai, you can run it and the user can download the resulting `.skill` file.
-
----
-
-## Cowork-Specific Instructions
-
-If you're in Cowork, the main things to know are:
-
-- You have subagents, so the main workflow (spawn test cases in parallel, run baselines, grade, etc.) all works. (However, if you run into severe problems with timeouts, it's OK to run the test prompts in series rather than parallel.)
-- You don't have a browser or display, so when generating the eval viewer, use `--static <output_path>` to write a standalone HTML file instead of starting a server. Then proffer a link that the user can click to open the HTML in their browser.
-- For whatever reason, the Cowork setup seems to disincline Claude from generating the eval viewer after running the tests, so just to reiterate: whether you're in Cowork or in Claude Code, after running tests, you should always generate the eval viewer for the human to look at examples before revising the skill yourself and trying to make corrections, using `generate_review.py` (not writing your own boutique html code). Sorry in advance but I'm gonna go all caps here: GENERATE THE EVAL VIEWER *BEFORE* evaluating inputs yourself. You want to get them in front of the human ASAP!
-- Feedback works differently: since there's no running server, the viewer's "Submit All Reviews" button will download `feedback.json` as a file. You can then read it from there (you may have to request access first).
-- Packaging works — `package_skill.py` just needs Python and a filesystem.
-- Description optimization (`run_loop.py` / `run_eval.py`) should work in Cowork just fine since it uses `claude -p` via subprocess, not a browser, but please save it until you've fully finished making the skill and the user agrees it's in good shape.
-
----
-
-## Reference files
-
-The agents/ directory contains instructions for specialized subagents. Read them when you need to spawn the relevant subagent.
-
-- `agents/grader.md` — How to evaluate assertions against outputs
-- `agents/comparator.md` — How to do blind A/B comparison between two outputs
-- `agents/analyzer.md` — How to analyze why one version beat another
-
-The references/ directory has additional documentation:
-- `references/schemas.md` — JSON structures for evals.json, grading.json, etc.
-
----
-
-Repeating one more time the core loop here for emphasis:
-
-- Figure out what the skill is about
-- Draft or edit the skill
-- Run claude-with-access-to-the-skill on test prompts
-- With the user, evaluate the outputs:
-  - Create benchmark.json and run `eval-viewer/generate_review.py` to help the user review them
-  - Run quantitative evals
-- Repeat until you and the user are satisfied
-- Package the final skill and return it to the user.
-
-Please add steps to your TodoList, if you have such a thing, to make sure you don't forget. If you're in Cowork, please specifically put "Create evals JSON and run `eval-viewer/generate_review.py` so human can review test cases" in your TodoList to make sure it happens.
-
-Good luck!
diff --git a/.github/skills/skill-creator/agents/analyzer.md b/.github/skills/skill-creator/agents/analyzer.md
deleted file mode 100644
index 14e41d6..0000000
--- a/.github/skills/skill-creator/agents/analyzer.md
+++ /dev/null
@@ -1,274 +0,0 @@
-# Post-hoc Analyzer Agent
-
-Analyze blind comparison results to understand WHY the winner won and generate improvement suggestions.
-
-## Role
-
-After the blind comparator determines a winner, the Post-hoc Analyzer "unblids" the results by examining the skills and transcripts. The goal is to extract actionable insights: what made the winner better, and how can the loser be improved?
-
-## Inputs
-
-You receive these parameters in your prompt:
-
-- **winner**: "A" or "B" (from blind comparison)
-- **winner_skill_path**: Path to the skill that produced the winning output
-- **winner_transcript_path**: Path to the execution transcript for the winner
-- **loser_skill_path**: Path to the skill that produced the losing output
-- **loser_transcript_path**: Path to the execution transcript for the loser
-- **comparison_result_path**: Path to the blind comparator's output JSON
-- **output_path**: Where to save the analysis results
-
-## Process
-
-### Step 1: Read Comparison Result
-
-1. Read the blind comparator's output at comparison_result_path
-2. Note the winning side (A or B), the reasoning, and any scores
-3. Understand what the comparator valued in the winning output
-
-### Step 2: Read Both Skills
-
-1. Read the winner skill's SKILL.md and key referenced files
-2. Read the loser skill's SKILL.md and key referenced files
-3. Identify structural differences:
-   - Instructions clarity and specificity
-   - Script/tool usage patterns
-   - Example coverage
-   - Edge case handling
-
-### Step 3: Read Both Transcripts
-
-1. Read the winner's transcript
-2. Read the loser's transcript
-3. Compare execution patterns:
-   - How closely did each follow their skill's instructions?
-   - What tools were used differently?
-   - Where did the loser diverge from optimal behavior?
-   - Did either encounter errors or make recovery attempts?
-
-### Step 4: Analyze Instruction Following
-
-For each transcript, evaluate:
-- Did the agent follow the skill's explicit instructions?
-- Did the agent use the skill's provided tools/scripts?
-- Were there missed opportunities to leverage skill content?
-- Did the agent add unnecessary steps not in the skill?
-
-Score instruction following 1-10 and note specific issues.
-
-### Step 5: Identify Winner Strengths
-
-Determine what made the winner better:
-- Clearer instructions that led to better behavior?
-- Better scripts/tools that produced better output?
-- More comprehensive examples that guided edge cases?
-- Better error handling guidance?
-
-Be specific. Quote from skills/transcripts where relevant.
-
-### Step 6: Identify Loser Weaknesses
-
-Determine what held the loser back:
-- Ambiguous instructions that led to suboptimal choices?
-- Missing tools/scripts that forced workarounds?
-- Gaps in edge case coverage?
-- Poor error handling that caused failures?
-
-### Step 7: Generate Improvement Suggestions
-
-Based on the analysis, produce actionable suggestions for improving the loser skill:
-- Specific instruction changes to make
-- Tools/scripts to add or modify
-- Examples to include
-- Edge cases to address
-
-Prioritize by impact. Focus on changes that would have changed the outcome.
-
-### Step 8: Write Analysis Results
-
-Save structured analysis to `{output_path}`.
-
-## Output Format
-
-Write a JSON file with this structure:
-
-```json
-{
-  "comparison_summary": {
-    "winner": "A",
-    "winner_skill": "path/to/winner/skill",
-    "loser_skill": "path/to/loser/skill",
-    "comparator_reasoning": "Brief summary of why comparator chose winner"
-  },
-  "winner_strengths": [
-    "Clear step-by-step instructions for handling multi-page documents",
-    "Included validation script that caught formatting errors",
-    "Explicit guidance on fallback behavior when OCR fails"
-  ],
-  "loser_weaknesses": [
-    "Vague instruction 'process the document appropriately' led to inconsistent behavior",
-    "No script for validation, agent had to improvise and made errors",
-    "No guidance on OCR failure, agent gave up instead of trying alternatives"
-  ],
-  "instruction_following": {
-    "winner": {
-      "score": 9,
-      "issues": [
-        "Minor: skipped optional logging step"
-      ]
-    },
-    "loser": {
-      "score": 6,
-      "issues": [
-        "Did not use the skill's formatting template",
-        "Invented own approach instead of following step 3",
-        "Missed the 'always validate output' instruction"
-      ]
-    }
-  },
-  "improvement_suggestions": [
-    {
-      "priority": "high",
-      "category": "instructions",
-      "suggestion": "Replace 'process the document appropriately' with explicit steps: 1) Extract text, 2) Identify sections, 3) Format per template",
-      "expected_impact": "Would eliminate ambiguity that caused inconsistent behavior"
-    },
-    {
-      "priority": "high",
-      "category": "tools",
-      "suggestion": "Add validate_output.py script similar to winner skill's validation approach",
-      "expected_impact": "Would catch formatting errors before final output"
-    },
-    {
-      "priority": "medium",
-      "category": "error_handling",
-      "suggestion": "Add fallback instructions: 'If OCR fails, try: 1) different resolution, 2) image preprocessing, 3) manual extraction'",
-      "expected_impact": "Would prevent early failure on difficult documents"
-    }
-  ],
-  "transcript_insights": {
-    "winner_execution_pattern": "Read skill -> Followed 5-step process -> Used validation script -> Fixed 2 issues -> Produced output",
-    "loser_execution_pattern": "Read skill -> Unclear on approach -> Tried 3 different methods -> No validation -> Output had errors"
-  }
-}
-```
-
-## Guidelines
-
-- **Be specific**: Quote from skills and transcripts, don't just say "instructions were unclear"
-- **Be actionable**: Suggestions should be concrete changes, not vague advice
-- **Focus on skill improvements**: The goal is to improve the losing skill, not critique the agent
-- **Prioritize by impact**: Which changes would most likely have changed the outcome?
-- **Consider causation**: Did the skill weakness actually cause the worse output, or is it incidental?
-- **Stay objective**: Analyze what happened, don't editorialize
-- **Think about generalization**: Would this improvement help on other evals too?
-
-## Categories for Suggestions
-
-Use these categories to organize improvement suggestions:
-
-| Category | Description |
-|----------|-------------|
-| `instructions` | Changes to the skill's prose instructions |
-| `tools` | Scripts, templates, or utilities to add/modify |
-| `examples` | Example inputs/outputs to include |
-| `error_handling` | Guidance for handling failures |
-| `structure` | Reorganization of skill content |
-| `references` | External docs or resources to add |
-
-## Priority Levels
-
-- **high**: Would likely change the outcome of this comparison
-- **medium**: Would improve quality but may not change win/loss
-- **low**: Nice to have, marginal improvement
-
----
-
-# Analyzing Benchmark Results
-
-When analyzing benchmark results, the analyzer's purpose is to **surface patterns and anomalies** across multiple runs, not suggest skill improvements.
-
-## Role
-
-Review all benchmark run results and generate freeform notes that help the user understand skill performance. Focus on patterns that wouldn't be visible from aggregate metrics alone.
-
-## Inputs
-
-You receive these parameters in your prompt:
-
-- **benchmark_data_path**: Path to the in-progress benchmark.json with all run results
-- **skill_path**: Path to the skill being benchmarked
-- **output_path**: Where to save the notes (as JSON array of strings)
-
-## Process
-
-### Step 1: Read Benchmark Data
-
-1. Read the benchmark.json containing all run results
-2. Note the configurations tested (with_skill, without_skill)
-3. Understand the run_summary aggregates already calculated
-
-### Step 2: Analyze Per-Assertion Patterns
-
-For each expectation across all runs:
-- Does it **always pass** in both configurations? (may not differentiate skill value)
-- Does it **always fail** in both configurations? (may be broken or beyond capability)
-- Does it **always pass with skill but fail without**? (skill clearly adds value here)
-- Does it **always fail with skill but pass without**? (skill may be hurting)
-- Is it **highly variable**? (flaky expectation or non-deterministic behavior)
-
-### Step 3: Analyze Cross-Eval Patterns
-
-Look for patterns across evals:
-- Are certain eval types consistently harder/easier?
-- Do some evals show high variance while others are stable?
-- Are there surprising results that contradict expectations?
-
-### Step 4: Analyze Metrics Patterns
-
-Look at time_seconds, tokens, tool_calls:
-- Does the skill significantly increase execution time?
-- Is there high variance in resource usage?
-- Are there outlier runs that skew the aggregates?
-
-### Step 5: Generate Notes
-
-Write freeform observations as a list of strings. Each note should:
-- State a specific observation
-- Be grounded in the data (not speculation)
-- Help the user understand something the aggregate metrics don't show
-
-Examples:
-- "Assertion 'Output is a PDF file' passes 100% in both configurations - may not differentiate skill value"
-- "Eval 3 shows high variance (50% ± 40%) - run 2 had an unusual failure that may be flaky"
-- "Without-skill runs consistently fail on table extraction expectations (0% pass rate)"
-- "Skill adds 13s average execution time but improves pass rate by 50%"
-- "Token usage is 80% higher with skill, primarily due to script output parsing"
-- "All 3 without-skill runs for eval 1 produced empty output"
-
-### Step 6: Write Notes
-
-Save notes to `{output_path}` as a JSON array of strings:
-
-```json
-[
-  "Assertion 'Output is a PDF file' passes 100% in both configurations - may not differentiate skill value",
-  "Eval 3 shows high variance (50% ± 40%) - run 2 had an unusual failure",
-  "Without-skill runs consistently fail on table extraction expectations",
-  "Skill adds 13s average execution time but improves pass rate by 50%"
-]
-```
-
-## Guidelines
-
-**DO:**
-- Report what you observe in the data
-- Be specific about which evals, expectations, or runs you're referring to
-- Note patterns that aggregate metrics would hide
-- Provide context that helps interpret the numbers
-
-**DO NOT:**
-- Suggest improvements to the skill (that's for the improvement step, not benchmarking)
-- Make subjective quality judgments ("the output was good/bad")
-- Speculate about causes without evidence
-- Repeat information already in the run_summary aggregates
diff --git a/.github/skills/skill-creator/agents/comparator.md b/.github/skills/skill-creator/agents/comparator.md
deleted file mode 100644
index 80e00eb..0000000
--- a/.github/skills/skill-creator/agents/comparator.md
+++ /dev/null
@@ -1,202 +0,0 @@
-# Blind Comparator Agent
-
-Compare two outputs WITHOUT knowing which skill produced them.
-
-## Role
-
-The Blind Comparator judges which output better accomplishes the eval task. You receive two outputs labeled A and B, but you do NOT know which skill produced which. This prevents bias toward a particular skill or approach.
-
-Your judgment is based purely on output quality and task completion.
-
-## Inputs
-
-You receive these parameters in your prompt:
-
-- **output_a_path**: Path to the first output file or directory
-- **output_b_path**: Path to the second output file or directory
-- **eval_prompt**: The original task/prompt that was executed
-- **expectations**: List of expectations to check (optional - may be empty)
-
-## Process
-
-### Step 1: Read Both Outputs
-
-1. Examine output A (file or directory)
-2. Examine output B (file or directory)
-3. Note the type, structure, and content of each
-4. If outputs are directories, examine all relevant files inside
-
-### Step 2: Understand the Task
-
-1. Read the eval_prompt carefully
-2. Identify what the task requires:
-   - What should be produced?
-   - What qualities matter (accuracy, completeness, format)?
-   - What would distinguish a good output from a poor one?
-
-### Step 3: Generate Evaluation Rubric
-
-Based on the task, generate a rubric with two dimensions:
-
-**Content Rubric** (what the output contains):
-| Criterion | 1 (Poor) | 3 (Acceptable) | 5 (Excellent) |
-|-----------|----------|----------------|---------------|
-| Correctness | Major errors | Minor errors | Fully correct |
-| Completeness | Missing key elements | Mostly complete | All elements present |
-| Accuracy | Significant inaccuracies | Minor inaccuracies | Accurate throughout |
-
-**Structure Rubric** (how the output is organized):
-| Criterion | 1 (Poor) | 3 (Acceptable) | 5 (Excellent) |
-|-----------|----------|----------------|---------------|
-| Organization | Disorganized | Reasonably organized | Clear, logical structure |
-| Formatting | Inconsistent/broken | Mostly consistent | Professional, polished |
-| Usability | Difficult to use | Usable with effort | Easy to use |
-
-Adapt criteria to the specific task. For example:
-- PDF form → "Field alignment", "Text readability", "Data placement"
-- Document → "Section structure", "Heading hierarchy", "Paragraph flow"
-- Data output → "Schema correctness", "Data types", "Completeness"
-
-### Step 4: Evaluate Each Output Against the Rubric
-
-For each output (A and B):
-
-1. **Score each criterion** on the rubric (1-5 scale)
-2. **Calculate dimension totals**: Content score, Structure score
-3. **Calculate overall score**: Average of dimension scores, scaled to 1-10
-
-### Step 5: Check Assertions (if provided)
-
-If expectations are provided:
-
-1. Check each expectation against output A
-2. Check each expectation against output B
-3. Count pass rates for each output
-4. Use expectation scores as secondary evidence (not the primary decision factor)
-
-### Step 6: Determine the Winner
-
-Compare A and B based on (in priority order):
-
-1. **Primary**: Overall rubric score (content + structure)
-2. **Secondary**: Assertion pass rates (if applicable)
-3. **Tiebreaker**: If truly equal, declare a TIE
-
-Be decisive - ties should be rare. One output is usually better, even if marginally.
-
-### Step 7: Write Comparison Results
-
-Save results to a JSON file at the path specified (or `comparison.json` if not specified).
-
-## Output Format
-
-Write a JSON file with this structure:
-
-```json
-{
-  "winner": "A",
-  "reasoning": "Output A provides a complete solution with proper formatting and all required fields. Output B is missing the date field and has formatting inconsistencies.",
-  "rubric": {
-    "A": {
-      "content": {
-        "correctness": 5,
-        "completeness": 5,
-        "accuracy": 4
-      },
-      "structure": {
-        "organization": 4,
-        "formatting": 5,
-        "usability": 4
-      },
-      "content_score": 4.7,
-      "structure_score": 4.3,
-      "overall_score": 9.0
-    },
-    "B": {
-      "content": {
-        "correctness": 3,
-        "completeness": 2,
-        "accuracy": 3
-      },
-      "structure": {
-        "organization": 3,
-        "formatting": 2,
-        "usability": 3
-      },
-      "content_score": 2.7,
-      "structure_score": 2.7,
-      "overall_score": 5.4
-    }
-  },
-  "output_quality": {
-    "A": {
-      "score": 9,
-      "strengths": ["Complete solution", "Well-formatted", "All fields present"],
-      "weaknesses": ["Minor style inconsistency in header"]
-    },
-    "B": {
-      "score": 5,
-      "strengths": ["Readable output", "Correct basic structure"],
-      "weaknesses": ["Missing date field", "Formatting inconsistencies", "Partial data extraction"]
-    }
-  },
-  "expectation_results": {
-    "A": {
-      "passed": 4,
-      "total": 5,
-      "pass_rate": 0.80,
-      "details": [
-        {"text": "Output includes name", "passed": true},
-        {"text": "Output includes date", "passed": true},
-        {"text": "Format is PDF", "passed": true},
-        {"text": "Contains signature", "passed": false},
-        {"text": "Readable text", "passed": true}
-      ]
-    },
-    "B": {
-      "passed": 3,
-      "total": 5,
-      "pass_rate": 0.60,
-      "details": [
-        {"text": "Output includes name", "passed": true},
-        {"text": "Output includes date", "passed": false},
-        {"text": "Format is PDF", "passed": true},
-        {"text": "Contains signature", "passed": false},
-        {"text": "Readable text", "passed": true}
-      ]
-    }
-  }
-}
-```
-
-If no expectations were provided, omit the `expectation_results` field entirely.
-
-## Field Descriptions
-
-- **winner**: "A", "B", or "TIE"
-- **reasoning**: Clear explanation of why the winner was chosen (or why it's a tie)
-- **rubric**: Structured rubric evaluation for each output
-  - **content**: Scores for content criteria (correctness, completeness, accuracy)
-  - **structure**: Scores for structure criteria (organization, formatting, usability)
-  - **content_score**: Average of content criteria (1-5)
-  - **structure_score**: Average of structure criteria (1-5)
-  - **overall_score**: Combined score scaled to 1-10
-- **output_quality**: Summary quality assessment
-  - **score**: 1-10 rating (should match rubric overall_score)
-  - **strengths**: List of positive aspects
-  - **weaknesses**: List of issues or shortcomings
-- **expectation_results**: (Only if expectations provided)
-  - **passed**: Number of expectations that passed
-  - **total**: Total number of expectations
-  - **pass_rate**: Fraction passed (0.0 to 1.0)
-  - **details**: Individual expectation results
-
-## Guidelines
-
-- **Stay blind**: DO NOT try to infer which skill produced which output. Judge purely on output quality.
-- **Be specific**: Cite specific examples when explaining strengths and weaknesses.
-- **Be decisive**: Choose a winner unless outputs are genuinely equivalent.
-- **Output quality first**: Assertion scores are secondary to overall task completion.
-- **Be objective**: Don't favor outputs based on style preferences; focus on correctness and completeness.
-- **Explain your reasoning**: The reasoning field should make it clear why you chose the winner.
-- **Handle edge cases**: If both outputs fail, pick the one that fails less badly. If both are excellent, pick the one that's marginally better.
diff --git a/.github/skills/skill-creator/agents/grader.md b/.github/skills/skill-creator/agents/grader.md
deleted file mode 100644
index 558ab05..0000000
--- a/.github/skills/skill-creator/agents/grader.md
+++ /dev/null
@@ -1,223 +0,0 @@
-# Grader Agent
-
-Evaluate expectations against an execution transcript and outputs.
-
-## Role
-
-The Grader reviews a transcript and output files, then determines whether each expectation passes or fails. Provide clear evidence for each judgment.
-
-You have two jobs: grade the outputs, and critique the evals themselves. A passing grade on a weak assertion is worse than useless — it creates false confidence. When you notice an assertion that's trivially satisfied, or an important outcome that no assertion checks, say so.
-
-## Inputs
-
-You receive these parameters in your prompt:
-
-- **expectations**: List of expectations to evaluate (strings)
-- **transcript_path**: Path to the execution transcript (markdown file)
-- **outputs_dir**: Directory containing output files from execution
-
-## Process
-
-### Step 1: Read the Transcript
-
-1. Read the transcript file completely
-2. Note the eval prompt, execution steps, and final result
-3. Identify any issues or errors documented
-
-### Step 2: Examine Output Files
-
-1. List files in outputs_dir
-2. Read/examine each file relevant to the expectations. If outputs aren't plain text, use the inspection tools provided in your prompt — don't rely solely on what the transcript says the executor produced.
-3. Note contents, structure, and quality
-
-### Step 3: Evaluate Each Assertion
-
-For each expectation:
-
-1. **Search for evidence** in the transcript and outputs
-2. **Determine verdict**:
-   - **PASS**: Clear evidence the expectation is true AND the evidence reflects genuine task completion, not just surface-level compliance
-   - **FAIL**: No evidence, or evidence contradicts the expectation, or the evidence is superficial (e.g., correct filename but empty/wrong content)
-3. **Cite the evidence**: Quote the specific text or describe what you found
-
-### Step 4: Extract and Verify Claims
-
-Beyond the predefined expectations, extract implicit claims from the outputs and verify them:
-
-1. **Extract claims** from the transcript and outputs:
-   - Factual statements ("The form has 12 fields")
-   - Process claims ("Used pypdf to fill the form")
-   - Quality claims ("All fields were filled correctly")
-
-2. **Verify each claim**:
-   - **Factual claims**: Can be checked against the outputs or external sources
-   - **Process claims**: Can be verified from the transcript
-   - **Quality claims**: Evaluate whether the claim is justified
-
-3. **Flag unverifiable claims**: Note claims that cannot be verified with available information
-
-This catches issues that predefined expectations might miss.
-
-### Step 5: Read User Notes
-
-If `{outputs_dir}/user_notes.md` exists:
-1. Read it and note any uncertainties or issues flagged by the executor
-2. Include relevant concerns in the grading output
-3. These may reveal problems even when expectations pass
-
-### Step 6: Critique the Evals
-
-After grading, consider whether the evals themselves could be improved. Only surface suggestions when there's a clear gap.
-
-Good suggestions test meaningful outcomes — assertions that are hard to satisfy without actually doing the work correctly. Think about what makes an assertion *discriminating*: it passes when the skill genuinely succeeds and fails when it doesn't.
-
-Suggestions worth raising:
-- An assertion that passed but would also pass for a clearly wrong output (e.g., checking filename existence but not file content)
-- An important outcome you observed — good or bad — that no assertion covers at all
-- An assertion that can't actually be verified from the available outputs
-
-Keep the bar high. The goal is to flag things the eval author would say "good catch" about, not to nitpick every assertion.
-
-### Step 7: Write Grading Results
-
-Save results to `{outputs_dir}/../grading.json` (sibling to outputs_dir).
-
-## Grading Criteria
-
-**PASS when**:
-- The transcript or outputs clearly demonstrate the expectation is true
-- Specific evidence can be cited
-- The evidence reflects genuine substance, not just surface compliance (e.g., a file exists AND contains correct content, not just the right filename)
-
-**FAIL when**:
-- No evidence found for the expectation
-- Evidence contradicts the expectation
-- The expectation cannot be verified from available information
-- The evidence is superficial — the assertion is technically satisfied but the underlying task outcome is wrong or incomplete
-- The output appears to meet the assertion by coincidence rather than by actually doing the work
-
-**When uncertain**: The burden of proof to pass is on the expectation.
-
-### Step 8: Read Executor Metrics and Timing
-
-1. If `{outputs_dir}/metrics.json` exists, read it and include in grading output
-2. If `{outputs_dir}/../timing.json` exists, read it and include timing data
-
-## Output Format
-
-Write a JSON file with this structure:
-
-```json
-{
-  "expectations": [
-    {
-      "text": "The output includes the name 'John Smith'",
-      "passed": true,
-      "evidence": "Found in transcript Step 3: 'Extracted names: John Smith, Sarah Johnson'"
-    },
-    {
-      "text": "The spreadsheet has a SUM formula in cell B10",
-      "passed": false,
-      "evidence": "No spreadsheet was created. The output was a text file."
-    },
-    {
-      "text": "The assistant used the skill's OCR script",
-      "passed": true,
-      "evidence": "Transcript Step 2 shows: 'Tool: Bash - python ocr_script.py image.png'"
-    }
-  ],
-  "summary": {
-    "passed": 2,
-    "failed": 1,
-    "total": 3,
-    "pass_rate": 0.67
-  },
-  "execution_metrics": {
-    "tool_calls": {
-      "Read": 5,
-      "Write": 2,
-      "Bash": 8
-    },
-    "total_tool_calls": 15,
-    "total_steps": 6,
-    "errors_encountered": 0,
-    "output_chars": 12450,
-    "transcript_chars": 3200
-  },
-  "timing": {
-    "executor_duration_seconds": 165.0,
-    "grader_duration_seconds": 26.0,
-    "total_duration_seconds": 191.0
-  },
-  "claims": [
-    {
-      "claim": "The form has 12 fillable fields",
-      "type": "factual",
-      "verified": true,
-      "evidence": "Counted 12 fields in field_info.json"
-    },
-    {
-      "claim": "All required fields were populated",
-      "type": "quality",
-      "verified": false,
-      "evidence": "Reference section was left blank despite data being available"
-    }
-  ],
-  "user_notes_summary": {
-    "uncertainties": ["Used 2023 data, may be stale"],
-    "needs_review": [],
-    "workarounds": ["Fell back to text overlay for non-fillable fields"]
-  },
-  "eval_feedback": {
-    "suggestions": [
-      {
-        "assertion": "The output includes the name 'John Smith'",
-        "reason": "A hallucinated document that mentions the name would also pass — consider checking it appears as the primary contact with matching phone and email from the input"
-      },
-      {
-        "reason": "No assertion checks whether the extracted phone numbers match the input — I observed incorrect numbers in the output that went uncaught"
-      }
-    ],
-    "overall": "Assertions check presence but not correctness. Consider adding content verification."
-  }
-}
-```
-
-## Field Descriptions
-
-- **expectations**: Array of graded expectations
-  - **text**: The original expectation text
-  - **passed**: Boolean - true if expectation passes
-  - **evidence**: Specific quote or description supporting the verdict
-- **summary**: Aggregate statistics
-  - **passed**: Count of passed expectations
-  - **failed**: Count of failed expectations
-  - **total**: Total expectations evaluated
-  - **pass_rate**: Fraction passed (0.0 to 1.0)
-- **execution_metrics**: Copied from executor's metrics.json (if available)
-  - **output_chars**: Total character count of output files (proxy for tokens)
-  - **transcript_chars**: Character count of transcript
-- **timing**: Wall clock timing from timing.json (if available)
-  - **executor_duration_seconds**: Time spent in executor subagent
-  - **total_duration_seconds**: Total elapsed time for the run
-- **claims**: Extracted and verified claims from the output
-  - **claim**: The statement being verified
-  - **type**: "factual", "process", or "quality"
-  - **verified**: Boolean - whether the claim holds
-  - **evidence**: Supporting or contradicting evidence
-- **user_notes_summary**: Issues flagged by the executor
-  - **uncertainties**: Things the executor wasn't sure about
-  - **needs_review**: Items requiring human attention
-  - **workarounds**: Places where the skill didn't work as expected
-- **eval_feedback**: Improvement suggestions for the evals (only when warranted)
-  - **suggestions**: List of concrete suggestions, each with a `reason` and optionally an `assertion` it relates to
-  - **overall**: Brief assessment — can be "No suggestions, evals look solid" if nothing to flag
-
-## Guidelines
-
-- **Be objective**: Base verdicts on evidence, not assumptions
-- **Be specific**: Quote the exact text that supports your verdict
-- **Be thorough**: Check both transcript and output files
-- **Be consistent**: Apply the same standard to each expectation
-- **Explain failures**: Make it clear why evidence was insufficient
-- **No partial credit**: Each expectation is pass or fail, not partial
diff --git a/.github/skills/skill-creator/assets/eval_review.html b/.github/skills/skill-creator/assets/eval_review.html
deleted file mode 100644
index 938ff32..0000000
--- a/.github/skills/skill-creator/assets/eval_review.html
+++ /dev/null
@@ -1,146 +0,0 @@
-<!DOCTYPE html>
-<html lang="en">
-<head>
-  <meta charset="UTF-8">
-  <meta name="viewport" content="width=device-width, initial-scale=1.0">
-  <title>Eval Set Review - __SKILL_NAME_PLACEHOLDER__</title>
-  <link rel="preconnect" href="https://fonts.googleapis.com">
-  <link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
-  <link href="https://fonts.googleapis.com/css2?family=Poppins:wght@500;600&family=Lora:wght@400;500&display=swap" rel="stylesheet">
-  <style>
-    * { box-sizing: border-box; margin: 0; padding: 0; }
-    body { font-family: 'Lora', Georgia, serif; background: #faf9f5; padding: 2rem; color: #141413; }
-    h1 { font-family: 'Poppins', sans-serif; margin-bottom: 0.5rem; font-size: 1.5rem; }
-    .description { color: #b0aea5; margin-bottom: 1.5rem; font-style: italic; max-width: 900px; }
-    .controls { margin-bottom: 1rem; display: flex; gap: 0.5rem; }
-    .btn { font-family: 'Poppins', sans-serif; padding: 0.5rem 1rem; border: none; border-radius: 6px; cursor: pointer; font-size: 0.875rem; font-weight: 500; }
-    .btn-add { background: #6a9bcc; color: white; }
-    .btn-add:hover { background: #5889b8; }
-    .btn-export { background: #d97757; color: white; }
-    .btn-export:hover { background: #c4613f; }
-    table { width: 100%; max-width: 1100px; border-collapse: collapse; background: white; border-radius: 6px; overflow: hidden; box-shadow: 0 1px 3px rgba(0,0,0,0.08); }
-    th { font-family: 'Poppins', sans-serif; background: #141413; color: #faf9f5; padding: 0.75rem 1rem; text-align: left; font-size: 0.875rem; }
-    td { padding: 0.75rem 1rem; border-bottom: 1px solid #e8e6dc; vertical-align: top; }
-    tr:nth-child(even) td { background: #faf9f5; }
-    tr:hover td { background: #f3f1ea; }
-    .section-header td { background: #e8e6dc; font-family: 'Poppins', sans-serif; font-weight: 500; font-size: 0.8rem; color: #141413; text-transform: uppercase; letter-spacing: 0.05em; }
-    .query-input { width: 100%; padding: 0.4rem; border: 1px solid #e8e6dc; border-radius: 4px; font-size: 0.875rem; font-family: 'Lora', Georgia, serif; resize: vertical; min-height: 60px; }
-    .query-input:focus { outline: none; border-color: #d97757; box-shadow: 0 0 0 2px rgba(217,119,87,0.15); }
-    .toggle { position: relative; display: inline-block; width: 44px; height: 24px; }
-    .toggle input { opacity: 0; width: 0; height: 0; }
-    .toggle .slider { position: absolute; inset: 0; background: #b0aea5; border-radius: 24px; cursor: pointer; transition: 0.2s; }
-    .toggle .slider::before { content: ""; position: absolute; width: 18px; height: 18px; left: 3px; bottom: 3px; background: white; border-radius: 50%; transition: 0.2s; }
-    .toggle input:checked + .slider { background: #d97757; }
-    .toggle input:checked + .slider::before { transform: translateX(20px); }
-    .btn-delete { background: #c44; color: white; padding: 0.3rem 0.6rem; border: none; border-radius: 4px; cursor: pointer; font-size: 0.75rem; font-family: 'Poppins', sans-serif; }
-    .btn-delete:hover { background: #a33; }
-    .summary { margin-top: 1rem; color: #b0aea5; font-size: 0.875rem; }
-  </style>
-</head>
-<body>
-  <h1>Eval Set Review: <span id="skill-name">__SKILL_NAME_PLACEHOLDER__</span></h1>
-  <p class="description">Current description: <span id="skill-desc">__SKILL_DESCRIPTION_PLACEHOLDER__</span></p>
-
-  <div class="controls">
-    <button class="btn btn-add" onclick="addRow()">+ Add Query</button>
-    <button class="btn btn-export" onclick="exportEvalSet()">Export Eval Set</button>
-  </div>
-
-  <table>
-    <thead>
-      <tr>
-        <th style="width:65%">Query</th>
-        <th style="width:18%">Should Trigger</th>
-        <th style="width:10%">Actions</th>
-      </tr>
-    </thead>
-    <tbody id="eval-body"></tbody>
-  </table>
-
-  <p class="summary" id="summary"></p>
-
-  <script>
-    const EVAL_DATA = __EVAL_DATA_PLACEHOLDER__;
-
-    let evalItems = [...EVAL_DATA];
-
-    function render() {
-      const tbody = document.getElementById('eval-body');
-      tbody.innerHTML = '';
-
-      // Sort: should-trigger first, then should-not-trigger
-      const sorted = evalItems
-        .map((item, origIdx) => ({ ...item, origIdx }))
-        .sort((a, b) => (b.should_trigger ? 1 : 0) - (a.should_trigger ? 1 : 0));
-
-      let lastGroup = null;
-      sorted.forEach(item => {
-        const group = item.should_trigger ? 'trigger' : 'no-trigger';
-        if (group !== lastGroup) {
-          const headerRow = document.createElement('tr');
-          headerRow.className = 'section-header';
-          headerRow.innerHTML = `<td colspan="3">${item.should_trigger ? 'Should Trigger' : 'Should NOT Trigger'}</td>`;
-          tbody.appendChild(headerRow);
-          lastGroup = group;
-        }
-
-        const idx = item.origIdx;
-        const tr = document.createElement('tr');
-        tr.innerHTML = `
-          <td><textarea class="query-input" onchange="updateQuery(${idx}, this.value)">${escapeHtml(item.query)}</textarea></td>
-          <td>
-            <label class="toggle">
-              <input type="checkbox" ${item.should_trigger ? 'checked' : ''} onchange="updateTrigger(${idx}, this.checked)">
-              <span class="slider"></span>
-            </label>
-            <span style="margin-left:8px;font-size:0.8rem;color:#b0aea5">${item.should_trigger ? 'Yes' : 'No'}</span>
-          </td>
-          <td><button class="btn-delete" onclick="deleteRow(${idx})">Delete</button></td>
-        `;
-        tbody.appendChild(tr);
-      });
-      updateSummary();
-    }
-
-    function escapeHtml(text) {
-      const div = document.createElement('div');
-      div.textContent = text;
-      return div.innerHTML;
-    }
-
-    function updateQuery(idx, value) { evalItems[idx].query = value; updateSummary(); }
-    function updateTrigger(idx, value) { evalItems[idx].should_trigger = value; render(); }
-    function deleteRow(idx) { evalItems.splice(idx, 1); render(); }
-
-    function addRow() {
-      evalItems.push({ query: '', should_trigger: true });
-      render();
-      const inputs = document.querySelectorAll('.query-input');
-      inputs[inputs.length - 1].focus();
-    }
-
-    function updateSummary() {
-      const trigger = evalItems.filter(i => i.should_trigger).length;
-      const noTrigger = evalItems.filter(i => !i.should_trigger).length;
-      document.getElementById('summary').textContent =
-        `${evalItems.length} queries total: ${trigger} should trigger, ${noTrigger} should not trigger`;
-    }
-
-    function exportEvalSet() {
-      const valid = evalItems.filter(i => i.query.trim() !== '');
-      const data = valid.map(i => ({ query: i.query.trim(), should_trigger: i.should_trigger }));
-      const blob = new Blob([JSON.stringify(data, null, 2)], { type: 'application/json' });
-      const url = URL.createObjectURL(blob);
-      const a = document.createElement('a');
-      a.href = url;
-      a.download = 'eval_set.json';
-      document.body.appendChild(a);
-      a.click();
-      document.body.removeChild(a);
-      URL.revokeObjectURL(url);
-    }
-
-    render();
-  </script>
-</body>
-</html>
diff --git a/.github/skills/skill-creator/eval-viewer/generate_review.py b/.github/skills/skill-creator/eval-viewer/generate_review.py
deleted file mode 100644
index 7fa5978..0000000
--- a/.github/skills/skill-creator/eval-viewer/generate_review.py
+++ /dev/null
@@ -1,471 +0,0 @@
-#!/usr/bin/env python3
-"""Generate and serve a review page for eval results.
-
-Reads the workspace directory, discovers runs (directories with outputs/),
-embeds all output data into a self-contained HTML page, and serves it via
-a tiny HTTP server. Feedback auto-saves to feedback.json in the workspace.
-
-Usage:
-    python generate_review.py <workspace-path> [--port PORT] [--skill-name NAME]
-    python generate_review.py <workspace-path> --previous-feedback /path/to/old/feedback.json
-
-No dependencies beyond the Python stdlib are required.
-"""
-
-import argparse
-import base64
-import json
-import mimetypes
-import os
-import re
-import signal
-import subprocess
-import sys
-import time
-import webbrowser
-from functools import partial
-from http.server import HTTPServer, BaseHTTPRequestHandler
-from pathlib import Path
-
-# Files to exclude from output listings
-METADATA_FILES = {"transcript.md", "user_notes.md", "metrics.json"}
-
-# Extensions we render as inline text
-TEXT_EXTENSIONS = {
-    ".txt", ".md", ".json", ".csv", ".py", ".js", ".ts", ".tsx", ".jsx",
-    ".yaml", ".yml", ".xml", ".html", ".css", ".sh", ".rb", ".go", ".rs",
-    ".java", ".c", ".cpp", ".h", ".hpp", ".sql", ".r", ".toml",
-}
-
-# Extensions we render as inline images
-IMAGE_EXTENSIONS = {".png", ".jpg", ".jpeg", ".gif", ".svg", ".webp"}
-
-# MIME type overrides for common types
-MIME_OVERRIDES = {
-    ".svg": "image/svg+xml",
-    ".xlsx": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
-    ".docx": "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
-    ".pptx": "application/vnd.openxmlformats-officedocument.presentationml.presentation",
-}
-
-
-def get_mime_type(path: Path) -> str:
-    ext = path.suffix.lower()
-    if ext in MIME_OVERRIDES:
-        return MIME_OVERRIDES[ext]
-    mime, _ = mimetypes.guess_type(str(path))
-    return mime or "application/octet-stream"
-
-
-def find_runs(workspace: Path) -> list[dict]:
-    """Recursively find directories that contain an outputs/ subdirectory."""
-    runs: list[dict] = []
-    _find_runs_recursive(workspace, workspace, runs)
-    runs.sort(key=lambda r: (r.get("eval_id", float("inf")), r["id"]))
-    return runs
-
-
-def _find_runs_recursive(root: Path, current: Path, runs: list[dict]) -> None:
-    if not current.is_dir():
-        return
-
-    outputs_dir = current / "outputs"
-    if outputs_dir.is_dir():
-        run = build_run(root, current)
-        if run:
-            runs.append(run)
-        return
-
-    skip = {"node_modules", ".git", "__pycache__", "skill", "inputs"}
-    for child in sorted(current.iterdir()):
-        if child.is_dir() and child.name not in skip:
-            _find_runs_recursive(root, child, runs)
-
-
-def build_run(root: Path, run_dir: Path) -> dict | None:
-    """Build a run dict with prompt, outputs, and grading data."""
-    prompt = ""
-    eval_id = None
-
-    # Try eval_metadata.json
-    for candidate in [run_dir / "eval_metadata.json", run_dir.parent / "eval_metadata.json"]:
-        if candidate.exists():
-            try:
-                metadata = json.loads(candidate.read_text())
-                prompt = metadata.get("prompt", "")
-                eval_id = metadata.get("eval_id")
-            except (json.JSONDecodeError, OSError):
-                pass
-            if prompt:
-                break
-
-    # Fall back to transcript.md
-    if not prompt:
-        for candidate in [run_dir / "transcript.md", run_dir / "outputs" / "transcript.md"]:
-            if candidate.exists():
-                try:
-                    text = candidate.read_text()
-                    match = re.search(r"## Eval Prompt\n\n([\s\S]*?)(?=\n##|$)", text)
-                    if match:
-                        prompt = match.group(1).strip()
-                except OSError:
-                    pass
-                if prompt:
-                    break
-
-    if not prompt:
-        prompt = "(No prompt found)"
-
-    run_id = str(run_dir.relative_to(root)).replace("/", "-").replace("\\", "-")
-
-    # Collect output files
-    outputs_dir = run_dir / "outputs"
-    output_files: list[dict] = []
-    if outputs_dir.is_dir():
-        for f in sorted(outputs_dir.iterdir()):
-            if f.is_file() and f.name not in METADATA_FILES:
-                output_files.append(embed_file(f))
-
-    # Load grading if present
-    grading = None
-    for candidate in [run_dir / "grading.json", run_dir.parent / "grading.json"]:
-        if candidate.exists():
-            try:
-                grading = json.loads(candidate.read_text())
-            except (json.JSONDecodeError, OSError):
-                pass
-            if grading:
-                break
-
-    return {
-        "id": run_id,
-        "prompt": prompt,
-        "eval_id": eval_id,
-        "outputs": output_files,
-        "grading": grading,
-    }
-
-
-def embed_file(path: Path) -> dict:
-    """Read a file and return an embedded representation."""
-    ext = path.suffix.lower()
-    mime = get_mime_type(path)
-
-    if ext in TEXT_EXTENSIONS:
-        try:
-            content = path.read_text(errors="replace")
-        except OSError:
-            content = "(Error reading file)"
-        return {
-            "name": path.name,
-            "type": "text",
-            "content": content,
-        }
-    elif ext in IMAGE_EXTENSIONS:
-        try:
-            raw = path.read_bytes()
-            b64 = base64.b64encode(raw).decode("ascii")
-        except OSError:
-            return {"name": path.name, "type": "error", "content": "(Error reading file)"}
-        return {
-            "name": path.name,
-            "type": "image",
-            "mime": mime,
-            "data_uri": f"data:{mime};base64,{b64}",
-        }
-    elif ext == ".pdf":
-        try:
-            raw = path.read_bytes()
-            b64 = base64.b64encode(raw).decode("ascii")
-        except OSError:
-            return {"name": path.name, "type": "error", "content": "(Error reading file)"}
-        return {
-            "name": path.name,
-            "type": "pdf",
-            "data_uri": f"data:{mime};base64,{b64}",
-        }
-    elif ext == ".xlsx":
-        try:
-            raw = path.read_bytes()
-            b64 = base64.b64encode(raw).decode("ascii")
-        except OSError:
-            return {"name": path.name, "type": "error", "content": "(Error reading file)"}
-        return {
-            "name": path.name,
-            "type": "xlsx",
-            "data_b64": b64,
-        }
-    else:
-        # Binary / unknown — base64 download link
-        try:
-            raw = path.read_bytes()
-            b64 = base64.b64encode(raw).decode("ascii")
-        except OSError:
-            return {"name": path.name, "type": "error", "content": "(Error reading file)"}
-        return {
-            "name": path.name,
-            "type": "binary",
-            "mime": mime,
-            "data_uri": f"data:{mime};base64,{b64}",
-        }
-
-
-def load_previous_iteration(workspace: Path) -> dict[str, dict]:
-    """Load previous iteration's feedback and outputs.
-
-    Returns a map of run_id -> {"feedback": str, "outputs": list[dict]}.
-    """
-    result: dict[str, dict] = {}
-
-    # Load feedback
-    feedback_map: dict[str, str] = {}
-    feedback_path = workspace / "feedback.json"
-    if feedback_path.exists():
-        try:
-            data = json.loads(feedback_path.read_text())
-            feedback_map = {
-                r["run_id"]: r["feedback"]
-                for r in data.get("reviews", [])
-                if r.get("feedback", "").strip()
-            }
-        except (json.JSONDecodeError, OSError, KeyError):
-            pass
-
-    # Load runs (to get outputs)
-    prev_runs = find_runs(workspace)
-    for run in prev_runs:
-        result[run["id"]] = {
-            "feedback": feedback_map.get(run["id"], ""),
-            "outputs": run.get("outputs", []),
-        }
-
-    # Also add feedback for run_ids that had feedback but no matching run
-    for run_id, fb in feedback_map.items():
-        if run_id not in result:
-            result[run_id] = {"feedback": fb, "outputs": []}
-
-    return result
-
-
-def generate_html(
-    runs: list[dict],
-    skill_name: str,
-    previous: dict[str, dict] | None = None,
-    benchmark: dict | None = None,
-) -> str:
-    """Generate the complete standalone HTML page with embedded data."""
-    template_path = Path(__file__).parent / "viewer.html"
-    template = template_path.read_text()
-
-    # Build previous_feedback and previous_outputs maps for the template
-    previous_feedback: dict[str, str] = {}
-    previous_outputs: dict[str, list[dict]] = {}
-    if previous:
-        for run_id, data in previous.items():
-            if data.get("feedback"):
-                previous_feedback[run_id] = data["feedback"]
-            if data.get("outputs"):
-                previous_outputs[run_id] = data["outputs"]
-
-    embedded = {
-        "skill_name": skill_name,
-        "runs": runs,
-        "previous_feedback": previous_feedback,
-        "previous_outputs": previous_outputs,
-    }
-    if benchmark:
-        embedded["benchmark"] = benchmark
-
-    data_json = json.dumps(embedded)
-
-    return template.replace("/*__EMBEDDED_DATA__*/", f"const EMBEDDED_DATA = {data_json};")
-
-
-# ---------------------------------------------------------------------------
-# HTTP server (stdlib only, zero dependencies)
-# ---------------------------------------------------------------------------
-
-def _kill_port(port: int) -> None:
-    """Kill any process listening on the given port."""
-    try:
-        result = subprocess.run(
-            ["lsof", "-ti", f":{port}"],
-            capture_output=True, text=True, timeout=5,
-        )
-        for pid_str in result.stdout.strip().split("\n"):
-            if pid_str.strip():
-                try:
-                    os.kill(int(pid_str.strip()), signal.SIGTERM)
-                except (ProcessLookupError, ValueError):
-                    pass
-        if result.stdout.strip():
-            time.sleep(0.5)
-    except subprocess.TimeoutExpired:
-        pass
-    except FileNotFoundError:
-        print("Note: lsof not found, cannot check if port is in use", file=sys.stderr)
-
-class ReviewHandler(BaseHTTPRequestHandler):
-    """Serves the review HTML and handles feedback saves.
-
-    Regenerates the HTML on each page load so that refreshing the browser
-    picks up new eval outputs without restarting the server.
-    """
-
-    def __init__(
-        self,
-        workspace: Path,
-        skill_name: str,
-        feedback_path: Path,
-        previous: dict[str, dict],
-        benchmark_path: Path | None,
-        *args,
-        **kwargs,
-    ):
-        self.workspace = workspace
-        self.skill_name = skill_name
-        self.feedback_path = feedback_path
-        self.previous = previous
-        self.benchmark_path = benchmark_path
-        super().__init__(*args, **kwargs)
-
-    def do_GET(self) -> None:
-        if self.path == "/" or self.path == "/index.html":
-            # Regenerate HTML on each request (re-scans workspace for new outputs)
-            runs = find_runs(self.workspace)
-            benchmark = None
-            if self.benchmark_path and self.benchmark_path.exists():
-                try:
-                    benchmark = json.loads(self.benchmark_path.read_text())
-                except (json.JSONDecodeError, OSError):
-                    pass
-            html = generate_html(runs, self.skill_name, self.previous, benchmark)
-            content = html.encode("utf-8")
-            self.send_response(200)
-            self.send_header("Content-Type", "text/html; charset=utf-8")
-            self.send_header("Content-Length", str(len(content)))
-            self.end_headers()
-            self.wfile.write(content)
-        elif self.path == "/api/feedback":
-            data = b"{}"
-            if self.feedback_path.exists():
-                data = self.feedback_path.read_bytes()
-            self.send_response(200)
-            self.send_header("Content-Type", "application/json")
-            self.send_header("Content-Length", str(len(data)))
-            self.end_headers()
-            self.wfile.write(data)
-        else:
-            self.send_error(404)
-
-    def do_POST(self) -> None:
-        if self.path == "/api/feedback":
-            length = int(self.headers.get("Content-Length", 0))
-            body = self.rfile.read(length)
-            try:
-                data = json.loads(body)
-                if not isinstance(data, dict) or "reviews" not in data:
-                    raise ValueError("Expected JSON object with 'reviews' key")
-                self.feedback_path.write_text(json.dumps(data, indent=2) + "\n")
-                resp = b'{"ok":true}'
-                self.send_response(200)
-            except (json.JSONDecodeError, OSError, ValueError) as e:
-                resp = json.dumps({"error": str(e)}).encode()
-                self.send_response(500)
-            self.send_header("Content-Type", "application/json")
-            self.send_header("Content-Length", str(len(resp)))
-            self.end_headers()
-            self.wfile.write(resp)
-        else:
-            self.send_error(404)
-
-    def log_message(self, format: str, *args: object) -> None:
-        # Suppress request logging to keep terminal clean
-        pass
-
-
-def main() -> None:
-    parser = argparse.ArgumentParser(description="Generate and serve eval review")
-    parser.add_argument("workspace", type=Path, help="Path to workspace directory")
-    parser.add_argument("--port", "-p", type=int, default=3117, help="Server port (default: 3117)")
-    parser.add_argument("--skill-name", "-n", type=str, default=None, help="Skill name for header")
-    parser.add_argument(
-        "--previous-workspace", type=Path, default=None,
-        help="Path to previous iteration's workspace (shows old outputs and feedback as context)",
-    )
-    parser.add_argument(
-        "--benchmark", type=Path, default=None,
-        help="Path to benchmark.json to show in the Benchmark tab",
-    )
-    parser.add_argument(
-        "--static", "-s", type=Path, default=None,
-        help="Write standalone HTML to this path instead of starting a server",
-    )
-    args = parser.parse_args()
-
-    workspace = args.workspace.resolve()
-    if not workspace.is_dir():
-        print(f"Error: {workspace} is not a directory", file=sys.stderr)
-        sys.exit(1)
-
-    runs = find_runs(workspace)
-    if not runs:
-        print(f"No runs found in {workspace}", file=sys.stderr)
-        sys.exit(1)
-
-    skill_name = args.skill_name or workspace.name.replace("-workspace", "")
-    feedback_path = workspace / "feedback.json"
-
-    previous: dict[str, dict] = {}
-    if args.previous_workspace:
-        previous = load_previous_iteration(args.previous_workspace.resolve())
-
-    benchmark_path = args.benchmark.resolve() if args.benchmark else None
-    benchmark = None
-    if benchmark_path and benchmark_path.exists():
-        try:
-            benchmark = json.loads(benchmark_path.read_text())
-        except (json.JSONDecodeError, OSError):
-            pass
-
-    if args.static:
-        html = generate_html(runs, skill_name, previous, benchmark)
-        args.static.parent.mkdir(parents=True, exist_ok=True)
-        args.static.write_text(html)
-        print(f"\n  Static viewer written to: {args.static}\n")
-        sys.exit(0)
-
-    # Kill any existing process on the target port
-    port = args.port
-    _kill_port(port)
-    handler = partial(ReviewHandler, workspace, skill_name, feedback_path, previous, benchmark_path)
-    try:
-        server = HTTPServer(("127.0.0.1", port), handler)
-    except OSError:
-        # Port still in use after kill attempt — find a free one
-        server = HTTPServer(("127.0.0.1", 0), handler)
-        port = server.server_address[1]
-
-    url = f"http://localhost:{port}"
-    print(f"\n  Eval Viewer")
-    print(f"  ─────────────────────────────────")
-    print(f"  URL:       {url}")
-    print(f"  Workspace: {workspace}")
-    print(f"  Feedback:  {feedback_path}")
-    if previous:
-        print(f"  Previous:  {args.previous_workspace} ({len(previous)} runs)")
-    if benchmark_path:
-        print(f"  Benchmark: {benchmark_path}")
-    print(f"\n  Press Ctrl+C to stop.\n")
-
-    webbrowser.open(url)
-
-    try:
-        server.serve_forever()
-    except KeyboardInterrupt:
-        print("\nStopped.")
-        server.server_close()
-
-
-if __name__ == "__main__":
-    main()
diff --git a/.github/skills/skill-creator/eval-viewer/viewer.html b/.github/skills/skill-creator/eval-viewer/viewer.html
deleted file mode 100644
index 6d8e963..0000000
--- a/.github/skills/skill-creator/eval-viewer/viewer.html
+++ /dev/null
@@ -1,1325 +0,0 @@
-<!DOCTYPE html>
-<html lang="en">
-<head>
-  <meta charset="UTF-8">
-  <meta name="viewport" content="width=device-width, initial-scale=1.0">
-  <title>Eval Review</title>
-  <link rel="preconnect" href="https://fonts.googleapis.com">
-  <link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
-  <link href="https://fonts.googleapis.com/css2?family=Poppins:wght@500;600&family=Lora:wght@400;500&display=swap" rel="stylesheet">
-  <script src="https://cdn.sheetjs.com/xlsx-0.20.3/package/dist/xlsx.full.min.js" integrity="sha384-EnyY0/GSHQGSxSgMwaIPzSESbqoOLSexfnSMN2AP+39Ckmn92stwABZynq1JyzdT" crossorigin="anonymous"></script>
-  <style>
-    :root {
-      --bg: #faf9f5;
-      --surface: #ffffff;
-      --border: #e8e6dc;
-      --text: #141413;
-      --text-muted: #b0aea5;
-      --accent: #d97757;
-      --accent-hover: #c4613f;
-      --green: #788c5d;
-      --green-bg: #eef2e8;
-      --red: #c44;
-      --red-bg: #fceaea;
-      --header-bg: #141413;
-      --header-text: #faf9f5;
-      --radius: 6px;
-    }
-
-    * { box-sizing: border-box; margin: 0; padding: 0; }
-
-    body {
-      font-family: 'Lora', Georgia, serif;
-      background: var(--bg);
-      color: var(--text);
-      height: 100vh;
-      display: flex;
-      flex-direction: column;
-    }
-
-    /* ---- Header ---- */
-    .header {
-      background: var(--header-bg);
-      color: var(--header-text);
-      padding: 1rem 2rem;
-      display: flex;
-      justify-content: space-between;
-      align-items: center;
-      flex-shrink: 0;
-    }
-    .header h1 {
-      font-family: 'Poppins', sans-serif;
-      font-size: 1.25rem;
-      font-weight: 600;
-    }
-    .header .instructions {
-      font-size: 0.8rem;
-      opacity: 0.7;
-      margin-top: 0.25rem;
-    }
-    .header .progress {
-      font-size: 0.875rem;
-      opacity: 0.8;
-      text-align: right;
-    }
-
-    /* ---- Main content ---- */
-    .main {
-      flex: 1;
-      overflow-y: auto;
-      padding: 1.5rem 2rem;
-      display: flex;
-      flex-direction: column;
-      gap: 1.25rem;
-    }
-
-    /* ---- Sections ---- */
-    .section {
-      background: var(--surface);
-      border: 1px solid var(--border);
-      border-radius: var(--radius);
-      flex-shrink: 0;
-    }
-    .section-header {
-      font-family: 'Poppins', sans-serif;
-      padding: 0.75rem 1rem;
-      font-size: 0.75rem;
-      font-weight: 500;
-      text-transform: uppercase;
-      letter-spacing: 0.05em;
-      color: var(--text-muted);
-      border-bottom: 1px solid var(--border);
-      background: var(--bg);
-    }
-    .section-body {
-      padding: 1rem;
-    }
-
-    /* ---- Config badge ---- */
-    .config-badge {
-      display: inline-block;
-      padding: 0.2rem 0.625rem;
-      border-radius: 9999px;
-      font-family: 'Poppins', sans-serif;
-      font-size: 0.6875rem;
-      font-weight: 600;
-      text-transform: uppercase;
-      letter-spacing: 0.03em;
-      margin-left: 0.75rem;
-      vertical-align: middle;
-    }
-    .config-badge.config-primary {
-      background: rgba(33, 150, 243, 0.12);
-      color: #1976d2;
-    }
-    .config-badge.config-baseline {
-      background: rgba(255, 193, 7, 0.15);
-      color: #f57f17;
-    }
-
-    /* ---- Prompt ---- */
-    .prompt-text {
-      white-space: pre-wrap;
-      font-size: 0.9375rem;
-      line-height: 1.6;
-    }
-
-    /* ---- Outputs ---- */
-    .output-file {
-      border: 1px solid var(--border);
-      border-radius: var(--radius);
-      overflow: hidden;
-    }
-    .output-file + .output-file {
-      margin-top: 1rem;
-    }
-    .output-file-header {
-      padding: 0.5rem 0.75rem;
-      font-size: 0.8rem;
-      font-weight: 600;
-      color: var(--text-muted);
-      background: var(--bg);
-      border-bottom: 1px solid var(--border);
-      font-family: 'SF Mono', SFMono-Regular, Consolas, 'Liberation Mono', Menlo, monospace;
-      display: flex;
-      justify-content: space-between;
-      align-items: center;
-    }
-    .output-file-header .dl-btn {
-      font-size: 0.7rem;
-      color: var(--accent);
-      text-decoration: none;
-      cursor: pointer;
-      font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', sans-serif;
-      font-weight: 500;
-      opacity: 0.8;
-    }
-    .output-file-header .dl-btn:hover {
-      opacity: 1;
-      text-decoration: underline;
-    }
-    .output-file-content {
-      padding: 0.75rem;
-      overflow-x: auto;
-    }
-    .output-file-content pre {
-      font-size: 0.8125rem;
-      line-height: 1.5;
-      white-space: pre-wrap;
-      word-break: break-word;
-      font-family: 'SF Mono', SFMono-Regular, Consolas, 'Liberation Mono', Menlo, monospace;
-    }
-    .output-file-content img {
-      max-width: 100%;
-      height: auto;
-      border-radius: 4px;
-    }
-    .output-file-content iframe {
-      width: 100%;
-      height: 600px;
-      border: none;
-    }
-    .output-file-content table {
-      border-collapse: collapse;
-      font-size: 0.8125rem;
-      width: 100%;
-    }
-    .output-file-content table td,
-    .output-file-content table th {
-      border: 1px solid var(--border);
-      padding: 0.375rem 0.5rem;
-      text-align: left;
-    }
-    .output-file-content table th {
-      background: var(--bg);
-      font-weight: 600;
-    }
-    .output-file-content .download-link {
-      display: inline-flex;
-      align-items: center;
-      gap: 0.5rem;
-      padding: 0.5rem 1rem;
-      background: var(--bg);
-      border: 1px solid var(--border);
-      border-radius: 4px;
-      color: var(--accent);
-      text-decoration: none;
-      font-size: 0.875rem;
-      cursor: pointer;
-    }
-    .output-file-content .download-link:hover {
-      background: var(--border);
-    }
-    .empty-state {
-      color: var(--text-muted);
-      font-style: italic;
-      padding: 2rem;
-      text-align: center;
-    }
-
-    /* ---- Feedback ---- */
-    .prev-feedback {
-      background: var(--bg);
-      border: 1px solid var(--border);
-      border-radius: 4px;
-      padding: 0.625rem 0.75rem;
-      margin-top: 0.75rem;
-      font-size: 0.8125rem;
-      color: var(--text-muted);
-      line-height: 1.5;
-    }
-    .prev-feedback-label {
-      font-size: 0.7rem;
-      font-weight: 600;
-      text-transform: uppercase;
-      letter-spacing: 0.04em;
-      margin-bottom: 0.25rem;
-      color: var(--text-muted);
-    }
-    .feedback-textarea {
-      width: 100%;
-      min-height: 100px;
-      padding: 0.75rem;
-      border: 1px solid var(--border);
-      border-radius: 4px;
-      font-family: inherit;
-      font-size: 0.9375rem;
-      line-height: 1.5;
-      resize: vertical;
-      color: var(--text);
-    }
-    .feedback-textarea:focus {
-      outline: none;
-      border-color: var(--accent);
-      box-shadow: 0 0 0 3px rgba(37, 99, 235, 0.1);
-    }
-    .feedback-status {
-      font-size: 0.75rem;
-      color: var(--text-muted);
-      margin-top: 0.5rem;
-      min-height: 1.1em;
-    }
-
-    /* ---- Grades (collapsible) ---- */
-    .grades-toggle {
-      display: flex;
-      align-items: center;
-      cursor: pointer;
-      user-select: none;
-    }
-    .grades-toggle:hover {
-      color: var(--accent);
-    }
-    .grades-toggle .arrow {
-      margin-right: 0.5rem;
-      transition: transform 0.15s;
-      font-size: 0.75rem;
-    }
-    .grades-toggle .arrow.open {
-      transform: rotate(90deg);
-    }
-    .grades-content {
-      display: none;
-      margin-top: 0.75rem;
-    }
-    .grades-content.open {
-      display: block;
-    }
-    .grades-summary {
-      font-size: 0.875rem;
-      margin-bottom: 0.75rem;
-      display: flex;
-      align-items: center;
-      gap: 0.5rem;
-    }
-    .grade-badge {
-      display: inline-block;
-      padding: 0.125rem 0.5rem;
-      border-radius: 9999px;
-      font-size: 0.75rem;
-      font-weight: 600;
-    }
-    .grade-pass { background: var(--green-bg); color: var(--green); }
-    .grade-fail { background: var(--red-bg); color: var(--red); }
-    .assertion-list {
-      list-style: none;
-    }
-    .assertion-item {
-      padding: 0.625rem 0;
-      border-bottom: 1px solid var(--border);
-      font-size: 0.8125rem;
-    }
-    .assertion-item:last-child { border-bottom: none; }
-    .assertion-status {
-      font-weight: 600;
-      margin-right: 0.5rem;
-    }
-    .assertion-status.pass { color: var(--green); }
-    .assertion-status.fail { color: var(--red); }
-    .assertion-evidence {
-      color: var(--text-muted);
-      font-size: 0.75rem;
-      margin-top: 0.25rem;
-      padding-left: 1.5rem;
-    }
-
-    /* ---- View tabs ---- */
-    .view-tabs {
-      display: flex;
-      gap: 0;
-      padding: 0 2rem;
-      background: var(--bg);
-      border-bottom: 1px solid var(--border);
-      flex-shrink: 0;
-    }
-    .view-tab {
-      font-family: 'Poppins', sans-serif;
-      padding: 0.625rem 1.25rem;
-      font-size: 0.8125rem;
-      font-weight: 500;
-      cursor: pointer;
-      border: none;
-      background: none;
-      color: var(--text-muted);
-      border-bottom: 2px solid transparent;
-      transition: all 0.15s;
-    }
-    .view-tab:hover { color: var(--text); }
-    .view-tab.active {
-      color: var(--accent);
-      border-bottom-color: var(--accent);
-    }
-    .view-panel { display: none; }
-    .view-panel.active { display: flex; flex-direction: column; flex: 1; overflow: hidden; }
-
-    /* ---- Benchmark view ---- */
-    .benchmark-view {
-      padding: 1.5rem 2rem;
-      overflow-y: auto;
-      flex: 1;
-    }
-    .benchmark-table {
-      border-collapse: collapse;
-      background: var(--surface);
-      border: 1px solid var(--border);
-      border-radius: var(--radius);
-      font-size: 0.8125rem;
-      width: 100%;
-      margin-bottom: 1.5rem;
-    }
-    .benchmark-table th, .benchmark-table td {
-      padding: 0.625rem 0.75rem;
-      text-align: left;
-      border: 1px solid var(--border);
-    }
-    .benchmark-table th {
-      font-family: 'Poppins', sans-serif;
-      background: var(--header-bg);
-      color: var(--header-text);
-      font-weight: 500;
-      font-size: 0.75rem;
-      text-transform: uppercase;
-      letter-spacing: 0.04em;
-    }
-    .benchmark-table tr:hover { background: var(--bg); }
-    .benchmark-table tr.benchmark-row-with { background: rgba(33, 150, 243, 0.06); }
-    .benchmark-table tr.benchmark-row-without { background: rgba(255, 193, 7, 0.06); }
-    .benchmark-table tr.benchmark-row-with:hover { background: rgba(33, 150, 243, 0.12); }
-    .benchmark-table tr.benchmark-row-without:hover { background: rgba(255, 193, 7, 0.12); }
-    .benchmark-table tr.benchmark-row-avg { font-weight: 600; border-top: 2px solid var(--border); }
-    .benchmark-table tr.benchmark-row-avg.benchmark-row-with { background: rgba(33, 150, 243, 0.12); }
-    .benchmark-table tr.benchmark-row-avg.benchmark-row-without { background: rgba(255, 193, 7, 0.12); }
-    .benchmark-delta-positive { color: var(--green); font-weight: 600; }
-    .benchmark-delta-negative { color: var(--red); font-weight: 600; }
-    .benchmark-notes {
-      background: var(--surface);
-      border: 1px solid var(--border);
-      border-radius: var(--radius);
-      padding: 1rem;
-    }
-    .benchmark-notes h3 {
-      font-family: 'Poppins', sans-serif;
-      font-size: 0.875rem;
-      margin-bottom: 0.75rem;
-    }
-    .benchmark-notes ul {
-      list-style: disc;
-      padding-left: 1.25rem;
-    }
-    .benchmark-notes li {
-      font-size: 0.8125rem;
-      line-height: 1.6;
-      margin-bottom: 0.375rem;
-    }
-    .benchmark-empty {
-      color: var(--text-muted);
-      font-style: italic;
-      text-align: center;
-      padding: 3rem;
-    }
-
-    /* ---- Navigation ---- */
-    .nav {
-      display: flex;
-      justify-content: space-between;
-      align-items: center;
-      padding: 1rem 2rem;
-      border-top: 1px solid var(--border);
-      background: var(--surface);
-      flex-shrink: 0;
-    }
-    .nav-btn {
-      font-family: 'Poppins', sans-serif;
-      padding: 0.5rem 1.25rem;
-      border: 1px solid var(--border);
-      border-radius: var(--radius);
-      background: var(--surface);
-      cursor: pointer;
-      font-size: 0.875rem;
-      font-weight: 500;
-      color: var(--text);
-      transition: all 0.15s;
-    }
-    .nav-btn:hover:not(:disabled) {
-      background: var(--bg);
-      border-color: var(--text-muted);
-    }
-    .nav-btn:disabled {
-      opacity: 0.4;
-      cursor: not-allowed;
-    }
-    .done-btn {
-      font-family: 'Poppins', sans-serif;
-      padding: 0.5rem 1.5rem;
-      border: 1px solid var(--border);
-      border-radius: var(--radius);
-      background: var(--surface);
-      color: var(--text);
-      cursor: pointer;
-      font-size: 0.875rem;
-      font-weight: 500;
-      transition: all 0.15s;
-    }
-    .done-btn:hover {
-      background: var(--bg);
-      border-color: var(--text-muted);
-    }
-    .done-btn.ready {
-      border: none;
-      background: var(--accent);
-      color: white;
-      font-weight: 600;
-    }
-    .done-btn.ready:hover {
-      background: var(--accent-hover);
-    }
-    /* ---- Done overlay ---- */
-    .done-overlay {
-      display: none;
-      position: fixed;
-      inset: 0;
-      background: rgba(0, 0, 0, 0.5);
-      z-index: 100;
-      justify-content: center;
-      align-items: center;
-    }
-    .done-overlay.visible {
-      display: flex;
-    }
-    .done-card {
-      background: var(--surface);
-      border-radius: 12px;
-      padding: 2rem 3rem;
-      text-align: center;
-      box-shadow: 0 20px 60px rgba(0, 0, 0, 0.3);
-      max-width: 500px;
-    }
-    .done-card h2 {
-      font-size: 1.5rem;
-      margin-bottom: 0.5rem;
-    }
-    .done-card p {
-      color: var(--text-muted);
-      margin-bottom: 1.5rem;
-      line-height: 1.5;
-    }
-    .done-card .btn-row {
-      display: flex;
-      gap: 0.5rem;
-      justify-content: center;
-    }
-    .done-card button {
-      padding: 0.5rem 1.25rem;
-      border: 1px solid var(--border);
-      border-radius: var(--radius);
-      background: var(--surface);
-      cursor: pointer;
-      font-size: 0.875rem;
-    }
-    .done-card button:hover {
-      background: var(--bg);
-    }
-    /* ---- Toast ---- */
-    .toast {
-      position: fixed;
-      bottom: 5rem;
-      left: 50%;
-      transform: translateX(-50%);
-      background: var(--header-bg);
-      color: var(--header-text);
-      padding: 0.625rem 1.25rem;
-      border-radius: var(--radius);
-      font-size: 0.875rem;
-      opacity: 0;
-      transition: opacity 0.3s;
-      pointer-events: none;
-      z-index: 200;
-    }
-    .toast.visible {
-      opacity: 1;
-    }
-  </style>
-</head>
-<body>
-  <div id="app" style="height:100vh; display:flex; flex-direction:column;">
-    <div class="header">
-      <div>
-        <h1>Eval Review: <span id="skill-name"></span></h1>
-        <div class="instructions">Review each output and leave feedback below. Navigate with arrow keys or buttons. When done, copy feedback and paste into Claude Code.</div>
-      </div>
-      <div class="progress" id="progress"></div>
-    </div>
-
-    <!-- View tabs (only shown when benchmark data exists) -->
-    <div class="view-tabs" id="view-tabs" style="display:none;">
-      <button class="view-tab active" onclick="switchView('outputs')">Outputs</button>
-      <button class="view-tab" onclick="switchView('benchmark')">Benchmark</button>
-    </div>
-
-    <!-- Outputs panel (qualitative review) -->
-    <div class="view-panel active" id="panel-outputs">
-    <div class="main">
-      <!-- Prompt -->
-      <div class="section">
-        <div class="section-header">Prompt <span class="config-badge" id="config-badge" style="display:none;"></span></div>
-        <div class="section-body">
-          <div class="prompt-text" id="prompt-text"></div>
-        </div>
-      </div>
-
-      <!-- Outputs -->
-      <div class="section">
-        <div class="section-header">Output</div>
-        <div class="section-body" id="outputs-body">
-          <div class="empty-state">No output files found</div>
-        </div>
-      </div>
-
-      <!-- Previous Output (collapsible) -->
-      <div class="section" id="prev-outputs-section" style="display:none;">
-        <div class="section-header">
-          <div class="grades-toggle" onclick="togglePrevOutputs()">
-            <span class="arrow" id="prev-outputs-arrow">&#9654;</span>
-            Previous Output
-          </div>
-        </div>
-        <div class="grades-content" id="prev-outputs-content"></div>
-      </div>
-
-      <!-- Grades (collapsible) -->
-      <div class="section" id="grades-section" style="display:none;">
-        <div class="section-header">
-          <div class="grades-toggle" onclick="toggleGrades()">
-            <span class="arrow" id="grades-arrow">&#9654;</span>
-            Formal Grades
-          </div>
-        </div>
-        <div class="grades-content" id="grades-content"></div>
-      </div>
-
-      <!-- Feedback -->
-      <div class="section">
-        <div class="section-header">Your Feedback</div>
-        <div class="section-body">
-          <textarea
-            class="feedback-textarea"
-            id="feedback"
-            placeholder="What do you think of this output? Any issues, suggestions, or things that look great?"
-          ></textarea>
-          <div class="feedback-status" id="feedback-status"></div>
-          <div class="prev-feedback" id="prev-feedback" style="display:none;">
-            <div class="prev-feedback-label">Previous feedback</div>
-            <div id="prev-feedback-text"></div>
-          </div>
-        </div>
-      </div>
-    </div>
-
-    <div class="nav" id="outputs-nav">
-      <button class="nav-btn" id="prev-btn" onclick="navigate(-1)">&#8592; Previous</button>
-      <button class="done-btn" id="done-btn" onclick="showDoneDialog()">Submit All Reviews</button>
-      <button class="nav-btn" id="next-btn" onclick="navigate(1)">Next &#8594;</button>
-    </div>
-    </div><!-- end panel-outputs -->
-
-    <!-- Benchmark panel (quantitative stats) -->
-    <div class="view-panel" id="panel-benchmark">
-      <div class="benchmark-view" id="benchmark-content">
-        <div class="benchmark-empty">No benchmark data available. Run a benchmark to see quantitative results here.</div>
-      </div>
-    </div>
-  </div>
-
-  <!-- Done overlay -->
-  <div class="done-overlay" id="done-overlay">
-    <div class="done-card">
-      <h2>Review Complete</h2>
-      <p>Your feedback has been saved. Go back to your Claude Code session and tell Claude you're done reviewing.</p>
-      <div class="btn-row">
-        <button onclick="closeDoneDialog()">OK</button>
-      </div>
-    </div>
-  </div>
-
-  <!-- Toast -->
-  <div class="toast" id="toast"></div>
-
-  <script>
-    // ---- Embedded data (injected by generate_review.py) ----
-    /*__EMBEDDED_DATA__*/
-
-    // ---- State ----
-    let feedbackMap = {};  // run_id -> feedback text
-    let currentIndex = 0;
-    let visitedRuns = new Set();
-
-    // ---- Init ----
-    async function init() {
-      // Load saved feedback from server — but only if this isn't a fresh
-      // iteration (indicated by previous_feedback being present). When
-      // previous feedback exists, the feedback.json on disk is stale from
-      // the prior iteration and should not pre-fill the textareas.
-      const hasPrevious = Object.keys(EMBEDDED_DATA.previous_feedback || {}).length > 0
-        || Object.keys(EMBEDDED_DATA.previous_outputs || {}).length > 0;
-      if (!hasPrevious) {
-        try {
-          const resp = await fetch("/api/feedback");
-          const data = await resp.json();
-          if (data.reviews) {
-            for (const r of data.reviews) feedbackMap[r.run_id] = r.feedback;
-          }
-        } catch { /* first run, no feedback yet */ }
-      }
-
-      document.getElementById("skill-name").textContent = EMBEDDED_DATA.skill_name;
-      showRun(0);
-
-      // Wire up feedback auto-save
-      const textarea = document.getElementById("feedback");
-      let saveTimeout = null;
-      textarea.addEventListener("input", () => {
-        clearTimeout(saveTimeout);
-        document.getElementById("feedback-status").textContent = "";
-        saveTimeout = setTimeout(() => saveCurrentFeedback(), 800);
-      });
-    }
-
-    // ---- Navigation ----
-    function navigate(delta) {
-      const newIndex = currentIndex + delta;
-      if (newIndex >= 0 && newIndex < EMBEDDED_DATA.runs.length) {
-        saveCurrentFeedback();
-        showRun(newIndex);
-      }
-    }
-
-    function updateNavButtons() {
-      document.getElementById("prev-btn").disabled = currentIndex === 0;
-      document.getElementById("next-btn").disabled =
-        currentIndex === EMBEDDED_DATA.runs.length - 1;
-    }
-
-    // ---- Show a run ----
-    function showRun(index) {
-      currentIndex = index;
-      const run = EMBEDDED_DATA.runs[index];
-
-      // Progress
-      document.getElementById("progress").textContent =
-        `${index + 1} of ${EMBEDDED_DATA.runs.length}`;
-
-      // Prompt
-      document.getElementById("prompt-text").textContent = run.prompt;
-
-      // Config badge
-      const badge = document.getElementById("config-badge");
-      const configMatch = run.id.match(/(with_skill|without_skill|new_skill|old_skill)/);
-      if (configMatch) {
-        const config = configMatch[1];
-        const isBaseline = config === "without_skill" || config === "old_skill";
-        badge.textContent = config.replace(/_/g, " ");
-        badge.className = "config-badge " + (isBaseline ? "config-baseline" : "config-primary");
-        badge.style.display = "inline-block";
-      } else {
-        badge.style.display = "none";
-      }
-
-      // Outputs
-      renderOutputs(run);
-
-      // Previous outputs
-      renderPrevOutputs(run);
-
-      // Grades
-      renderGrades(run);
-
-      // Previous feedback
-      const prevFb = (EMBEDDED_DATA.previous_feedback || {})[run.id];
-      const prevEl = document.getElementById("prev-feedback");
-      if (prevFb) {
-        document.getElementById("prev-feedback-text").textContent = prevFb;
-        prevEl.style.display = "block";
-      } else {
-        prevEl.style.display = "none";
-      }
-
-      // Feedback
-      document.getElementById("feedback").value = feedbackMap[run.id] || "";
-      document.getElementById("feedback-status").textContent = "";
-
-      updateNavButtons();
-
-      // Track visited runs and promote done button when all visited
-      visitedRuns.add(index);
-      const doneBtn = document.getElementById("done-btn");
-      if (visitedRuns.size >= EMBEDDED_DATA.runs.length) {
-        doneBtn.classList.add("ready");
-      }
-
-      // Scroll main content to top
-      document.querySelector(".main").scrollTop = 0;
-    }
-
-    // ---- Render outputs ----
-    function renderOutputs(run) {
-      const container = document.getElementById("outputs-body");
-      container.innerHTML = "";
-
-      const outputs = run.outputs || [];
-      if (outputs.length === 0) {
-        container.innerHTML = '<div class="empty-state">No output files</div>';
-        return;
-      }
-
-      for (const file of outputs) {
-        const fileDiv = document.createElement("div");
-        fileDiv.className = "output-file";
-
-        // Always show file header with download link
-        const header = document.createElement("div");
-        header.className = "output-file-header";
-        const nameSpan = document.createElement("span");
-        nameSpan.textContent = file.name;
-        header.appendChild(nameSpan);
-        const dlBtn = document.createElement("a");
-        dlBtn.className = "dl-btn";
-        dlBtn.textContent = "Download";
-        dlBtn.download = file.name;
-        dlBtn.href = getDownloadUri(file);
-        header.appendChild(dlBtn);
-        fileDiv.appendChild(header);
-
-        const content = document.createElement("div");
-        content.className = "output-file-content";
-
-        if (file.type === "text") {
-          const pre = document.createElement("pre");
-          pre.textContent = file.content;
-          content.appendChild(pre);
-        } else if (file.type === "image") {
-          const img = document.createElement("img");
-          img.src = file.data_uri;
-          img.alt = file.name;
-          content.appendChild(img);
-        } else if (file.type === "pdf") {
-          const iframe = document.createElement("iframe");
-          iframe.src = file.data_uri;
-          content.appendChild(iframe);
-        } else if (file.type === "xlsx") {
-          renderXlsx(content, file.data_b64);
-        } else if (file.type === "binary") {
-          const a = document.createElement("a");
-          a.className = "download-link";
-          a.href = file.data_uri;
-          a.download = file.name;
-          a.textContent = "Download " + file.name;
-          content.appendChild(a);
-        } else if (file.type === "error") {
-          const pre = document.createElement("pre");
-          pre.textContent = file.content;
-          pre.style.color = "var(--red)";
-          content.appendChild(pre);
-        }
-
-        fileDiv.appendChild(content);
-        container.appendChild(fileDiv);
-      }
-    }
-
-    // ---- XLSX rendering via SheetJS ----
-    function renderXlsx(container, b64Data) {
-      try {
-        const raw = Uint8Array.from(atob(b64Data), c => c.charCodeAt(0));
-        const wb = XLSX.read(raw, { type: "array" });
-
-        for (let i = 0; i < wb.SheetNames.length; i++) {
-          const sheetName = wb.SheetNames[i];
-          const ws = wb.Sheets[sheetName];
-
-          if (wb.SheetNames.length > 1) {
-            const sheetLabel = document.createElement("div");
-            sheetLabel.style.cssText =
-              "font-weight:600; font-size:0.8rem; color:#b0aea5; margin-top:0.5rem; margin-bottom:0.25rem;";
-            sheetLabel.textContent = "Sheet: " + sheetName;
-            container.appendChild(sheetLabel);
-          }
-
-          const htmlStr = XLSX.utils.sheet_to_html(ws, { editable: false });
-          const wrapper = document.createElement("div");
-          wrapper.innerHTML = htmlStr;
-          container.appendChild(wrapper);
-        }
-      } catch (err) {
-        container.textContent = "Error rendering spreadsheet: " + err.message;
-      }
-    }
-
-    // ---- Grades ----
-    function renderGrades(run) {
-      const section = document.getElementById("grades-section");
-      const content = document.getElementById("grades-content");
-
-      if (!run.grading) {
-        section.style.display = "none";
-        return;
-      }
-
-      const grading = run.grading;
-      section.style.display = "block";
-      // Reset to collapsed
-      content.classList.remove("open");
-      document.getElementById("grades-arrow").classList.remove("open");
-
-      const summary = grading.summary || {};
-      const expectations = grading.expectations || [];
-
-      let html = '<div style="padding: 1rem;">';
-
-      // Summary line
-      const passRate = summary.pass_rate != null
-        ? Math.round(summary.pass_rate * 100) + "%"
-        : "?";
-      const badgeClass = summary.pass_rate >= 0.8 ? "grade-pass" : summary.pass_rate >= 0.5 ? "" : "grade-fail";
-      html += '<div class="grades-summary">';
-      html += '<span class="grade-badge ' + badgeClass + '">' + passRate + '</span>';
-      html += '<span>' + (summary.passed || 0) + ' passed, ' + (summary.failed || 0) + ' failed of ' + (summary.total || 0) + '</span>';
-      html += '</div>';
-
-      // Assertions list
-      html += '<ul class="assertion-list">';
-      for (const exp of expectations) {
-        const statusClass = exp.passed ? "pass" : "fail";
-        const statusIcon = exp.passed ? "\u2713" : "\u2717";
-        html += '<li class="assertion-item">';
-        html += '<span class="assertion-status ' + statusClass + '">' + statusIcon + '</span>';
-        html += '<span>' + escapeHtml(exp.text) + '</span>';
-        if (exp.evidence) {
-          html += '<div class="assertion-evidence">' + escapeHtml(exp.evidence) + '</div>';
-        }
-        html += '</li>';
-      }
-      html += '</ul>';
-
-      html += '</div>';
-      content.innerHTML = html;
-    }
-
-    function toggleGrades() {
-      const content = document.getElementById("grades-content");
-      const arrow = document.getElementById("grades-arrow");
-      content.classList.toggle("open");
-      arrow.classList.toggle("open");
-    }
-
-    // ---- Previous outputs (collapsible) ----
-    function renderPrevOutputs(run) {
-      const section = document.getElementById("prev-outputs-section");
-      const content = document.getElementById("prev-outputs-content");
-      const prevOutputs = (EMBEDDED_DATA.previous_outputs || {})[run.id];
-
-      if (!prevOutputs || prevOutputs.length === 0) {
-        section.style.display = "none";
-        return;
-      }
-
-      section.style.display = "block";
-      // Reset to collapsed
-      content.classList.remove("open");
-      document.getElementById("prev-outputs-arrow").classList.remove("open");
-
-      // Render the files into the content area
-      content.innerHTML = "";
-      const wrapper = document.createElement("div");
-      wrapper.style.padding = "1rem";
-
-      for (const file of prevOutputs) {
-        const fileDiv = document.createElement("div");
-        fileDiv.className = "output-file";
-
-        const header = document.createElement("div");
-        header.className = "output-file-header";
-        const nameSpan = document.createElement("span");
-        nameSpan.textContent = file.name;
-        header.appendChild(nameSpan);
-        const dlBtn = document.createElement("a");
-        dlBtn.className = "dl-btn";
-        dlBtn.textContent = "Download";
-        dlBtn.download = file.name;
-        dlBtn.href = getDownloadUri(file);
-        header.appendChild(dlBtn);
-        fileDiv.appendChild(header);
-
-        const fc = document.createElement("div");
-        fc.className = "output-file-content";
-
-        if (file.type === "text") {
-          const pre = document.createElement("pre");
-          pre.textContent = file.content;
-          fc.appendChild(pre);
-        } else if (file.type === "image") {
-          const img = document.createElement("img");
-          img.src = file.data_uri;
-          img.alt = file.name;
-          fc.appendChild(img);
-        } else if (file.type === "pdf") {
-          const iframe = document.createElement("iframe");
-          iframe.src = file.data_uri;
-          fc.appendChild(iframe);
-        } else if (file.type === "xlsx") {
-          renderXlsx(fc, file.data_b64);
-        } else if (file.type === "binary") {
-          const a = document.createElement("a");
-          a.className = "download-link";
-          a.href = file.data_uri;
-          a.download = file.name;
-          a.textContent = "Download " + file.name;
-          fc.appendChild(a);
-        }
-
-        fileDiv.appendChild(fc);
-        wrapper.appendChild(fileDiv);
-      }
-
-      content.appendChild(wrapper);
-    }
-
-    function togglePrevOutputs() {
-      const content = document.getElementById("prev-outputs-content");
-      const arrow = document.getElementById("prev-outputs-arrow");
-      content.classList.toggle("open");
-      arrow.classList.toggle("open");
-    }
-
-    // ---- Feedback (saved to server -> feedback.json) ----
-    function saveCurrentFeedback() {
-      const run = EMBEDDED_DATA.runs[currentIndex];
-      const text = document.getElementById("feedback").value;
-
-      if (text.trim() === "") {
-        delete feedbackMap[run.id];
-      } else {
-        feedbackMap[run.id] = text;
-      }
-
-      // Build reviews array from map
-      const reviews = [];
-      for (const [run_id, feedback] of Object.entries(feedbackMap)) {
-        if (feedback.trim()) {
-          reviews.push({ run_id, feedback, timestamp: new Date().toISOString() });
-        }
-      }
-
-      fetch("/api/feedback", {
-        method: "POST",
-        headers: { "Content-Type": "application/json" },
-        body: JSON.stringify({ reviews, status: "in_progress" }),
-      }).then(() => {
-        document.getElementById("feedback-status").textContent = "Saved";
-      }).catch(() => {
-        // Static mode or server unavailable — no-op on auto-save,
-        // feedback will be downloaded on final submit
-        document.getElementById("feedback-status").textContent = "Will download on submit";
-      });
-    }
-
-    // ---- Done ----
-    function showDoneDialog() {
-      // Save current textarea to feedbackMap (but don't POST yet)
-      const run = EMBEDDED_DATA.runs[currentIndex];
-      const text = document.getElementById("feedback").value;
-      if (text.trim() === "") {
-        delete feedbackMap[run.id];
-      } else {
-        feedbackMap[run.id] = text;
-      }
-
-      // POST once with status: complete — include ALL runs so the model
-      // can distinguish "no feedback" (looks good) from "not reviewed"
-      const reviews = [];
-      const ts = new Date().toISOString();
-      for (const r of EMBEDDED_DATA.runs) {
-        reviews.push({ run_id: r.id, feedback: feedbackMap[r.id] || "", timestamp: ts });
-      }
-      const payload = JSON.stringify({ reviews, status: "complete" }, null, 2);
-      fetch("/api/feedback", {
-        method: "POST",
-        headers: { "Content-Type": "application/json" },
-        body: payload,
-      }).then(() => {
-        document.getElementById("done-overlay").classList.add("visible");
-      }).catch(() => {
-        // Server not available (static mode) — download as file
-        const blob = new Blob([payload], { type: "application/json" });
-        const url = URL.createObjectURL(blob);
-        const a = document.createElement("a");
-        a.href = url;
-        a.download = "feedback.json";
-        a.click();
-        URL.revokeObjectURL(url);
-        document.getElementById("done-overlay").classList.add("visible");
-      });
-    }
-
-    function closeDoneDialog() {
-      // Reset status back to in_progress
-      saveCurrentFeedback();
-      document.getElementById("done-overlay").classList.remove("visible");
-    }
-
-    // ---- Toast ----
-    function showToast(message) {
-      const toast = document.getElementById("toast");
-      toast.textContent = message;
-      toast.classList.add("visible");
-      setTimeout(() => toast.classList.remove("visible"), 2000);
-    }
-
-    // ---- Keyboard nav ----
-    document.addEventListener("keydown", (e) => {
-      // Don't capture when typing in textarea
-      if (e.target.tagName === "TEXTAREA") return;
-
-      if (e.key === "ArrowLeft" || e.key === "ArrowUp") {
-        e.preventDefault();
-        navigate(-1);
-      } else if (e.key === "ArrowRight" || e.key === "ArrowDown") {
-        e.preventDefault();
-        navigate(1);
-      }
-    });
-
-    // ---- Util ----
-    function getDownloadUri(file) {
-      if (file.data_uri) return file.data_uri;
-      if (file.data_b64) return "data:application/octet-stream;base64," + file.data_b64;
-      if (file.type === "text") return "data:text/plain;charset=utf-8," + encodeURIComponent(file.content);
-      return "#";
-    }
-
-    function escapeHtml(text) {
-      const div = document.createElement("div");
-      div.textContent = text;
-      return div.innerHTML;
-    }
-
-    // ---- View switching ----
-    function switchView(view) {
-      document.querySelectorAll(".view-tab").forEach(t => t.classList.remove("active"));
-      document.querySelectorAll(".view-panel").forEach(p => p.classList.remove("active"));
-      document.querySelector(`[onclick="switchView('${view}')"]`).classList.add("active");
-      document.getElementById("panel-" + view).classList.add("active");
-    }
-
-    // ---- Benchmark rendering ----
-    function renderBenchmark() {
-      const data = EMBEDDED_DATA.benchmark;
-      if (!data) return;
-
-      // Show the tabs
-      document.getElementById("view-tabs").style.display = "flex";
-
-      const container = document.getElementById("benchmark-content");
-      const summary = data.run_summary || {};
-      const metadata = data.metadata || {};
-      const notes = data.notes || [];
-
-      let html = "";
-
-      // Header
-      html += "<h2 style='font-family: Poppins, sans-serif; margin-bottom: 0.5rem;'>Benchmark Results</h2>";
-      html += "<p style='color: var(--text-muted); font-size: 0.875rem; margin-bottom: 1.25rem;'>";
-      if (metadata.skill_name) html += "<strong>" + escapeHtml(metadata.skill_name) + "</strong> &mdash; ";
-      if (metadata.timestamp) html += metadata.timestamp + " &mdash; ";
-      if (metadata.evals_run) html += "Evals: " + metadata.evals_run.join(", ") + " &mdash; ";
-      html += (metadata.runs_per_configuration || "?") + " runs per configuration";
-      html += "</p>";
-
-      // Summary table
-      html += '<table class="benchmark-table">';
-
-      function fmtStat(stat, pct) {
-        if (!stat) return "—";
-        const suffix = pct ? "%" : "";
-        const m = pct ? (stat.mean * 100).toFixed(0) : stat.mean.toFixed(1);
-        const s = pct ? (stat.stddev * 100).toFixed(0) : stat.stddev.toFixed(1);
-        return m + suffix + " ± " + s + suffix;
-      }
-
-      function deltaClass(val) {
-        if (!val) return "";
-        const n = parseFloat(val);
-        if (n > 0) return "benchmark-delta-positive";
-        if (n < 0) return "benchmark-delta-negative";
-        return "";
-      }
-
-      // Discover config names dynamically (everything except "delta")
-      const configs = Object.keys(summary).filter(k => k !== "delta");
-      const configA = configs[0] || "config_a";
-      const configB = configs[1] || "config_b";
-      const labelA = configA.replace(/_/g, " ").replace(/\b\w/g, c => c.toUpperCase());
-      const labelB = configB.replace(/_/g, " ").replace(/\b\w/g, c => c.toUpperCase());
-      const a = summary[configA] || {};
-      const b = summary[configB] || {};
-      const delta = summary.delta || {};
-
-      html += "<thead><tr><th>Metric</th><th>" + escapeHtml(labelA) + "</th><th>" + escapeHtml(labelB) + "</th><th>Delta</th></tr></thead>";
-      html += "<tbody>";
-
-      html += "<tr><td><strong>Pass Rate</strong></td>";
-      html += "<td>" + fmtStat(a.pass_rate, true) + "</td>";
-      html += "<td>" + fmtStat(b.pass_rate, true) + "</td>";
-      html += '<td class="' + deltaClass(delta.pass_rate) + '">' + (delta.pass_rate || "—") + "</td></tr>";
-
-      // Time (only show row if data exists)
-      if (a.time_seconds || b.time_seconds) {
-        html += "<tr><td><strong>Time (s)</strong></td>";
-        html += "<td>" + fmtStat(a.time_seconds, false) + "</td>";
-        html += "<td>" + fmtStat(b.time_seconds, false) + "</td>";
-        html += '<td class="' + deltaClass(delta.time_seconds) + '">' + (delta.time_seconds ? delta.time_seconds + "s" : "—") + "</td></tr>";
-      }
-
-      // Tokens (only show row if data exists)
-      if (a.tokens || b.tokens) {
-        html += "<tr><td><strong>Tokens</strong></td>";
-        html += "<td>" + fmtStat(a.tokens, false) + "</td>";
-        html += "<td>" + fmtStat(b.tokens, false) + "</td>";
-        html += '<td class="' + deltaClass(delta.tokens) + '">' + (delta.tokens || "—") + "</td></tr>";
-      }
-
-      html += "</tbody></table>";
-
-      // Per-eval breakdown (if runs data available)
-      const runs = data.runs || [];
-      if (runs.length > 0) {
-        const evalIds = [...new Set(runs.map(r => r.eval_id))].sort((a, b) => a - b);
-
-        html += "<h3 style='font-family: Poppins, sans-serif; margin-bottom: 0.75rem;'>Per-Eval Breakdown</h3>";
-
-        const hasTime = runs.some(r => r.result && r.result.time_seconds != null);
-        const hasErrors = runs.some(r => r.result && r.result.errors > 0);
-
-        for (const evalId of evalIds) {
-          const evalRuns = runs.filter(r => r.eval_id === evalId);
-          const evalName = evalRuns[0] && evalRuns[0].eval_name ? evalRuns[0].eval_name : "Eval " + evalId;
-
-          html += "<h4 style='font-family: Poppins, sans-serif; margin: 1rem 0 0.5rem; color: var(--text);'>" + escapeHtml(evalName) + "</h4>";
-          html += '<table class="benchmark-table">';
-          html += "<thead><tr><th>Config</th><th>Run</th><th>Pass Rate</th>";
-          if (hasTime) html += "<th>Time (s)</th>";
-          if (hasErrors) html += "<th>Crashes During Execution</th>";
-          html += "</tr></thead>";
-          html += "<tbody>";
-
-          // Group by config and render with average rows
-          const configGroups = [...new Set(evalRuns.map(r => r.configuration))];
-          for (let ci = 0; ci < configGroups.length; ci++) {
-            const config = configGroups[ci];
-            const configRuns = evalRuns.filter(r => r.configuration === config);
-            if (configRuns.length === 0) continue;
-
-            const rowClass = ci === 0 ? "benchmark-row-with" : "benchmark-row-without";
-            const configLabel = config.replace(/_/g, " ").replace(/\b\w/g, c => c.toUpperCase());
-
-            for (const run of configRuns) {
-              const r = run.result || {};
-              const prClass = r.pass_rate >= 0.8 ? "benchmark-delta-positive" : r.pass_rate < 0.5 ? "benchmark-delta-negative" : "";
-              html += '<tr class="' + rowClass + '">';
-              html += "<td>" + configLabel + "</td>";
-              html += "<td>" + run.run_number + "</td>";
-              html += '<td class="' + prClass + '">' + ((r.pass_rate || 0) * 100).toFixed(0) + "% (" + (r.passed || 0) + "/" + (r.total || 0) + ")</td>";
-              if (hasTime) html += "<td>" + (r.time_seconds != null ? r.time_seconds.toFixed(1) : "—") + "</td>";
-              if (hasErrors) html += "<td>" + (r.errors || 0) + "</td>";
-              html += "</tr>";
-            }
-
-            // Average row
-            const rates = configRuns.map(r => (r.result || {}).pass_rate || 0);
-            const avgRate = rates.reduce((a, b) => a + b, 0) / rates.length;
-            const avgPrClass = avgRate >= 0.8 ? "benchmark-delta-positive" : avgRate < 0.5 ? "benchmark-delta-negative" : "";
-            html += '<tr class="benchmark-row-avg ' + rowClass + '">';
-            html += "<td>" + configLabel + "</td>";
-            html += "<td>Avg</td>";
-            html += '<td class="' + avgPrClass + '">' + (avgRate * 100).toFixed(0) + "%</td>";
-            if (hasTime) {
-              const times = configRuns.map(r => (r.result || {}).time_seconds).filter(t => t != null);
-              html += "<td>" + (times.length ? (times.reduce((a, b) => a + b, 0) / times.length).toFixed(1) : "—") + "</td>";
-            }
-            if (hasErrors) html += "<td></td>";
-            html += "</tr>";
-          }
-          html += "</tbody></table>";
-
-          // Per-assertion detail for this eval
-          const runsWithExpectations = {};
-          for (const config of configGroups) {
-            runsWithExpectations[config] = evalRuns.filter(r => r.configuration === config && r.expectations && r.expectations.length > 0);
-          }
-          const hasAnyExpectations = Object.values(runsWithExpectations).some(runs => runs.length > 0);
-          if (hasAnyExpectations) {
-            // Collect all unique assertion texts across all configs
-            const allAssertions = [];
-            const seen = new Set();
-            for (const config of configGroups) {
-              for (const run of runsWithExpectations[config]) {
-                for (const exp of (run.expectations || [])) {
-                  if (!seen.has(exp.text)) {
-                    seen.add(exp.text);
-                    allAssertions.push(exp.text);
-                  }
-                }
-              }
-            }
-
-            html += '<table class="benchmark-table" style="margin-top: 0.5rem;">';
-            html += "<thead><tr><th>Assertion</th>";
-            for (const config of configGroups) {
-              const label = config.replace(/_/g, " ").replace(/\b\w/g, c => c.toUpperCase());
-              html += "<th>" + escapeHtml(label) + "</th>";
-            }
-            html += "</tr></thead><tbody>";
-
-            for (const assertionText of allAssertions) {
-              html += "<tr><td>" + escapeHtml(assertionText) + "</td>";
-
-              for (const config of configGroups) {
-                html += "<td>";
-                for (const run of runsWithExpectations[config]) {
-                  const exp = (run.expectations || []).find(e => e.text === assertionText);
-                  if (exp) {
-                    const cls = exp.passed ? "benchmark-delta-positive" : "benchmark-delta-negative";
-                    const icon = exp.passed ? "\u2713" : "\u2717";
-                    html += '<span class="' + cls + '" title="Run ' + run.run_number + ': ' + escapeHtml(exp.evidence || "") + '">' + icon + "</span> ";
-                  } else {
-                    html += "— ";
-                  }
-                }
-                html += "</td>";
-              }
-              html += "</tr>";
-            }
-            html += "</tbody></table>";
-          }
-        }
-      }
-
-      // Notes
-      if (notes.length > 0) {
-        html += '<div class="benchmark-notes">';
-        html += "<h3>Analysis Notes</h3>";
-        html += "<ul>";
-        for (const note of notes) {
-          html += "<li>" + escapeHtml(note) + "</li>";
-        }
-        html += "</ul></div>";
-      }
-
-      container.innerHTML = html;
-    }
-
-    // ---- Start ----
-    init();
-    renderBenchmark();
-  </script>
-</body>
-</html>
diff --git a/.github/skills/skill-creator/references/schemas.md b/.github/skills/skill-creator/references/schemas.md
deleted file mode 100644
index b6eeaa2..0000000
--- a/.github/skills/skill-creator/references/schemas.md
+++ /dev/null
@@ -1,430 +0,0 @@
-# JSON Schemas
-
-This document defines the JSON schemas used by skill-creator.
-
----
-
-## evals.json
-
-Defines the evals for a skill. Located at `evals/evals.json` within the skill directory.
-
-```json
-{
-  "skill_name": "example-skill",
-  "evals": [
-    {
-      "id": 1,
-      "prompt": "User's example prompt",
-      "expected_output": "Description of expected result",
-      "files": ["evals/files/sample1.pdf"],
-      "expectations": [
-        "The output includes X",
-        "The skill used script Y"
-      ]
-    }
-  ]
-}
-```
-
-**Fields:**
-- `skill_name`: Name matching the skill's frontmatter
-- `evals[].id`: Unique integer identifier
-- `evals[].prompt`: The task to execute
-- `evals[].expected_output`: Human-readable description of success
-- `evals[].files`: Optional list of input file paths (relative to skill root)
-- `evals[].expectations`: List of verifiable statements
-
----
-
-## history.json
-
-Tracks version progression in Improve mode. Located at workspace root.
-
-```json
-{
-  "started_at": "2026-01-15T10:30:00Z",
-  "skill_name": "pdf",
-  "current_best": "v2",
-  "iterations": [
-    {
-      "version": "v0",
-      "parent": null,
-      "expectation_pass_rate": 0.65,
-      "grading_result": "baseline",
-      "is_current_best": false
-    },
-    {
-      "version": "v1",
-      "parent": "v0",
-      "expectation_pass_rate": 0.75,
-      "grading_result": "won",
-      "is_current_best": false
-    },
-    {
-      "version": "v2",
-      "parent": "v1",
-      "expectation_pass_rate": 0.85,
-      "grading_result": "won",
-      "is_current_best": true
-    }
-  ]
-}
-```
-
-**Fields:**
-- `started_at`: ISO timestamp of when improvement started
-- `skill_name`: Name of the skill being improved
-- `current_best`: Version identifier of the best performer
-- `iterations[].version`: Version identifier (v0, v1, ...)
-- `iterations[].parent`: Parent version this was derived from
-- `iterations[].expectation_pass_rate`: Pass rate from grading
-- `iterations[].grading_result`: "baseline", "won", "lost", or "tie"
-- `iterations[].is_current_best`: Whether this is the current best version
-
----
-
-## grading.json
-
-Output from the grader agent. Located at `<run-dir>/grading.json`.
-
-```json
-{
-  "expectations": [
-    {
-      "text": "The output includes the name 'John Smith'",
-      "passed": true,
-      "evidence": "Found in transcript Step 3: 'Extracted names: John Smith, Sarah Johnson'"
-    },
-    {
-      "text": "The spreadsheet has a SUM formula in cell B10",
-      "passed": false,
-      "evidence": "No spreadsheet was created. The output was a text file."
-    }
-  ],
-  "summary": {
-    "passed": 2,
-    "failed": 1,
-    "total": 3,
-    "pass_rate": 0.67
-  },
-  "execution_metrics": {
-    "tool_calls": {
-      "Read": 5,
-      "Write": 2,
-      "Bash": 8
-    },
-    "total_tool_calls": 15,
-    "total_steps": 6,
-    "errors_encountered": 0,
-    "output_chars": 12450,
-    "transcript_chars": 3200
-  },
-  "timing": {
-    "executor_duration_seconds": 165.0,
-    "grader_duration_seconds": 26.0,
-    "total_duration_seconds": 191.0
-  },
-  "claims": [
-    {
-      "claim": "The form has 12 fillable fields",
-      "type": "factual",
-      "verified": true,
-      "evidence": "Counted 12 fields in field_info.json"
-    }
-  ],
-  "user_notes_summary": {
-    "uncertainties": ["Used 2023 data, may be stale"],
-    "needs_review": [],
-    "workarounds": ["Fell back to text overlay for non-fillable fields"]
-  },
-  "eval_feedback": {
-    "suggestions": [
-      {
-        "assertion": "The output includes the name 'John Smith'",
-        "reason": "A hallucinated document that mentions the name would also pass"
-      }
-    ],
-    "overall": "Assertions check presence but not correctness."
-  }
-}
-```
-
-**Fields:**
-- `expectations[]`: Graded expectations with evidence
-- `summary`: Aggregate pass/fail counts
-- `execution_metrics`: Tool usage and output size (from executor's metrics.json)
-- `timing`: Wall clock timing (from timing.json)
-- `claims`: Extracted and verified claims from the output
-- `user_notes_summary`: Issues flagged by the executor
-- `eval_feedback`: (optional) Improvement suggestions for the evals, only present when the grader identifies issues worth raising
-
----
-
-## metrics.json
-
-Output from the executor agent. Located at `<run-dir>/outputs/metrics.json`.
-
-```json
-{
-  "tool_calls": {
-    "Read": 5,
-    "Write": 2,
-    "Bash": 8,
-    "Edit": 1,
-    "Glob": 2,
-    "Grep": 0
-  },
-  "total_tool_calls": 18,
-  "total_steps": 6,
-  "files_created": ["filled_form.pdf", "field_values.json"],
-  "errors_encountered": 0,
-  "output_chars": 12450,
-  "transcript_chars": 3200
-}
-```
-
-**Fields:**
-- `tool_calls`: Count per tool type
-- `total_tool_calls`: Sum of all tool calls
-- `total_steps`: Number of major execution steps
-- `files_created`: List of output files created
-- `errors_encountered`: Number of errors during execution
-- `output_chars`: Total character count of output files
-- `transcript_chars`: Character count of transcript
-
----
-
-## timing.json
-
-Wall clock timing for a run. Located at `<run-dir>/timing.json`.
-
-**How to capture:** When a subagent task completes, the task notification includes `total_tokens` and `duration_ms`. Save these immediately — they are not persisted anywhere else and cannot be recovered after the fact.
-
-```json
-{
-  "total_tokens": 84852,
-  "duration_ms": 23332,
-  "total_duration_seconds": 23.3,
-  "executor_start": "2026-01-15T10:30:00Z",
-  "executor_end": "2026-01-15T10:32:45Z",
-  "executor_duration_seconds": 165.0,
-  "grader_start": "2026-01-15T10:32:46Z",
-  "grader_end": "2026-01-15T10:33:12Z",
-  "grader_duration_seconds": 26.0
-}
-```
-
----
-
-## benchmark.json
-
-Output from Benchmark mode. Located at `benchmarks/<timestamp>/benchmark.json`.
-
-```json
-{
-  "metadata": {
-    "skill_name": "pdf",
-    "skill_path": "/path/to/pdf",
-    "executor_model": "claude-sonnet-4-20250514",
-    "analyzer_model": "most-capable-model",
-    "timestamp": "2026-01-15T10:30:00Z",
-    "evals_run": [1, 2, 3],
-    "runs_per_configuration": 3
-  },
-
-  "runs": [
-    {
-      "eval_id": 1,
-      "eval_name": "Ocean",
-      "configuration": "with_skill",
-      "run_number": 1,
-      "result": {
-        "pass_rate": 0.85,
-        "passed": 6,
-        "failed": 1,
-        "total": 7,
-        "time_seconds": 42.5,
-        "tokens": 3800,
-        "tool_calls": 18,
-        "errors": 0
-      },
-      "expectations": [
-        {"text": "...", "passed": true, "evidence": "..."}
-      ],
-      "notes": [
-        "Used 2023 data, may be stale",
-        "Fell back to text overlay for non-fillable fields"
-      ]
-    }
-  ],
-
-  "run_summary": {
-    "with_skill": {
-      "pass_rate": {"mean": 0.85, "stddev": 0.05, "min": 0.80, "max": 0.90},
-      "time_seconds": {"mean": 45.0, "stddev": 12.0, "min": 32.0, "max": 58.0},
-      "tokens": {"mean": 3800, "stddev": 400, "min": 3200, "max": 4100}
-    },
-    "without_skill": {
-      "pass_rate": {"mean": 0.35, "stddev": 0.08, "min": 0.28, "max": 0.45},
-      "time_seconds": {"mean": 32.0, "stddev": 8.0, "min": 24.0, "max": 42.0},
-      "tokens": {"mean": 2100, "stddev": 300, "min": 1800, "max": 2500}
-    },
-    "delta": {
-      "pass_rate": "+0.50",
-      "time_seconds": "+13.0",
-      "tokens": "+1700"
-    }
-  },
-
-  "notes": [
-    "Assertion 'Output is a PDF file' passes 100% in both configurations - may not differentiate skill value",
-    "Eval 3 shows high variance (50% ± 40%) - may be flaky or model-dependent",
-    "Without-skill runs consistently fail on table extraction expectations",
-    "Skill adds 13s average execution time but improves pass rate by 50%"
-  ]
-}
-```
-
-**Fields:**
-- `metadata`: Information about the benchmark run
-  - `skill_name`: Name of the skill
-  - `timestamp`: When the benchmark was run
-  - `evals_run`: List of eval names or IDs
-  - `runs_per_configuration`: Number of runs per config (e.g. 3)
-- `runs[]`: Individual run results
-  - `eval_id`: Numeric eval identifier
-  - `eval_name`: Human-readable eval name (used as section header in the viewer)
-  - `configuration`: Must be `"with_skill"` or `"without_skill"` (the viewer uses this exact string for grouping and color coding)
-  - `run_number`: Integer run number (1, 2, 3...)
-  - `result`: Nested object with `pass_rate`, `passed`, `total`, `time_seconds`, `tokens`, `errors`
-- `run_summary`: Statistical aggregates per configuration
-  - `with_skill` / `without_skill`: Each contains `pass_rate`, `time_seconds`, `tokens` objects with `mean` and `stddev` fields
-  - `delta`: Difference strings like `"+0.50"`, `"+13.0"`, `"+1700"`
-- `notes`: Freeform observations from the analyzer
-
-**Important:** The viewer reads these field names exactly. Using `config` instead of `configuration`, or putting `pass_rate` at the top level of a run instead of nested under `result`, will cause the viewer to show empty/zero values. Always reference this schema when generating benchmark.json manually.
-
----
-
-## comparison.json
-
-Output from blind comparator. Located at `<grading-dir>/comparison-N.json`.
-
-```json
-{
-  "winner": "A",
-  "reasoning": "Output A provides a complete solution with proper formatting and all required fields. Output B is missing the date field and has formatting inconsistencies.",
-  "rubric": {
-    "A": {
-      "content": {
-        "correctness": 5,
-        "completeness": 5,
-        "accuracy": 4
-      },
-      "structure": {
-        "organization": 4,
-        "formatting": 5,
-        "usability": 4
-      },
-      "content_score": 4.7,
-      "structure_score": 4.3,
-      "overall_score": 9.0
-    },
-    "B": {
-      "content": {
-        "correctness": 3,
-        "completeness": 2,
-        "accuracy": 3
-      },
-      "structure": {
-        "organization": 3,
-        "formatting": 2,
-        "usability": 3
-      },
-      "content_score": 2.7,
-      "structure_score": 2.7,
-      "overall_score": 5.4
-    }
-  },
-  "output_quality": {
-    "A": {
-      "score": 9,
-      "strengths": ["Complete solution", "Well-formatted", "All fields present"],
-      "weaknesses": ["Minor style inconsistency in header"]
-    },
-    "B": {
-      "score": 5,
-      "strengths": ["Readable output", "Correct basic structure"],
-      "weaknesses": ["Missing date field", "Formatting inconsistencies", "Partial data extraction"]
-    }
-  },
-  "expectation_results": {
-    "A": {
-      "passed": 4,
-      "total": 5,
-      "pass_rate": 0.80,
-      "details": [
-        {"text": "Output includes name", "passed": true}
-      ]
-    },
-    "B": {
-      "passed": 3,
-      "total": 5,
-      "pass_rate": 0.60,
-      "details": [
-        {"text": "Output includes name", "passed": true}
-      ]
-    }
-  }
-}
-```
-
----
-
-## analysis.json
-
-Output from post-hoc analyzer. Located at `<grading-dir>/analysis.json`.
-
-```json
-{
-  "comparison_summary": {
-    "winner": "A",
-    "winner_skill": "path/to/winner/skill",
-    "loser_skill": "path/to/loser/skill",
-    "comparator_reasoning": "Brief summary of why comparator chose winner"
-  },
-  "winner_strengths": [
-    "Clear step-by-step instructions for handling multi-page documents",
-    "Included validation script that caught formatting errors"
-  ],
-  "loser_weaknesses": [
-    "Vague instruction 'process the document appropriately' led to inconsistent behavior",
-    "No script for validation, agent had to improvise"
-  ],
-  "instruction_following": {
-    "winner": {
-      "score": 9,
-      "issues": ["Minor: skipped optional logging step"]
-    },
-    "loser": {
-      "score": 6,
-      "issues": [
-        "Did not use the skill's formatting template",
-        "Invented own approach instead of following step 3"
-      ]
-    }
-  },
-  "improvement_suggestions": [
-    {
-      "priority": "high",
-      "category": "instructions",
-      "suggestion": "Replace 'process the document appropriately' with explicit steps",
-      "expected_impact": "Would eliminate ambiguity that caused inconsistent behavior"
-    }
-  ],
-  "transcript_insights": {
-    "winner_execution_pattern": "Read skill -> Followed 5-step process -> Used validation script",
-    "loser_execution_pattern": "Read skill -> Unclear on approach -> Tried 3 different methods"
-  }
-}
-```
diff --git a/.github/skills/skill-creator/scripts/__init__.py b/.github/skills/skill-creator/scripts/__init__.py
deleted file mode 100644
index 5d8369d..0000000
--- a/.github/skills/skill-creator/scripts/__init__.py
+++ /dev/null
@@ -1,2 +0,0 @@
-# Copyright (c) Cratis. All rights reserved.
-# Licensed under the MIT license. See LICENSE file in the project root for full license information.
diff --git a/.github/skills/skill-creator/scripts/aggregate_benchmark.py b/.github/skills/skill-creator/scripts/aggregate_benchmark.py
deleted file mode 100644
index 3e66e8c..0000000
--- a/.github/skills/skill-creator/scripts/aggregate_benchmark.py
+++ /dev/null
@@ -1,401 +0,0 @@
-#!/usr/bin/env python3
-"""
-Aggregate individual run results into benchmark summary statistics.
-
-Reads grading.json files from run directories and produces:
-- run_summary with mean, stddev, min, max for each metric
-- delta between with_skill and without_skill configurations
-
-Usage:
-    python aggregate_benchmark.py <benchmark_dir>
-
-Example:
-    python aggregate_benchmark.py benchmarks/2026-01-15T10-30-00/
-
-The script supports two directory layouts:
-
-    Workspace layout (from skill-creator iterations):
-    <benchmark_dir>/
-    └── eval-N/
-        ├── with_skill/
-        │   ├── run-1/grading.json
-        │   └── run-2/grading.json
-        └── without_skill/
-            ├── run-1/grading.json
-            └── run-2/grading.json
-
-    Legacy layout (with runs/ subdirectory):
-    <benchmark_dir>/
-    └── runs/
-        └── eval-N/
-            ├── with_skill/
-            │   └── run-1/grading.json
-            └── without_skill/
-                └── run-1/grading.json
-"""
-
-import argparse
-import json
-import math
-import sys
-from datetime import datetime, timezone
-from pathlib import Path
-
-
-def calculate_stats(values: list[float]) -> dict:
-    """Calculate mean, stddev, min, max for a list of values."""
-    if not values:
-        return {"mean": 0.0, "stddev": 0.0, "min": 0.0, "max": 0.0}
-
-    n = len(values)
-    mean = sum(values) / n
-
-    if n > 1:
-        variance = sum((x - mean) ** 2 for x in values) / (n - 1)
-        stddev = math.sqrt(variance)
-    else:
-        stddev = 0.0
-
-    return {
-        "mean": round(mean, 4),
-        "stddev": round(stddev, 4),
-        "min": round(min(values), 4),
-        "max": round(max(values), 4)
-    }
-
-
-def load_run_results(benchmark_dir: Path) -> dict:
-    """
-    Load all run results from a benchmark directory.
-
-    Returns dict keyed by config name (e.g. "with_skill"/"without_skill",
-    or "new_skill"/"old_skill"), each containing a list of run results.
-    """
-    # Support both layouts: eval dirs directly under benchmark_dir, or under runs/
-    runs_dir = benchmark_dir / "runs"
-    if runs_dir.exists():
-        search_dir = runs_dir
-    elif list(benchmark_dir.glob("eval-*")):
-        search_dir = benchmark_dir
-    else:
-        print(f"No eval directories found in {benchmark_dir} or {benchmark_dir / 'runs'}")
-        return {}
-
-    results: dict[str, list] = {}
-
-    for eval_idx, eval_dir in enumerate(sorted(search_dir.glob("eval-*"))):
-        metadata_path = eval_dir / "eval_metadata.json"
-        if metadata_path.exists():
-            try:
-                with open(metadata_path) as mf:
-                    eval_id = json.load(mf).get("eval_id", eval_idx)
-            except (json.JSONDecodeError, OSError):
-                eval_id = eval_idx
-        else:
-            try:
-                eval_id = int(eval_dir.name.split("-")[1])
-            except ValueError:
-                eval_id = eval_idx
-
-        # Discover config directories dynamically rather than hardcoding names
-        for config_dir in sorted(eval_dir.iterdir()):
-            if not config_dir.is_dir():
-                continue
-            # Skip non-config directories (inputs, outputs, etc.)
-            if not list(config_dir.glob("run-*")):
-                continue
-            config = config_dir.name
-            if config not in results:
-                results[config] = []
-
-            for run_dir in sorted(config_dir.glob("run-*")):
-                run_number = int(run_dir.name.split("-")[1])
-                grading_file = run_dir / "grading.json"
-
-                if not grading_file.exists():
-                    print(f"Warning: grading.json not found in {run_dir}")
-                    continue
-
-                try:
-                    with open(grading_file) as f:
-                        grading = json.load(f)
-                except json.JSONDecodeError as e:
-                    print(f"Warning: Invalid JSON in {grading_file}: {e}")
-                    continue
-
-                # Extract metrics
-                result = {
-                    "eval_id": eval_id,
-                    "run_number": run_number,
-                    "pass_rate": grading.get("summary", {}).get("pass_rate", 0.0),
-                    "passed": grading.get("summary", {}).get("passed", 0),
-                    "failed": grading.get("summary", {}).get("failed", 0),
-                    "total": grading.get("summary", {}).get("total", 0),
-                }
-
-                # Extract timing — check grading.json first, then sibling timing.json
-                timing = grading.get("timing", {})
-                result["time_seconds"] = timing.get("total_duration_seconds", 0.0)
-                timing_file = run_dir / "timing.json"
-                if result["time_seconds"] == 0.0 and timing_file.exists():
-                    try:
-                        with open(timing_file) as tf:
-                            timing_data = json.load(tf)
-                        result["time_seconds"] = timing_data.get("total_duration_seconds", 0.0)
-                        result["tokens"] = timing_data.get("total_tokens", 0)
-                    except json.JSONDecodeError:
-                        pass
-
-                # Extract metrics if available
-                metrics = grading.get("execution_metrics", {})
-                result["tool_calls"] = metrics.get("total_tool_calls", 0)
-                if not result.get("tokens"):
-                    result["tokens"] = metrics.get("output_chars", 0)
-                result["errors"] = metrics.get("errors_encountered", 0)
-
-                # Extract expectations — viewer requires fields: text, passed, evidence
-                raw_expectations = grading.get("expectations", [])
-                for exp in raw_expectations:
-                    if "text" not in exp or "passed" not in exp:
-                        print(f"Warning: expectation in {grading_file} missing required fields (text, passed, evidence): {exp}")
-                result["expectations"] = raw_expectations
-
-                # Extract notes from user_notes_summary
-                notes_summary = grading.get("user_notes_summary", {})
-                notes = []
-                notes.extend(notes_summary.get("uncertainties", []))
-                notes.extend(notes_summary.get("needs_review", []))
-                notes.extend(notes_summary.get("workarounds", []))
-                result["notes"] = notes
-
-                results[config].append(result)
-
-    return results
-
-
-def aggregate_results(results: dict) -> dict:
-    """
-    Aggregate run results into summary statistics.
-
-    Returns run_summary with stats for each configuration and delta.
-    """
-    run_summary = {}
-    configs = list(results.keys())
-
-    for config in configs:
-        runs = results.get(config, [])
-
-        if not runs:
-            run_summary[config] = {
-                "pass_rate": {"mean": 0.0, "stddev": 0.0, "min": 0.0, "max": 0.0},
-                "time_seconds": {"mean": 0.0, "stddev": 0.0, "min": 0.0, "max": 0.0},
-                "tokens": {"mean": 0, "stddev": 0, "min": 0, "max": 0}
-            }
-            continue
-
-        pass_rates = [r["pass_rate"] for r in runs]
-        times = [r["time_seconds"] for r in runs]
-        tokens = [r.get("tokens", 0) for r in runs]
-
-        run_summary[config] = {
-            "pass_rate": calculate_stats(pass_rates),
-            "time_seconds": calculate_stats(times),
-            "tokens": calculate_stats(tokens)
-        }
-
-    # Calculate delta between the first two configs (if two exist)
-    if len(configs) >= 2:
-        primary = run_summary.get(configs[0], {})
-        baseline = run_summary.get(configs[1], {})
-    else:
-        primary = run_summary.get(configs[0], {}) if configs else {}
-        baseline = {}
-
-    delta_pass_rate = primary.get("pass_rate", {}).get("mean", 0) - baseline.get("pass_rate", {}).get("mean", 0)
-    delta_time = primary.get("time_seconds", {}).get("mean", 0) - baseline.get("time_seconds", {}).get("mean", 0)
-    delta_tokens = primary.get("tokens", {}).get("mean", 0) - baseline.get("tokens", {}).get("mean", 0)
-
-    run_summary["delta"] = {
-        "pass_rate": f"{delta_pass_rate:+.2f}",
-        "time_seconds": f"{delta_time:+.1f}",
-        "tokens": f"{delta_tokens:+.0f}"
-    }
-
-    return run_summary
-
-
-def generate_benchmark(benchmark_dir: Path, skill_name: str = "", skill_path: str = "") -> dict:
-    """
-    Generate complete benchmark.json from run results.
-    """
-    results = load_run_results(benchmark_dir)
-    run_summary = aggregate_results(results)
-
-    # Build runs array for benchmark.json
-    runs = []
-    for config in results:
-        for result in results[config]:
-            runs.append({
-                "eval_id": result["eval_id"],
-                "configuration": config,
-                "run_number": result["run_number"],
-                "result": {
-                    "pass_rate": result["pass_rate"],
-                    "passed": result["passed"],
-                    "failed": result["failed"],
-                    "total": result["total"],
-                    "time_seconds": result["time_seconds"],
-                    "tokens": result.get("tokens", 0),
-                    "tool_calls": result.get("tool_calls", 0),
-                    "errors": result.get("errors", 0)
-                },
-                "expectations": result["expectations"],
-                "notes": result["notes"]
-            })
-
-    # Determine eval IDs from results
-    eval_ids = sorted(set(
-        r["eval_id"]
-        for config in results.values()
-        for r in config
-    ))
-
-    benchmark = {
-        "metadata": {
-            "skill_name": skill_name or "<skill-name>",
-            "skill_path": skill_path or "<path/to/skill>",
-            "executor_model": "<model-name>",
-            "analyzer_model": "<model-name>",
-            "timestamp": datetime.now(timezone.utc).strftime("%Y-%m-%dT%H:%M:%SZ"),
-            "evals_run": eval_ids,
-            "runs_per_configuration": 3
-        },
-        "runs": runs,
-        "run_summary": run_summary,
-        "notes": []  # To be filled by analyzer
-    }
-
-    return benchmark
-
-
-def generate_markdown(benchmark: dict) -> str:
-    """Generate human-readable benchmark.md from benchmark data."""
-    metadata = benchmark["metadata"]
-    run_summary = benchmark["run_summary"]
-
-    # Determine config names (excluding "delta")
-    configs = [k for k in run_summary if k != "delta"]
-    config_a = configs[0] if len(configs) >= 1 else "config_a"
-    config_b = configs[1] if len(configs) >= 2 else "config_b"
-    label_a = config_a.replace("_", " ").title()
-    label_b = config_b.replace("_", " ").title()
-
-    lines = [
-        f"# Skill Benchmark: {metadata['skill_name']}",
-        "",
-        f"**Model**: {metadata['executor_model']}",
-        f"**Date**: {metadata['timestamp']}",
-        f"**Evals**: {', '.join(map(str, metadata['evals_run']))} ({metadata['runs_per_configuration']} runs each per configuration)",
-        "",
-        "## Summary",
-        "",
-        f"| Metric | {label_a} | {label_b} | Delta |",
-        "|--------|------------|---------------|-------|",
-    ]
-
-    a_summary = run_summary.get(config_a, {})
-    b_summary = run_summary.get(config_b, {})
-    delta = run_summary.get("delta", {})
-
-    # Format pass rate
-    a_pr = a_summary.get("pass_rate", {})
-    b_pr = b_summary.get("pass_rate", {})
-    lines.append(f"| Pass Rate | {a_pr.get('mean', 0)*100:.0f}% ± {a_pr.get('stddev', 0)*100:.0f}% | {b_pr.get('mean', 0)*100:.0f}% ± {b_pr.get('stddev', 0)*100:.0f}% | {delta.get('pass_rate', '—')} |")
-
-    # Format time
-    a_time = a_summary.get("time_seconds", {})
-    b_time = b_summary.get("time_seconds", {})
-    lines.append(f"| Time | {a_time.get('mean', 0):.1f}s ± {a_time.get('stddev', 0):.1f}s | {b_time.get('mean', 0):.1f}s ± {b_time.get('stddev', 0):.1f}s | {delta.get('time_seconds', '—')}s |")
-
-    # Format tokens
-    a_tokens = a_summary.get("tokens", {})
-    b_tokens = b_summary.get("tokens", {})
-    lines.append(f"| Tokens | {a_tokens.get('mean', 0):.0f} ± {a_tokens.get('stddev', 0):.0f} | {b_tokens.get('mean', 0):.0f} ± {b_tokens.get('stddev', 0):.0f} | {delta.get('tokens', '—')} |")
-
-    # Notes section
-    if benchmark.get("notes"):
-        lines.extend([
-            "",
-            "## Notes",
-            ""
-        ])
-        for note in benchmark["notes"]:
-            lines.append(f"- {note}")
-
-    return "\n".join(lines)
-
-
-def main():
-    parser = argparse.ArgumentParser(
-        description="Aggregate benchmark run results into summary statistics"
-    )
-    parser.add_argument(
-        "benchmark_dir",
-        type=Path,
-        help="Path to the benchmark directory"
-    )
-    parser.add_argument(
-        "--skill-name",
-        default="",
-        help="Name of the skill being benchmarked"
-    )
-    parser.add_argument(
-        "--skill-path",
-        default="",
-        help="Path to the skill being benchmarked"
-    )
-    parser.add_argument(
-        "--output", "-o",
-        type=Path,
-        help="Output path for benchmark.json (default: <benchmark_dir>/benchmark.json)"
-    )
-
-    args = parser.parse_args()
-
-    if not args.benchmark_dir.exists():
-        print(f"Directory not found: {args.benchmark_dir}")
-        sys.exit(1)
-
-    # Generate benchmark
-    benchmark = generate_benchmark(args.benchmark_dir, args.skill_name, args.skill_path)
-
-    # Determine output paths
-    output_json = args.output or (args.benchmark_dir / "benchmark.json")
-    output_md = output_json.with_suffix(".md")
-
-    # Write benchmark.json
-    with open(output_json, "w") as f:
-        json.dump(benchmark, f, indent=2)
-    print(f"Generated: {output_json}")
-
-    # Write benchmark.md
-    markdown = generate_markdown(benchmark)
-    with open(output_md, "w") as f:
-        f.write(markdown)
-    print(f"Generated: {output_md}")
-
-    # Print summary
-    run_summary = benchmark["run_summary"]
-    configs = [k for k in run_summary if k != "delta"]
-    delta = run_summary.get("delta", {})
-
-    print(f"\nSummary:")
-    for config in configs:
-        pr = run_summary[config]["pass_rate"]["mean"]
-        label = config.replace("_", " ").title()
-        print(f"  {label}: {pr*100:.1f}% pass rate")
-    print(f"  Delta:         {delta.get('pass_rate', '—')}")
-
-
-if __name__ == "__main__":
-    main()
diff --git a/.github/skills/skill-creator/scripts/generate_report.py b/.github/skills/skill-creator/scripts/generate_report.py
deleted file mode 100644
index 959e30a..0000000
--- a/.github/skills/skill-creator/scripts/generate_report.py
+++ /dev/null
@@ -1,326 +0,0 @@
-#!/usr/bin/env python3
-"""Generate an HTML report from run_loop.py output.
-
-Takes the JSON output from run_loop.py and generates a visual HTML report
-showing each description attempt with check/x for each test case.
-Distinguishes between train and test queries.
-"""
-
-import argparse
-import html
-import json
-import sys
-from pathlib import Path
-
-
-def generate_html(data: dict, auto_refresh: bool = False, skill_name: str = "") -> str:
-    """Generate HTML report from loop output data. If auto_refresh is True, adds a meta refresh tag."""
-    history = data.get("history", [])
-    holdout = data.get("holdout", 0)
-    title_prefix = html.escape(skill_name + " \u2014 ") if skill_name else ""
-
-    # Get all unique queries from train and test sets, with should_trigger info
-    train_queries: list[dict] = []
-    test_queries: list[dict] = []
-    if history:
-        for r in history[0].get("train_results", history[0].get("results", [])):
-            train_queries.append({"query": r["query"], "should_trigger": r.get("should_trigger", True)})
-        if history[0].get("test_results"):
-            for r in history[0].get("test_results", []):
-                test_queries.append({"query": r["query"], "should_trigger": r.get("should_trigger", True)})
-
-    refresh_tag = '    <meta http-equiv="refresh" content="5">\n' if auto_refresh else ""
-
-    html_parts = ["""<!DOCTYPE html>
-<html>
-<head>
-    <meta charset="utf-8">
-""" + refresh_tag + """    <title>""" + title_prefix + """Skill Description Optimization</title>
-    <link rel="preconnect" href="https://fonts.googleapis.com">
-    <link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
-    <link href="https://fonts.googleapis.com/css2?family=Poppins:wght@500;600&family=Lora:wght@400;500&display=swap" rel="stylesheet">
-    <style>
-        body {
-            font-family: 'Lora', Georgia, serif;
-            max-width: 100%;
-            margin: 0 auto;
-            padding: 20px;
-            background: #faf9f5;
-            color: #141413;
-        }
-        h1 { font-family: 'Poppins', sans-serif; color: #141413; }
-        .explainer {
-            background: white;
-            padding: 15px;
-            border-radius: 6px;
-            margin-bottom: 20px;
-            border: 1px solid #e8e6dc;
-            color: #b0aea5;
-            font-size: 0.875rem;
-            line-height: 1.6;
-        }
-        .summary {
-            background: white;
-            padding: 15px;
-            border-radius: 6px;
-            margin-bottom: 20px;
-            border: 1px solid #e8e6dc;
-        }
-        .summary p { margin: 5px 0; }
-        .best { color: #788c5d; font-weight: bold; }
-        .table-container {
-            overflow-x: auto;
-            width: 100%;
-        }
-        table {
-            border-collapse: collapse;
-            background: white;
-            border: 1px solid #e8e6dc;
-            border-radius: 6px;
-            font-size: 12px;
-            min-width: 100%;
-        }
-        th, td {
-            padding: 8px;
-            text-align: left;
-            border: 1px solid #e8e6dc;
-            white-space: normal;
-            word-wrap: break-word;
-        }
-        th {
-            font-family: 'Poppins', sans-serif;
-            background: #141413;
-            color: #faf9f5;
-            font-weight: 500;
-        }
-        th.test-col {
-            background: #6a9bcc;
-        }
-        th.query-col { min-width: 200px; }
-        td.description {
-            font-family: monospace;
-            font-size: 11px;
-            word-wrap: break-word;
-            max-width: 400px;
-        }
-        td.result {
-            text-align: center;
-            font-size: 16px;
-            min-width: 40px;
-        }
-        td.test-result {
-            background: #f0f6fc;
-        }
-        .pass { color: #788c5d; }
-        .fail { color: #c44; }
-        .rate {
-            font-size: 9px;
-            color: #b0aea5;
-            display: block;
-        }
-        tr:hover { background: #faf9f5; }
-        .score {
-            display: inline-block;
-            padding: 2px 6px;
-            border-radius: 4px;
-            font-weight: bold;
-            font-size: 11px;
-        }
-        .score-good { background: #eef2e8; color: #788c5d; }
-        .score-ok { background: #fef3c7; color: #d97706; }
-        .score-bad { background: #fceaea; color: #c44; }
-        .train-label { color: #b0aea5; font-size: 10px; }
-        .test-label { color: #6a9bcc; font-size: 10px; font-weight: bold; }
-        .best-row { background: #f5f8f2; }
-        th.positive-col { border-bottom: 3px solid #788c5d; }
-        th.negative-col { border-bottom: 3px solid #c44; }
-        th.test-col.positive-col { border-bottom: 3px solid #788c5d; }
-        th.test-col.negative-col { border-bottom: 3px solid #c44; }
-        .legend { font-family: 'Poppins', sans-serif; display: flex; gap: 20px; margin-bottom: 10px; font-size: 13px; align-items: center; }
-        .legend-item { display: flex; align-items: center; gap: 6px; }
-        .legend-swatch { width: 16px; height: 16px; border-radius: 3px; display: inline-block; }
-        .swatch-positive { background: #141413; border-bottom: 3px solid #788c5d; }
-        .swatch-negative { background: #141413; border-bottom: 3px solid #c44; }
-        .swatch-test { background: #6a9bcc; }
-        .swatch-train { background: #141413; }
-    </style>
-</head>
-<body>
-    <h1>""" + title_prefix + """Skill Description Optimization</h1>
-    <div class="explainer">
-        <strong>Optimizing your skill's description.</strong> This page updates automatically as Claude tests different versions of your skill's description. Each row is an iteration — a new description attempt. The columns show test queries: green checkmarks mean the skill triggered correctly (or correctly didn't trigger), red crosses mean it got it wrong. The "Train" score shows performance on queries used to improve the description; the "Test" score shows performance on held-out queries the optimizer hasn't seen. When it's done, Claude will apply the best-performing description to your skill.
-    </div>
-"""]
-
-    # Summary section
-    best_test_score = data.get('best_test_score')
-    best_train_score = data.get('best_train_score')
-    html_parts.append(f"""
-    <div class="summary">
-        <p><strong>Original:</strong> {html.escape(data.get('original_description', 'N/A'))}</p>
-        <p class="best"><strong>Best:</strong> {html.escape(data.get('best_description', 'N/A'))}</p>
-        <p><strong>Best Score:</strong> {data.get('best_score', 'N/A')} {'(test)' if best_test_score else '(train)'}</p>
-        <p><strong>Iterations:</strong> {data.get('iterations_run', 0)} | <strong>Train:</strong> {data.get('train_size', '?')} | <strong>Test:</strong> {data.get('test_size', '?')}</p>
-    </div>
-""")
-
-    # Legend
-    html_parts.append("""
-    <div class="legend">
-        <span style="font-weight:600">Query columns:</span>
-        <span class="legend-item"><span class="legend-swatch swatch-positive"></span> Should trigger</span>
-        <span class="legend-item"><span class="legend-swatch swatch-negative"></span> Should NOT trigger</span>
-        <span class="legend-item"><span class="legend-swatch swatch-train"></span> Train</span>
-        <span class="legend-item"><span class="legend-swatch swatch-test"></span> Test</span>
-    </div>
-""")
-
-    # Table header
-    html_parts.append("""
-    <div class="table-container">
-    <table>
-        <thead>
-            <tr>
-                <th>Iter</th>
-                <th>Train</th>
-                <th>Test</th>
-                <th class="query-col">Description</th>
-""")
-
-    # Add column headers for train queries
-    for qinfo in train_queries:
-        polarity = "positive-col" if qinfo["should_trigger"] else "negative-col"
-        html_parts.append(f'                <th class="{polarity}">{html.escape(qinfo["query"])}</th>\n')
-
-    # Add column headers for test queries (different color)
-    for qinfo in test_queries:
-        polarity = "positive-col" if qinfo["should_trigger"] else "negative-col"
-        html_parts.append(f'                <th class="test-col {polarity}">{html.escape(qinfo["query"])}</th>\n')
-
-    html_parts.append("""            </tr>
-        </thead>
-        <tbody>
-""")
-
-    # Find best iteration for highlighting
-    if test_queries:
-        best_iter = max(history, key=lambda h: h.get("test_passed") or 0).get("iteration")
-    else:
-        best_iter = max(history, key=lambda h: h.get("train_passed", h.get("passed", 0))).get("iteration")
-
-    # Add rows for each iteration
-    for h in history:
-        iteration = h.get("iteration", "?")
-        train_passed = h.get("train_passed", h.get("passed", 0))
-        train_total = h.get("train_total", h.get("total", 0))
-        test_passed = h.get("test_passed")
-        test_total = h.get("test_total")
-        description = h.get("description", "")
-        train_results = h.get("train_results", h.get("results", []))
-        test_results = h.get("test_results", [])
-
-        # Create lookups for results by query
-        train_by_query = {r["query"]: r for r in train_results}
-        test_by_query = {r["query"]: r for r in test_results} if test_results else {}
-
-        # Compute aggregate correct/total runs across all retries
-        def aggregate_runs(results: list[dict]) -> tuple[int, int]:
-            correct = 0
-            total = 0
-            for r in results:
-                runs = r.get("runs", 0)
-                triggers = r.get("triggers", 0)
-                total += runs
-                if r.get("should_trigger", True):
-                    correct += triggers
-                else:
-                    correct += runs - triggers
-            return correct, total
-
-        train_correct, train_runs = aggregate_runs(train_results)
-        test_correct, test_runs = aggregate_runs(test_results)
-
-        # Determine score classes
-        def score_class(correct: int, total: int) -> str:
-            if total > 0:
-                ratio = correct / total
-                if ratio >= 0.8:
-                    return "score-good"
-                elif ratio >= 0.5:
-                    return "score-ok"
-            return "score-bad"
-
-        train_class = score_class(train_correct, train_runs)
-        test_class = score_class(test_correct, test_runs)
-
-        row_class = "best-row" if iteration == best_iter else ""
-
-        html_parts.append(f"""            <tr class="{row_class}">
-                <td>{iteration}</td>
-                <td><span class="score {train_class}">{train_correct}/{train_runs}</span></td>
-                <td><span class="score {test_class}">{test_correct}/{test_runs}</span></td>
-                <td class="description">{html.escape(description)}</td>
-""")
-
-        # Add result for each train query
-        for qinfo in train_queries:
-            r = train_by_query.get(qinfo["query"], {})
-            did_pass = r.get("pass", False)
-            triggers = r.get("triggers", 0)
-            runs = r.get("runs", 0)
-
-            icon = "✓" if did_pass else "✗"
-            css_class = "pass" if did_pass else "fail"
-
-            html_parts.append(f'                <td class="result {css_class}">{icon}<span class="rate">{triggers}/{runs}</span></td>\n')
-
-        # Add result for each test query (with different background)
-        for qinfo in test_queries:
-            r = test_by_query.get(qinfo["query"], {})
-            did_pass = r.get("pass", False)
-            triggers = r.get("triggers", 0)
-            runs = r.get("runs", 0)
-
-            icon = "✓" if did_pass else "✗"
-            css_class = "pass" if did_pass else "fail"
-
-            html_parts.append(f'                <td class="result test-result {css_class}">{icon}<span class="rate">{triggers}/{runs}</span></td>\n')
-
-        html_parts.append("            </tr>\n")
-
-    html_parts.append("""        </tbody>
-    </table>
-    </div>
-""")
-
-    html_parts.append("""
-</body>
-</html>
-""")
-
-    return "".join(html_parts)
-
-
-def main():
-    parser = argparse.ArgumentParser(description="Generate HTML report from run_loop output")
-    parser.add_argument("input", help="Path to JSON output from run_loop.py (or - for stdin)")
-    parser.add_argument("-o", "--output", default=None, help="Output HTML file (default: stdout)")
-    parser.add_argument("--skill-name", default="", help="Skill name to include in the report title")
-    args = parser.parse_args()
-
-    if args.input == "-":
-        data = json.load(sys.stdin)
-    else:
-        data = json.loads(Path(args.input).read_text())
-
-    html_output = generate_html(data, skill_name=args.skill_name)
-
-    if args.output:
-        Path(args.output).write_text(html_output)
-        print(f"Report written to {args.output}", file=sys.stderr)
-    else:
-        print(html_output)
-
-
-if __name__ == "__main__":
-    main()
diff --git a/.github/skills/skill-creator/scripts/improve_description.py b/.github/skills/skill-creator/scripts/improve_description.py
deleted file mode 100644
index a270777..0000000
--- a/.github/skills/skill-creator/scripts/improve_description.py
+++ /dev/null
@@ -1,248 +0,0 @@
-#!/usr/bin/env python3
-"""Improve a skill description based on eval results.
-
-Takes eval results (from run_eval.py) and generates an improved description
-using Claude with extended thinking.
-"""
-
-import argparse
-import json
-import re
-import sys
-from pathlib import Path
-
-import anthropic
-
-from scripts.utils import parse_skill_md
-
-
-def improve_description(
-    client: anthropic.Anthropic,
-    skill_name: str,
-    skill_content: str,
-    current_description: str,
-    eval_results: dict,
-    history: list[dict],
-    model: str,
-    test_results: dict | None = None,
-    log_dir: Path | None = None,
-    iteration: int | None = None,
-) -> str:
-    """Call Claude to improve the description based on eval results."""
-    failed_triggers = [
-        r for r in eval_results["results"]
-        if r["should_trigger"] and not r["pass"]
-    ]
-    false_triggers = [
-        r for r in eval_results["results"]
-        if not r["should_trigger"] and not r["pass"]
-    ]
-
-    # Build scores summary
-    train_score = f"{eval_results['summary']['passed']}/{eval_results['summary']['total']}"
-    if test_results:
-        test_score = f"{test_results['summary']['passed']}/{test_results['summary']['total']}"
-        scores_summary = f"Train: {train_score}, Test: {test_score}"
-    else:
-        scores_summary = f"Train: {train_score}"
-
-    prompt = f"""You are optimizing a skill description for a Claude Code skill called "{skill_name}". A "skill" is sort of like a prompt, but with progressive disclosure -- there's a title and description that Claude sees when deciding whether to use the skill, and then if it does use the skill, it reads the .md file which has lots more details and potentially links to other resources in the skill folder like helper files and scripts and additional documentation or examples.
-
-The description appears in Claude's "available_skills" list. When a user sends a query, Claude decides whether to invoke the skill based solely on the title and on this description. Your goal is to write a description that triggers for relevant queries, and doesn't trigger for irrelevant ones.
-
-Here's the current description:
-<current_description>
-"{current_description}"
-</current_description>
-
-Current scores ({scores_summary}):
-<scores_summary>
-"""
-    if failed_triggers:
-        prompt += "FAILED TO TRIGGER (should have triggered but didn't):\n"
-        for r in failed_triggers:
-            prompt += f'  - "{r["query"]}" (triggered {r["triggers"]}/{r["runs"]} times)\n'
-        prompt += "\n"
-
-    if false_triggers:
-        prompt += "FALSE TRIGGERS (triggered but shouldn't have):\n"
-        for r in false_triggers:
-            prompt += f'  - "{r["query"]}" (triggered {r["triggers"]}/{r["runs"]} times)\n'
-        prompt += "\n"
-
-    if history:
-        prompt += "PREVIOUS ATTEMPTS (do NOT repeat these — try something structurally different):\n\n"
-        for h in history:
-            train_s = f"{h.get('train_passed', h.get('passed', 0))}/{h.get('train_total', h.get('total', 0))}"
-            test_s = f"{h.get('test_passed', '?')}/{h.get('test_total', '?')}" if h.get('test_passed') is not None else None
-            score_str = f"train={train_s}" + (f", test={test_s}" if test_s else "")
-            prompt += f'<attempt {score_str}>\n'
-            prompt += f'Description: "{h["description"]}"\n'
-            if "results" in h:
-                prompt += "Train results:\n"
-                for r in h["results"]:
-                    status = "PASS" if r["pass"] else "FAIL"
-                    prompt += f'  [{status}] "{r["query"][:80]}" (triggered {r["triggers"]}/{r["runs"]})\n'
-            if h.get("note"):
-                prompt += f'Note: {h["note"]}\n'
-            prompt += "</attempt>\n\n"
-
-    prompt += f"""</scores_summary>
-
-Skill content (for context on what the skill does):
-<skill_content>
-{skill_content}
-</skill_content>
-
-Based on the failures, write a new and improved description that is more likely to trigger correctly. When I say "based on the failures", it's a bit of a tricky line to walk because we don't want to overfit to the specific cases you're seeing. So what I DON'T want you to do is produce an ever-expanding list of specific queries that this skill should or shouldn't trigger for. Instead, try to generalize from the failures to broader categories of user intent and situations where this skill would be useful or not useful. The reason for this is twofold:
-
-1. Avoid overfitting
-2. The list might get loooong and it's injected into ALL queries and there might be a lot of skills, so we don't want to blow too much space on any given description.
-
-Concretely, your description should not be more than about 100-200 words, even if that comes at the cost of accuracy.
-
-Here are some tips that we've found to work well in writing these descriptions:
-- The skill should be phrased in the imperative -- "Use this skill for" rather than "this skill does"
-- The skill description should focus on the user's intent, what they are trying to achieve, vs. the implementation details of how the skill works.
-- The description competes with other skills for Claude's attention — make it distinctive and immediately recognizable.
-- If you're getting lots of failures after repeated attempts, change things up. Try different sentence structures or wordings.
-
-I'd encourage you to be creative and mix up the style in different iterations since you'll have multiple opportunities to try different approaches and we'll just grab the highest-scoring one at the end. 
-
-Please respond with only the new description text in <new_description> tags, nothing else."""
-
-    response = client.messages.create(
-        model=model,
-        max_tokens=16000,
-        thinking={
-            "type": "enabled",
-            "budget_tokens": 10000,
-        },
-        messages=[{"role": "user", "content": prompt}],
-    )
-
-    # Extract thinking and text from response
-    thinking_text = ""
-    text = ""
-    for block in response.content:
-        if block.type == "thinking":
-            thinking_text = block.thinking
-        elif block.type == "text":
-            text = block.text
-
-    # Parse out the <new_description> tags
-    match = re.search(r"<new_description>(.*?)</new_description>", text, re.DOTALL)
-    description = match.group(1).strip().strip('"') if match else text.strip().strip('"')
-
-    # Log the transcript
-    transcript: dict = {
-        "iteration": iteration,
-        "prompt": prompt,
-        "thinking": thinking_text,
-        "response": text,
-        "parsed_description": description,
-        "char_count": len(description),
-        "over_limit": len(description) > 1024,
-    }
-
-    # If over 1024 chars, ask the model to shorten it
-    if len(description) > 1024:
-        shorten_prompt = f"Your description is {len(description)} characters, which exceeds the hard 1024 character limit. Please rewrite it to be under 1024 characters while preserving the most important trigger words and intent coverage. Respond with only the new description in <new_description> tags."
-        shorten_response = client.messages.create(
-            model=model,
-            max_tokens=16000,
-            thinking={
-                "type": "enabled",
-                "budget_tokens": 10000,
-            },
-            messages=[
-                {"role": "user", "content": prompt},
-                {"role": "assistant", "content": text},
-                {"role": "user", "content": shorten_prompt},
-            ],
-        )
-
-        shorten_thinking = ""
-        shorten_text = ""
-        for block in shorten_response.content:
-            if block.type == "thinking":
-                shorten_thinking = block.thinking
-            elif block.type == "text":
-                shorten_text = block.text
-
-        match = re.search(r"<new_description>(.*?)</new_description>", shorten_text, re.DOTALL)
-        shortened = match.group(1).strip().strip('"') if match else shorten_text.strip().strip('"')
-
-        transcript["rewrite_prompt"] = shorten_prompt
-        transcript["rewrite_thinking"] = shorten_thinking
-        transcript["rewrite_response"] = shorten_text
-        transcript["rewrite_description"] = shortened
-        transcript["rewrite_char_count"] = len(shortened)
-        description = shortened
-
-    transcript["final_description"] = description
-
-    if log_dir:
-        log_dir.mkdir(parents=True, exist_ok=True)
-        log_file = log_dir / f"improve_iter_{iteration or 'unknown'}.json"
-        log_file.write_text(json.dumps(transcript, indent=2))
-
-    return description
-
-
-def main():
-    parser = argparse.ArgumentParser(description="Improve a skill description based on eval results")
-    parser.add_argument("--eval-results", required=True, help="Path to eval results JSON (from run_eval.py)")
-    parser.add_argument("--skill-path", required=True, help="Path to skill directory")
-    parser.add_argument("--history", default=None, help="Path to history JSON (previous attempts)")
-    parser.add_argument("--model", required=True, help="Model for improvement")
-    parser.add_argument("--verbose", action="store_true", help="Print thinking to stderr")
-    args = parser.parse_args()
-
-    skill_path = Path(args.skill_path)
-    if not (skill_path / "SKILL.md").exists():
-        print(f"Error: No SKILL.md found at {skill_path}", file=sys.stderr)
-        sys.exit(1)
-
-    eval_results = json.loads(Path(args.eval_results).read_text())
-    history = []
-    if args.history:
-        history = json.loads(Path(args.history).read_text())
-
-    name, _, content = parse_skill_md(skill_path)
-    current_description = eval_results["description"]
-
-    if args.verbose:
-        print(f"Current: {current_description}", file=sys.stderr)
-        print(f"Score: {eval_results['summary']['passed']}/{eval_results['summary']['total']}", file=sys.stderr)
-
-    client = anthropic.Anthropic()
-    new_description = improve_description(
-        client=client,
-        skill_name=name,
-        skill_content=content,
-        current_description=current_description,
-        eval_results=eval_results,
-        history=history,
-        model=args.model,
-    )
-
-    if args.verbose:
-        print(f"Improved: {new_description}", file=sys.stderr)
-
-    # Output as JSON with both the new description and updated history
-    output = {
-        "description": new_description,
-        "history": history + [{
-            "description": current_description,
-            "passed": eval_results["summary"]["passed"],
-            "failed": eval_results["summary"]["failed"],
-            "total": eval_results["summary"]["total"],
-            "results": eval_results["results"],
-        }],
-    }
-    print(json.dumps(output, indent=2))
-
-
-if __name__ == "__main__":
-    main()
diff --git a/.github/skills/skill-creator/scripts/package_skill.py b/.github/skills/skill-creator/scripts/package_skill.py
deleted file mode 100644
index f48eac4..0000000
--- a/.github/skills/skill-creator/scripts/package_skill.py
+++ /dev/null
@@ -1,136 +0,0 @@
-#!/usr/bin/env python3
-"""
-Skill Packager - Creates a distributable .skill file of a skill folder
-
-Usage:
-    python utils/package_skill.py <path/to/skill-folder> [output-directory]
-
-Example:
-    python utils/package_skill.py skills/public/my-skill
-    python utils/package_skill.py skills/public/my-skill ./dist
-"""
-
-import fnmatch
-import sys
-import zipfile
-from pathlib import Path
-from scripts.quick_validate import validate_skill
-
-# Patterns to exclude when packaging skills.
-EXCLUDE_DIRS = {"__pycache__", "node_modules"}
-EXCLUDE_GLOBS = {"*.pyc"}
-EXCLUDE_FILES = {".DS_Store"}
-# Directories excluded only at the skill root (not when nested deeper).
-ROOT_EXCLUDE_DIRS = {"evals"}
-
-
-def should_exclude(rel_path: Path) -> bool:
-    """Check if a path should be excluded from packaging."""
-    parts = rel_path.parts
-    if any(part in EXCLUDE_DIRS for part in parts):
-        return True
-    # rel_path is relative to skill_path.parent, so parts[0] is the skill
-    # folder name and parts[1] (if present) is the first subdir.
-    if len(parts) > 1 and parts[1] in ROOT_EXCLUDE_DIRS:
-        return True
-    name = rel_path.name
-    if name in EXCLUDE_FILES:
-        return True
-    return any(fnmatch.fnmatch(name, pat) for pat in EXCLUDE_GLOBS)
-
-
-def package_skill(skill_path, output_dir=None):
-    """
-    Package a skill folder into a .skill file.
-
-    Args:
-        skill_path: Path to the skill folder
-        output_dir: Optional output directory for the .skill file (defaults to current directory)
-
-    Returns:
-        Path to the created .skill file, or None if error
-    """
-    skill_path = Path(skill_path).resolve()
-
-    # Validate skill folder exists
-    if not skill_path.exists():
-        print(f"❌ Error: Skill folder not found: {skill_path}")
-        return None
-
-    if not skill_path.is_dir():
-        print(f"❌ Error: Path is not a directory: {skill_path}")
-        return None
-
-    # Validate SKILL.md exists
-    skill_md = skill_path / "SKILL.md"
-    if not skill_md.exists():
-        print(f"❌ Error: SKILL.md not found in {skill_path}")
-        return None
-
-    # Run validation before packaging
-    print("🔍 Validating skill...")
-    valid, message = validate_skill(skill_path)
-    if not valid:
-        print(f"❌ Validation failed: {message}")
-        print("   Please fix the validation errors before packaging.")
-        return None
-    print(f"✅ {message}\n")
-
-    # Determine output location
-    skill_name = skill_path.name
-    if output_dir:
-        output_path = Path(output_dir).resolve()
-        output_path.mkdir(parents=True, exist_ok=True)
-    else:
-        output_path = Path.cwd()
-
-    skill_filename = output_path / f"{skill_name}.skill"
-
-    # Create the .skill file (zip format)
-    try:
-        with zipfile.ZipFile(skill_filename, 'w', zipfile.ZIP_DEFLATED) as zipf:
-            # Walk through the skill directory, excluding build artifacts
-            for file_path in skill_path.rglob('*'):
-                if not file_path.is_file():
-                    continue
-                arcname = file_path.relative_to(skill_path.parent)
-                if should_exclude(arcname):
-                    print(f"  Skipped: {arcname}")
-                    continue
-                zipf.write(file_path, arcname)
-                print(f"  Added: {arcname}")
-
-        print(f"\n✅ Successfully packaged skill to: {skill_filename}")
-        return skill_filename
-
-    except Exception as e:
-        print(f"❌ Error creating .skill file: {e}")
-        return None
-
-
-def main():
-    if len(sys.argv) < 2:
-        print("Usage: python utils/package_skill.py <path/to/skill-folder> [output-directory]")
-        print("\nExample:")
-        print("  python utils/package_skill.py skills/public/my-skill")
-        print("  python utils/package_skill.py skills/public/my-skill ./dist")
-        sys.exit(1)
-
-    skill_path = sys.argv[1]
-    output_dir = sys.argv[2] if len(sys.argv) > 2 else None
-
-    print(f"📦 Packaging skill: {skill_path}")
-    if output_dir:
-        print(f"   Output directory: {output_dir}")
-    print()
-
-    result = package_skill(skill_path, output_dir)
-
-    if result:
-        sys.exit(0)
-    else:
-        sys.exit(1)
-
-
-if __name__ == "__main__":
-    main()
diff --git a/.github/skills/skill-creator/scripts/quick_validate.py b/.github/skills/skill-creator/scripts/quick_validate.py
deleted file mode 100644
index ed8e1dd..0000000
--- a/.github/skills/skill-creator/scripts/quick_validate.py
+++ /dev/null
@@ -1,103 +0,0 @@
-#!/usr/bin/env python3
-"""
-Quick validation script for skills - minimal version
-"""
-
-import sys
-import os
-import re
-import yaml
-from pathlib import Path
-
-def validate_skill(skill_path):
-    """Basic validation of a skill"""
-    skill_path = Path(skill_path)
-
-    # Check SKILL.md exists
-    skill_md = skill_path / 'SKILL.md'
-    if not skill_md.exists():
-        return False, "SKILL.md not found"
-
-    # Read and validate frontmatter
-    content = skill_md.read_text()
-    if not content.startswith('---'):
-        return False, "No YAML frontmatter found"
-
-    # Extract frontmatter
-    match = re.match(r'^---\n(.*?)\n---', content, re.DOTALL)
-    if not match:
-        return False, "Invalid frontmatter format"
-
-    frontmatter_text = match.group(1)
-
-    # Parse YAML frontmatter
-    try:
-        frontmatter = yaml.safe_load(frontmatter_text)
-        if not isinstance(frontmatter, dict):
-            return False, "Frontmatter must be a YAML dictionary"
-    except yaml.YAMLError as e:
-        return False, f"Invalid YAML in frontmatter: {e}"
-
-    # Define allowed properties
-    ALLOWED_PROPERTIES = {'name', 'description', 'license', 'allowed-tools', 'metadata', 'compatibility'}
-
-    # Check for unexpected properties (excluding nested keys under metadata)
-    unexpected_keys = set(frontmatter.keys()) - ALLOWED_PROPERTIES
-    if unexpected_keys:
-        return False, (
-            f"Unexpected key(s) in SKILL.md frontmatter: {', '.join(sorted(unexpected_keys))}. "
-            f"Allowed properties are: {', '.join(sorted(ALLOWED_PROPERTIES))}"
-        )
-
-    # Check required fields
-    if 'name' not in frontmatter:
-        return False, "Missing 'name' in frontmatter"
-    if 'description' not in frontmatter:
-        return False, "Missing 'description' in frontmatter"
-
-    # Extract name for validation
-    name = frontmatter.get('name', '')
-    if not isinstance(name, str):
-        return False, f"Name must be a string, got {type(name).__name__}"
-    name = name.strip()
-    if name:
-        # Check naming convention (kebab-case: lowercase with hyphens)
-        if not re.match(r'^[a-z0-9-]+$', name):
-            return False, f"Name '{name}' should be kebab-case (lowercase letters, digits, and hyphens only)"
-        if name.startswith('-') or name.endswith('-') or '--' in name:
-            return False, f"Name '{name}' cannot start/end with hyphen or contain consecutive hyphens"
-        # Check name length (max 64 characters per spec)
-        if len(name) > 64:
-            return False, f"Name is too long ({len(name)} characters). Maximum is 64 characters."
-
-    # Extract and validate description
-    description = frontmatter.get('description', '')
-    if not isinstance(description, str):
-        return False, f"Description must be a string, got {type(description).__name__}"
-    description = description.strip()
-    if description:
-        # Check for angle brackets
-        if '<' in description or '>' in description:
-            return False, "Description cannot contain angle brackets (< or >)"
-        # Check description length (max 1024 characters per spec)
-        if len(description) > 1024:
-            return False, f"Description is too long ({len(description)} characters). Maximum is 1024 characters."
-
-    # Validate compatibility field if present (optional)
-    compatibility = frontmatter.get('compatibility', '')
-    if compatibility:
-        if not isinstance(compatibility, str):
-            return False, f"Compatibility must be a string, got {type(compatibility).__name__}"
-        if len(compatibility) > 500:
-            return False, f"Compatibility is too long ({len(compatibility)} characters). Maximum is 500 characters."
-
-    return True, "Skill is valid!"
-
-if __name__ == "__main__":
-    if len(sys.argv) != 2:
-        print("Usage: python quick_validate.py <skill_directory>")
-        sys.exit(1)
-    
-    valid, message = validate_skill(sys.argv[1])
-    print(message)
-    sys.exit(0 if valid else 1)
\ No newline at end of file
diff --git a/.github/skills/skill-creator/scripts/restructure_evals.py b/.github/skills/skill-creator/scripts/restructure_evals.py
deleted file mode 100644
index d5a34ea..0000000
--- a/.github/skills/skill-creator/scripts/restructure_evals.py
+++ /dev/null
@@ -1,42 +0,0 @@
-#!/usr/bin/env python3
-"""Restructure eval workspace: move grading.json into run-1/ dirs and add summary fields."""
-import json
-import shutil
-from pathlib import Path
-
-base = Path("/Volumes/sourcecode/repos/cratis/Documentation/.github/skills/skills-eval-workspace/iteration-1")
-
-for grading_file in sorted(base.rglob("grading.json")):
-    parent = grading_file.parent
-    if parent.name not in ("with_skill", "without_skill"):
-        continue
-
-    with open(grading_file) as f:
-        grading = json.load(f)
-
-    expectations = grading.get("expectations", [])
-    passed = sum(1 for e in expectations if e.get("passed", False))
-    failed = len(expectations) - passed
-    total = len(expectations)
-    pass_rate = round(passed / total, 4) if total > 0 else 0.0
-
-    grading["summary"] = {
-        "pass_rate": pass_rate,
-        "passed": passed,
-        "failed": failed,
-        "total": total
-    }
-
-    run_dir = parent / "run-1"
-    run_dir.mkdir(exist_ok=True)
-
-    with open(run_dir / "grading.json", "w") as f:
-        json.dump(grading, f, indent=2)
-
-    timing_file = parent / "timing.json"
-    if timing_file.exists():
-        shutil.copy(timing_file, run_dir / "timing.json")
-
-    print(f"OK {parent.parent.parent.name}/{parent.parent.name}/{parent.name} pass_rate={pass_rate} ({passed}/{total})")
-
-print("Done!")
diff --git a/.github/skills/skill-creator/scripts/run_eval.py b/.github/skills/skill-creator/scripts/run_eval.py
deleted file mode 100644
index e58c70b..0000000
--- a/.github/skills/skill-creator/scripts/run_eval.py
+++ /dev/null
@@ -1,310 +0,0 @@
-#!/usr/bin/env python3
-"""Run trigger evaluation for a skill description.
-
-Tests whether a skill's description causes Claude to trigger (read the skill)
-for a set of queries. Outputs results as JSON.
-"""
-
-import argparse
-import json
-import os
-import select
-import subprocess
-import sys
-import time
-import uuid
-from concurrent.futures import ProcessPoolExecutor, as_completed
-from pathlib import Path
-
-from scripts.utils import parse_skill_md
-
-
-def find_project_root() -> Path:
-    """Find the project root by walking up from cwd looking for .claude/.
-
-    Mimics how Claude Code discovers its project root, so the command file
-    we create ends up where claude -p will look for it.
-    """
-    current = Path.cwd()
-    for parent in [current, *current.parents]:
-        if (parent / ".claude").is_dir():
-            return parent
-    return current
-
-
-def run_single_query(
-    query: str,
-    skill_name: str,
-    skill_description: str,
-    timeout: int,
-    project_root: str,
-    model: str | None = None,
-) -> bool:
-    """Run a single query and return whether the skill was triggered.
-
-    Creates a command file in .claude/commands/ so it appears in Claude's
-    available_skills list, then runs `claude -p` with the raw query.
-    Uses --include-partial-messages to detect triggering early from
-    stream events (content_block_start) rather than waiting for the
-    full assistant message, which only arrives after tool execution.
-    """
-    unique_id = uuid.uuid4().hex[:8]
-    clean_name = f"{skill_name}-skill-{unique_id}"
-    project_commands_dir = Path(project_root) / ".claude" / "commands"
-    command_file = project_commands_dir / f"{clean_name}.md"
-
-    try:
-        project_commands_dir.mkdir(parents=True, exist_ok=True)
-        # Use YAML block scalar to avoid breaking on quotes in description
-        indented_desc = "\n  ".join(skill_description.split("\n"))
-        command_content = (
-            f"---\n"
-            f"description: |\n"
-            f"  {indented_desc}\n"
-            f"---\n\n"
-            f"# {skill_name}\n\n"
-            f"This skill handles: {skill_description}\n"
-        )
-        command_file.write_text(command_content)
-
-        cmd = [
-            "claude",
-            "-p", query,
-            "--output-format", "stream-json",
-            "--verbose",
-            "--include-partial-messages",
-        ]
-        if model:
-            cmd.extend(["--model", model])
-
-        # Remove CLAUDECODE env var to allow nesting claude -p inside a
-        # Claude Code session. The guard is for interactive terminal conflicts;
-        # programmatic subprocess usage is safe.
-        env = {k: v for k, v in os.environ.items() if k != "CLAUDECODE"}
-
-        process = subprocess.Popen(
-            cmd,
-            stdout=subprocess.PIPE,
-            stderr=subprocess.DEVNULL,
-            cwd=project_root,
-            env=env,
-        )
-
-        triggered = False
-        start_time = time.time()
-        buffer = ""
-        # Track state for stream event detection
-        pending_tool_name = None
-        accumulated_json = ""
-
-        try:
-            while time.time() - start_time < timeout:
-                if process.poll() is not None:
-                    remaining = process.stdout.read()
-                    if remaining:
-                        buffer += remaining.decode("utf-8", errors="replace")
-                    break
-
-                ready, _, _ = select.select([process.stdout], [], [], 1.0)
-                if not ready:
-                    continue
-
-                chunk = os.read(process.stdout.fileno(), 8192)
-                if not chunk:
-                    break
-                buffer += chunk.decode("utf-8", errors="replace")
-
-                while "\n" in buffer:
-                    line, buffer = buffer.split("\n", 1)
-                    line = line.strip()
-                    if not line:
-                        continue
-
-                    try:
-                        event = json.loads(line)
-                    except json.JSONDecodeError:
-                        continue
-
-                    # Early detection via stream events
-                    if event.get("type") == "stream_event":
-                        se = event.get("event", {})
-                        se_type = se.get("type", "")
-
-                        if se_type == "content_block_start":
-                            cb = se.get("content_block", {})
-                            if cb.get("type") == "tool_use":
-                                tool_name = cb.get("name", "")
-                                if tool_name in ("Skill", "Read"):
-                                    pending_tool_name = tool_name
-                                    accumulated_json = ""
-                                else:
-                                    return False
-
-                        elif se_type == "content_block_delta" and pending_tool_name:
-                            delta = se.get("delta", {})
-                            if delta.get("type") == "input_json_delta":
-                                accumulated_json += delta.get("partial_json", "")
-                                if clean_name in accumulated_json:
-                                    return True
-
-                        elif se_type in ("content_block_stop", "message_stop"):
-                            if pending_tool_name:
-                                return clean_name in accumulated_json
-                            if se_type == "message_stop":
-                                return False
-
-                    # Fallback: full assistant message
-                    elif event.get("type") == "assistant":
-                        message = event.get("message", {})
-                        for content_item in message.get("content", []):
-                            if content_item.get("type") != "tool_use":
-                                continue
-                            tool_name = content_item.get("name", "")
-                            tool_input = content_item.get("input", {})
-                            if tool_name == "Skill" and clean_name in tool_input.get("skill", ""):
-                                triggered = True
-                            elif tool_name == "Read" and clean_name in tool_input.get("file_path", ""):
-                                triggered = True
-                            return triggered
-
-                    elif event.get("type") == "result":
-                        return triggered
-        finally:
-            # Clean up process on any exit path (return, exception, timeout)
-            if process.poll() is None:
-                process.kill()
-                process.wait()
-
-        return triggered
-    finally:
-        if command_file.exists():
-            command_file.unlink()
-
-
-def run_eval(
-    eval_set: list[dict],
-    skill_name: str,
-    description: str,
-    num_workers: int,
-    timeout: int,
-    project_root: Path,
-    runs_per_query: int = 1,
-    trigger_threshold: float = 0.5,
-    model: str | None = None,
-) -> dict:
-    """Run the full eval set and return results."""
-    results = []
-
-    with ProcessPoolExecutor(max_workers=num_workers) as executor:
-        future_to_info = {}
-        for item in eval_set:
-            for run_idx in range(runs_per_query):
-                future = executor.submit(
-                    run_single_query,
-                    item["query"],
-                    skill_name,
-                    description,
-                    timeout,
-                    str(project_root),
-                    model,
-                )
-                future_to_info[future] = (item, run_idx)
-
-        query_triggers: dict[str, list[bool]] = {}
-        query_items: dict[str, dict] = {}
-        for future in as_completed(future_to_info):
-            item, _ = future_to_info[future]
-            query = item["query"]
-            query_items[query] = item
-            if query not in query_triggers:
-                query_triggers[query] = []
-            try:
-                query_triggers[query].append(future.result())
-            except Exception as e:
-                print(f"Warning: query failed: {e}", file=sys.stderr)
-                query_triggers[query].append(False)
-
-    for query, triggers in query_triggers.items():
-        item = query_items[query]
-        trigger_rate = sum(triggers) / len(triggers)
-        should_trigger = item["should_trigger"]
-        if should_trigger:
-            did_pass = trigger_rate >= trigger_threshold
-        else:
-            did_pass = trigger_rate < trigger_threshold
-        results.append({
-            "query": query,
-            "should_trigger": should_trigger,
-            "trigger_rate": trigger_rate,
-            "triggers": sum(triggers),
-            "runs": len(triggers),
-            "pass": did_pass,
-        })
-
-    passed = sum(1 for r in results if r["pass"])
-    total = len(results)
-
-    return {
-        "skill_name": skill_name,
-        "description": description,
-        "results": results,
-        "summary": {
-            "total": total,
-            "passed": passed,
-            "failed": total - passed,
-        },
-    }
-
-
-def main():
-    parser = argparse.ArgumentParser(description="Run trigger evaluation for a skill description")
-    parser.add_argument("--eval-set", required=True, help="Path to eval set JSON file")
-    parser.add_argument("--skill-path", required=True, help="Path to skill directory")
-    parser.add_argument("--description", default=None, help="Override description to test")
-    parser.add_argument("--num-workers", type=int, default=10, help="Number of parallel workers")
-    parser.add_argument("--timeout", type=int, default=30, help="Timeout per query in seconds")
-    parser.add_argument("--runs-per-query", type=int, default=3, help="Number of runs per query")
-    parser.add_argument("--trigger-threshold", type=float, default=0.5, help="Trigger rate threshold")
-    parser.add_argument("--model", default=None, help="Model to use for claude -p (default: user's configured model)")
-    parser.add_argument("--verbose", action="store_true", help="Print progress to stderr")
-    args = parser.parse_args()
-
-    eval_set = json.loads(Path(args.eval_set).read_text())
-    skill_path = Path(args.skill_path)
-
-    if not (skill_path / "SKILL.md").exists():
-        print(f"Error: No SKILL.md found at {skill_path}", file=sys.stderr)
-        sys.exit(1)
-
-    name, original_description, content = parse_skill_md(skill_path)
-    description = args.description or original_description
-    project_root = find_project_root()
-
-    if args.verbose:
-        print(f"Evaluating: {description}", file=sys.stderr)
-
-    output = run_eval(
-        eval_set=eval_set,
-        skill_name=name,
-        description=description,
-        num_workers=args.num_workers,
-        timeout=args.timeout,
-        project_root=project_root,
-        runs_per_query=args.runs_per_query,
-        trigger_threshold=args.trigger_threshold,
-        model=args.model,
-    )
-
-    if args.verbose:
-        summary = output["summary"]
-        print(f"Results: {summary['passed']}/{summary['total']} passed", file=sys.stderr)
-        for r in output["results"]:
-            status = "PASS" if r["pass"] else "FAIL"
-            rate_str = f"{r['triggers']}/{r['runs']}"
-            print(f"  [{status}] rate={rate_str} expected={r['should_trigger']}: {r['query'][:70]}", file=sys.stderr)
-
-    print(json.dumps(output, indent=2))
-
-
-if __name__ == "__main__":
-    main()
diff --git a/.github/skills/skill-creator/scripts/run_loop.py b/.github/skills/skill-creator/scripts/run_loop.py
deleted file mode 100644
index 36f9b4e..0000000
--- a/.github/skills/skill-creator/scripts/run_loop.py
+++ /dev/null
@@ -1,332 +0,0 @@
-#!/usr/bin/env python3
-"""Run the eval + improve loop until all pass or max iterations reached.
-
-Combines run_eval.py and improve_description.py in a loop, tracking history
-and returning the best description found. Supports train/test split to prevent
-overfitting.
-"""
-
-import argparse
-import json
-import random
-import sys
-import tempfile
-import time
-import webbrowser
-from pathlib import Path
-
-import anthropic
-
-from scripts.generate_report import generate_html
-from scripts.improve_description import improve_description
-from scripts.run_eval import find_project_root, run_eval
-from scripts.utils import parse_skill_md
-
-
-def split_eval_set(eval_set: list[dict], holdout: float, seed: int = 42) -> tuple[list[dict], list[dict]]:
-    """Split eval set into train and test sets, stratified by should_trigger."""
-    random.seed(seed)
-
-    # Separate by should_trigger
-    trigger = [e for e in eval_set if e["should_trigger"]]
-    no_trigger = [e for e in eval_set if not e["should_trigger"]]
-
-    # Shuffle each group
-    random.shuffle(trigger)
-    random.shuffle(no_trigger)
-
-    # Calculate split points
-    n_trigger_test = max(1, int(len(trigger) * holdout))
-    n_no_trigger_test = max(1, int(len(no_trigger) * holdout))
-
-    # Split
-    test_set = trigger[:n_trigger_test] + no_trigger[:n_no_trigger_test]
-    train_set = trigger[n_trigger_test:] + no_trigger[n_no_trigger_test:]
-
-    return train_set, test_set
-
-
-def run_loop(
-    eval_set: list[dict],
-    skill_path: Path,
-    description_override: str | None,
-    num_workers: int,
-    timeout: int,
-    max_iterations: int,
-    runs_per_query: int,
-    trigger_threshold: float,
-    holdout: float,
-    model: str,
-    verbose: bool,
-    live_report_path: Path | None = None,
-    log_dir: Path | None = None,
-) -> dict:
-    """Run the eval + improvement loop."""
-    project_root = find_project_root()
-    name, original_description, content = parse_skill_md(skill_path)
-    current_description = description_override or original_description
-
-    # Split into train/test if holdout > 0
-    if holdout > 0:
-        train_set, test_set = split_eval_set(eval_set, holdout)
-        if verbose:
-            print(f"Split: {len(train_set)} train, {len(test_set)} test (holdout={holdout})", file=sys.stderr)
-    else:
-        train_set = eval_set
-        test_set = []
-
-    client = anthropic.Anthropic()
-    history = []
-    exit_reason = "unknown"
-
-    for iteration in range(1, max_iterations + 1):
-        if verbose:
-            print(f"\n{'='*60}", file=sys.stderr)
-            print(f"Iteration {iteration}/{max_iterations}", file=sys.stderr)
-            print(f"Description: {current_description}", file=sys.stderr)
-            print(f"{'='*60}", file=sys.stderr)
-
-        # Evaluate train + test together in one batch for parallelism
-        all_queries = train_set + test_set
-        t0 = time.time()
-        all_results = run_eval(
-            eval_set=all_queries,
-            skill_name=name,
-            description=current_description,
-            num_workers=num_workers,
-            timeout=timeout,
-            project_root=project_root,
-            runs_per_query=runs_per_query,
-            trigger_threshold=trigger_threshold,
-            model=model,
-        )
-        eval_elapsed = time.time() - t0
-
-        # Split results back into train/test by matching queries
-        train_queries_set = {q["query"] for q in train_set}
-        train_result_list = [r for r in all_results["results"] if r["query"] in train_queries_set]
-        test_result_list = [r for r in all_results["results"] if r["query"] not in train_queries_set]
-
-        train_passed = sum(1 for r in train_result_list if r["pass"])
-        train_total = len(train_result_list)
-        train_summary = {"passed": train_passed, "failed": train_total - train_passed, "total": train_total}
-        train_results = {"results": train_result_list, "summary": train_summary}
-
-        if test_set:
-            test_passed = sum(1 for r in test_result_list if r["pass"])
-            test_total = len(test_result_list)
-            test_summary = {"passed": test_passed, "failed": test_total - test_passed, "total": test_total}
-            test_results = {"results": test_result_list, "summary": test_summary}
-        else:
-            test_results = None
-            test_summary = None
-
-        history.append({
-            "iteration": iteration,
-            "description": current_description,
-            "train_passed": train_summary["passed"],
-            "train_failed": train_summary["failed"],
-            "train_total": train_summary["total"],
-            "train_results": train_results["results"],
-            "test_passed": test_summary["passed"] if test_summary else None,
-            "test_failed": test_summary["failed"] if test_summary else None,
-            "test_total": test_summary["total"] if test_summary else None,
-            "test_results": test_results["results"] if test_results else None,
-            # For backward compat with report generator
-            "passed": train_summary["passed"],
-            "failed": train_summary["failed"],
-            "total": train_summary["total"],
-            "results": train_results["results"],
-        })
-
-        # Write live report if path provided
-        if live_report_path:
-            partial_output = {
-                "original_description": original_description,
-                "best_description": current_description,
-                "best_score": "in progress",
-                "iterations_run": len(history),
-                "holdout": holdout,
-                "train_size": len(train_set),
-                "test_size": len(test_set),
-                "history": history,
-            }
-            live_report_path.write_text(generate_html(partial_output, auto_refresh=True, skill_name=name))
-
-        if verbose:
-            def print_eval_stats(label, results, elapsed):
-                pos = [r for r in results if r["should_trigger"]]
-                neg = [r for r in results if not r["should_trigger"]]
-                tp = sum(r["triggers"] for r in pos)
-                pos_runs = sum(r["runs"] for r in pos)
-                fn = pos_runs - tp
-                fp = sum(r["triggers"] for r in neg)
-                neg_runs = sum(r["runs"] for r in neg)
-                tn = neg_runs - fp
-                total = tp + tn + fp + fn
-                precision = tp / (tp + fp) if (tp + fp) > 0 else 1.0
-                recall = tp / (tp + fn) if (tp + fn) > 0 else 1.0
-                accuracy = (tp + tn) / total if total > 0 else 0.0
-                print(f"{label}: {tp+tn}/{total} correct, precision={precision:.0%} recall={recall:.0%} accuracy={accuracy:.0%} ({elapsed:.1f}s)", file=sys.stderr)
-                for r in results:
-                    status = "PASS" if r["pass"] else "FAIL"
-                    rate_str = f"{r['triggers']}/{r['runs']}"
-                    print(f"  [{status}] rate={rate_str} expected={r['should_trigger']}: {r['query'][:60]}", file=sys.stderr)
-
-            print_eval_stats("Train", train_results["results"], eval_elapsed)
-            if test_summary:
-                print_eval_stats("Test ", test_results["results"], 0)
-
-        if train_summary["failed"] == 0:
-            exit_reason = f"all_passed (iteration {iteration})"
-            if verbose:
-                print(f"\nAll train queries passed on iteration {iteration}!", file=sys.stderr)
-            break
-
-        if iteration == max_iterations:
-            exit_reason = f"max_iterations ({max_iterations})"
-            if verbose:
-                print(f"\nMax iterations reached ({max_iterations}).", file=sys.stderr)
-            break
-
-        # Improve the description based on train results
-        if verbose:
-            print(f"\nImproving description...", file=sys.stderr)
-
-        t0 = time.time()
-        # Strip test scores from history so improvement model can't see them
-        blinded_history = [
-            {k: v for k, v in h.items() if not k.startswith("test_")}
-            for h in history
-        ]
-        new_description = improve_description(
-            client=client,
-            skill_name=name,
-            skill_content=content,
-            current_description=current_description,
-            eval_results=train_results,
-            history=blinded_history,
-            model=model,
-            log_dir=log_dir,
-            iteration=iteration,
-        )
-        improve_elapsed = time.time() - t0
-
-        if verbose:
-            print(f"Proposed ({improve_elapsed:.1f}s): {new_description}", file=sys.stderr)
-
-        current_description = new_description
-
-    # Find the best iteration by TEST score (or train if no test set)
-    if test_set:
-        best = max(history, key=lambda h: h["test_passed"] or 0)
-        best_score = f"{best['test_passed']}/{best['test_total']}"
-    else:
-        best = max(history, key=lambda h: h["train_passed"])
-        best_score = f"{best['train_passed']}/{best['train_total']}"
-
-    if verbose:
-        print(f"\nExit reason: {exit_reason}", file=sys.stderr)
-        print(f"Best score: {best_score} (iteration {best['iteration']})", file=sys.stderr)
-
-    return {
-        "exit_reason": exit_reason,
-        "original_description": original_description,
-        "best_description": best["description"],
-        "best_score": best_score,
-        "best_train_score": f"{best['train_passed']}/{best['train_total']}",
-        "best_test_score": f"{best['test_passed']}/{best['test_total']}" if test_set else None,
-        "final_description": current_description,
-        "iterations_run": len(history),
-        "holdout": holdout,
-        "train_size": len(train_set),
-        "test_size": len(test_set),
-        "history": history,
-    }
-
-
-def main():
-    parser = argparse.ArgumentParser(description="Run eval + improve loop")
-    parser.add_argument("--eval-set", required=True, help="Path to eval set JSON file")
-    parser.add_argument("--skill-path", required=True, help="Path to skill directory")
-    parser.add_argument("--description", default=None, help="Override starting description")
-    parser.add_argument("--num-workers", type=int, default=10, help="Number of parallel workers")
-    parser.add_argument("--timeout", type=int, default=30, help="Timeout per query in seconds")
-    parser.add_argument("--max-iterations", type=int, default=5, help="Max improvement iterations")
-    parser.add_argument("--runs-per-query", type=int, default=3, help="Number of runs per query")
-    parser.add_argument("--trigger-threshold", type=float, default=0.5, help="Trigger rate threshold")
-    parser.add_argument("--holdout", type=float, default=0.4, help="Fraction of eval set to hold out for testing (0 to disable)")
-    parser.add_argument("--model", required=True, help="Model for improvement")
-    parser.add_argument("--verbose", action="store_true", help="Print progress to stderr")
-    parser.add_argument("--report", default="auto", help="Generate HTML report at this path (default: 'auto' for temp file, 'none' to disable)")
-    parser.add_argument("--results-dir", default=None, help="Save all outputs (results.json, report.html, log.txt) to a timestamped subdirectory here")
-    args = parser.parse_args()
-
-    eval_set = json.loads(Path(args.eval_set).read_text())
-    skill_path = Path(args.skill_path)
-
-    if not (skill_path / "SKILL.md").exists():
-        print(f"Error: No SKILL.md found at {skill_path}", file=sys.stderr)
-        sys.exit(1)
-
-    name, _, _ = parse_skill_md(skill_path)
-
-    # Set up live report path
-    if args.report != "none":
-        if args.report == "auto":
-            timestamp = time.strftime("%Y%m%d_%H%M%S")
-            live_report_path = Path(tempfile.gettempdir()) / f"skill_description_report_{skill_path.name}_{timestamp}.html"
-        else:
-            live_report_path = Path(args.report)
-        # Open the report immediately so the user can watch
-        live_report_path.write_text("<html><body><h1>Starting optimization loop...</h1><meta http-equiv='refresh' content='5'></body></html>")
-        webbrowser.open(str(live_report_path))
-    else:
-        live_report_path = None
-
-    # Determine output directory (create before run_loop so logs can be written)
-    if args.results_dir:
-        timestamp = time.strftime("%Y-%m-%d_%H%M%S")
-        results_dir = Path(args.results_dir) / timestamp
-        results_dir.mkdir(parents=True, exist_ok=True)
-    else:
-        results_dir = None
-
-    log_dir = results_dir / "logs" if results_dir else None
-
-    output = run_loop(
-        eval_set=eval_set,
-        skill_path=skill_path,
-        description_override=args.description,
-        num_workers=args.num_workers,
-        timeout=args.timeout,
-        max_iterations=args.max_iterations,
-        runs_per_query=args.runs_per_query,
-        trigger_threshold=args.trigger_threshold,
-        holdout=args.holdout,
-        model=args.model,
-        verbose=args.verbose,
-        live_report_path=live_report_path,
-        log_dir=log_dir,
-    )
-
-    # Save JSON output
-    json_output = json.dumps(output, indent=2)
-    print(json_output)
-    if results_dir:
-        (results_dir / "results.json").write_text(json_output)
-
-    # Write final HTML report (without auto-refresh)
-    if live_report_path:
-        live_report_path.write_text(generate_html(output, auto_refresh=False, skill_name=name))
-        print(f"\nReport: {live_report_path}", file=sys.stderr)
-
-    if results_dir and live_report_path:
-        (results_dir / "report.html").write_text(generate_html(output, auto_refresh=False, skill_name=name))
-
-    if results_dir:
-        print(f"Results saved to: {results_dir}", file=sys.stderr)
-
-
-if __name__ == "__main__":
-    main()
diff --git a/.github/skills/skill-creator/scripts/utils.py b/.github/skills/skill-creator/scripts/utils.py
deleted file mode 100644
index 51b6a07..0000000
--- a/.github/skills/skill-creator/scripts/utils.py
+++ /dev/null
@@ -1,47 +0,0 @@
-"""Shared utilities for skill-creator scripts."""
-
-from pathlib import Path
-
-
-
-def parse_skill_md(skill_path: Path) -> tuple[str, str, str]:
-    """Parse a SKILL.md file, returning (name, description, full_content)."""
-    content = (skill_path / "SKILL.md").read_text()
-    lines = content.split("\n")
-
-    if lines[0].strip() != "---":
-        raise ValueError("SKILL.md missing frontmatter (no opening ---)")
-
-    end_idx = None
-    for i, line in enumerate(lines[1:], start=1):
-        if line.strip() == "---":
-            end_idx = i
-            break
-
-    if end_idx is None:
-        raise ValueError("SKILL.md missing frontmatter (no closing ---)")
-
-    name = ""
-    description = ""
-    frontmatter_lines = lines[1:end_idx]
-    i = 0
-    while i < len(frontmatter_lines):
-        line = frontmatter_lines[i]
-        if line.startswith("name:"):
-            name = line[len("name:"):].strip().strip('"').strip("'")
-        elif line.startswith("description:"):
-            value = line[len("description:"):].strip()
-            # Handle YAML multiline indicators (>, |, >-, |-)
-            if value in (">", "|", ">-", "|-"):
-                continuation_lines: list[str] = []
-                i += 1
-                while i < len(frontmatter_lines) and (frontmatter_lines[i].startswith("  ") or frontmatter_lines[i].startswith("\t")):
-                    continuation_lines.append(frontmatter_lines[i].strip())
-                    i += 1
-                description = " ".join(continuation_lines)
-                continue
-            else:
-                description = value.strip('"').strip("'")
-        i += 1
-
-    return name, description, content
diff --git a/.github/skills/stepper-command-dialog/SKILL.md b/.github/skills/stepper-command-dialog/SKILL.md
deleted file mode 100644
index 736b068..0000000
--- a/.github/skills/stepper-command-dialog/SKILL.md
+++ /dev/null
@@ -1,234 +0,0 @@
----
-name: stepper-command-dialog
-description: Step-by-step guidance for building a multi-step wizard dialog (StepperCommandDialog) in a Cratis Arc application. Use whenever a command requires gathering information across multiple steps, implementing a wizard flow, breaking a complex form into named stages, or using StepperCommandDialog, StepperPanel, validateOnInit, or wizard-style navigation.
----
-
-# StepperCommandDialog — Wizard Dialogs
-
-`StepperCommandDialog` organizes a single command form across multiple named steps. Users navigate with **Previous** and **Next** buttons; **Submit** only appears on the last step when every field across all steps is valid.
-
-Use this instead of `CommandDialog` when:
-- The form has too many fields to show at once
-- Fields can be grouped into logical stages (e.g. "Contact Info → Project Details → Summary")
-- You want guided, linear input with per-step validation feedback
-- The operation feels like a wizard or an onboarding flow
-
----
-
-## Step 1 — Define the command
-
-A single command collects all fields across all steps. Each step contributes properties to the same command instance.
-
-```tsx
-// API/Projects/CreateProject.ts  (proxy-generated — run dotnet build first)
-// C# side:
-[Command]
-public record CreateProject(string Name, string Email, string Description, decimal Budget)
-{
-    public ProjectCreated Handle() => new(Name, Email, Description, Budget);
-}
-```
-
----
-
-## Step 2 — Build the dialog component
-
-```tsx
-import { StepperCommandDialog } from '@cratis/components/CommandDialog';
-import { StepperPanel } from 'primereact/stepperpanel';
-import { InputTextField, TextAreaField, NumberField } from '@cratis/components/CommandForm/fields';
-import { DialogResult, useDialogContext } from '@cratis/arc.react/dialogs';
-import { CreateProject } from '../api/Projects/CreateProject';
-
-const CreateProjectDialog = () => {
-    const { closeDialog } = useDialogContext();
-
-    return (
-        <StepperCommandDialog<CreateProject>
-            command={CreateProject}
-            title="Create New Project"
-            okLabel="Create"
-            onConfirm={() => closeDialog(DialogResult.Ok)}
-            onCancel={() => closeDialog(DialogResult.Cancelled)}
-        >
-            <StepperPanel header="Contact Info">
-                <InputTextField<CreateProject>
-                    value={c => c.email}
-                    title="Contact Email"
-                    placeholder="Enter contact email"
-                    type="email"
-                />
-            </StepperPanel>
-            <StepperPanel header="Project Details">
-                <InputTextField<CreateProject>
-                    value={c => c.name}
-                    title="Project Name"
-                    placeholder="Enter project name"
-                />
-                <TextAreaField<CreateProject>
-                    value={c => c.description}
-                    title="Description"
-                    placeholder="Describe the project"
-                    rows={4}
-                />
-            </StepperPanel>
-            <StepperPanel header="Budget">
-                <NumberField<CreateProject>
-                    value={c => c.budget}
-                    title="Budget"
-                    placeholder="Enter budget"
-                />
-            </StepperPanel>
-        </StepperCommandDialog>
-    );
-};
-```
-
-**Rules:**
-- Each `StepperPanel` takes a `header` string — this is the step label shown in the wizard navigation bar
-- All `CommandForm` fields inside any `StepperPanel` are bound to the **same** command instance
-- Fields map to command properties via the `value={c => c.propertyName}` accessor
-- The `Next` button is disabled while the current step has validation errors
-- `Submit` only appears on the **last** step when all fields (across all steps) are valid
-
----
-
-## Step 3 — Wire the dialog to a parent component
-
-```tsx
-import { useDialog } from '@cratis/arc.react/dialogs';
-
-export const ProjectsPage = () => {
-    const [CreateProjectDialogWrapper, showCreateProject] = useDialog(CreateProjectDialog);
-
-    return (
-        <>
-            <button
-                className="p-button p-component"
-                onClick={() => showCreateProject()}
-            >
-                New Project
-            </button>
-            <CreateProjectDialogWrapper />
-        </>
-    );
-};
-```
-
----
-
-## Navigation behavior
-
-| Step | Footer buttons shown |
-|------|---------------------|
-| First step | **Next** |
-| Middle step | **Previous**, **Next** |
-| Last step (any step invalid) | **Previous** |
-| Last step (all valid) | **Previous**, **Submit** |
-
-Cancel is always available via the **×** button in the dialog header.
-
----
-
-## Validation indicators
-
-Step number circles in the wizard navigation bar change color to reflect validity:
-
-- **Red circle** — the step contains at least one field with a validation error
-- **Green circle** — the step has been visited (navigated through) and all its fields are valid
-- **Default color** — the step has not been visited yet
-- **Dimmed** — a step that is not the currently active step
-
-To trigger validation immediately on open (before the user types anything), pass `validateOnInit`:
-
-```tsx
-<StepperCommandDialog
-    command={CreateProject}
-    validateOnInit
-    ...
->
-```
-
-This is useful when the dialog opens with pre-populated values that may already be invalid.
-
----
-
-## Customizing step labels
-
-The `okLabel`, `nextLabel`, and `previousLabel` props override the default button text:
-
-```tsx
-<StepperCommandDialog
-    command={CreateProject}
-    title="Register Employee"
-    okLabel="Register"
-    nextLabel="Continue →"
-    previousLabel="← Back"
-    ...
->
-```
-
----
-
-## Pre-populating values (edit wizard)
-
-Use `currentValues` for fields the user can change and `initialValues` for hidden/fixed fields (e.g. an ID):
-
-```tsx
-<StepperCommandDialog
-    command={UpdateProject}
-    currentValues={{ name: project.name, description: project.description }}
-    initialValues={{ projectId: project.id }}
-    ...
->
-```
-
----
-
-## Stepper orientation
-
-For longer wizards, vertical orientation can be more readable:
-
-```tsx
-<StepperCommandDialog orientation="vertical" ...>
-    <StepperPanel header="Step One"> ... </StepperPanel>
-    <StepperPanel header="Step Two"> ... </StepperPanel>
-</StepperCommandDialog>
-```
-
----
-
-## Props reference
-
-| Prop | Type | Default | Description |
-|------|------|---------|-------------|
-| `command` | `Constructor<TCommand>` | — | **Required.** Command class constructor |
-| `title` | `string` | — | **Required.** Dialog header title |
-| `children` | `StepperPanel[]` | — | **Required.** Wizard steps |
-| `okLabel` | `string` | `'Submit'` | Submit button label (last step) |
-| `nextLabel` | `string` | `'Next'` | Next button label |
-| `previousLabel` | `string` | `'Previous'` | Previous button label |
-| `visible` | `boolean` | `true` | Controls dialog visibility |
-| `width` | `string` | `'600px'` | Dialog width |
-| `isValid` | `boolean` | — | Extra validity gate combined with form validity |
-| `validateOnInit` | `boolean` | — | Run validation on mount to show errors immediately |
-| `orientation` | `'horizontal' \| 'vertical'` | `'horizontal'` | Stepper layout direction |
-| `linear` | `boolean` | `true` | Require steps to be completed in order |
-| `initialValues` | `Partial<TCommand>` | — | Fixed initial values (not shown to user as editable) |
-| `currentValues` | `Partial<TCommand>` | — | Pre-populated editable values |
-| `onConfirm` | `() => void \| Promise<void>` | — | Called after successful execute |
-| `onCancel` | `() => void \| Promise<void>` | — | Called when user dismisses dialog |
-| `onBeforeExecute` | `(values: TCommand) => TCommand` | — | Transform values before execution |
-| `pt` | `StepperProps['pt']` | — | PrimeReact PassThrough for deep DOM customization |
-
----
-
-## Common mistakes
-
-| Mistake | Fix |
-|---------|-----|
-| Putting a Cancel button in the footer | Don't — the × in the dialog header is the cancel action |
-| One step per field | Group related fields; aim for 2–5 fields per step |
-| Fields from different steps sharing the same `value` accessor | Each property should appear on exactly one step |
-| Forgetting `header` on `StepperPanel` | Always set `header` — it is the navigation label |
-| Using `CommandDialog` for a 4+ field form | Consider `StepperCommandDialog` to reduce cognitive load |
diff --git a/.github/skills/toolbar/SKILL.md b/.github/skills/toolbar/SKILL.md
deleted file mode 100644
index 7473699..0000000
--- a/.github/skills/toolbar/SKILL.md
+++ /dev/null
@@ -1,282 +0,0 @@
----
-name: toolbar
-description: Use this skill when asked to add or build a canvas-style icon toolbar using the @cratis/components Toolbar component. Covers Toolbar, ToolbarButton, ToolbarSeparator, ToolbarSection, ToolbarContext, and ToolbarFanOutItem. Use whenever building tool panels, drawing tool selectors, zoom controls, or any icon-button group with active states, context switching, or fan-out sub-panels.
----
-
-## When to use the Toolbar
-
-The `Toolbar` component is designed for **canvas-style tool panels** — the kind you find in drawing or diagram editors. It groups `ToolbarButton` elements into a pill-shaped bar with hover tooltips and active highlights. Use it when you need:
-
-- A tool selector with an "active tool" state
-- A set of action buttons that animate between tool modes
-- A toolbar that fans out sub-panels for grouped tools
-
-For a standard page action menu (Create / Edit / Delete over a data table), use `DataPage.MenuItems` instead.
-
----
-
-## Component overview
-
-| Component | Purpose |
-|---|---|
-| `Toolbar` | Container — renders a pill-shaped bar of buttons |
-| `ToolbarButton` | Icon or text button with tooltip and optional active state |
-| `ToolbarSeparator` | Visual divider between button groups |
-| `ToolbarSection` | Animated section that transitions between named contexts |
-| `ToolbarContext` | Named set of buttons inside a `ToolbarSection` |
-| `ToolbarFanOutItem` | Button that slides out a horizontal sub-panel on click |
-
-All components import from `@cratis/components`:
-
-```tsx
-import { Toolbar, ToolbarButton, ToolbarSeparator, ToolbarSection, ToolbarContext, ToolbarFanOutItem } from '@cratis/components';
-```
-
----
-
-## Step 1 — Basic vertical toolbar
-
-Place `ToolbarButton` elements inside `Toolbar`. The default orientation is vertical.
-
-```tsx
-import { Toolbar, ToolbarButton } from '@cratis/components';
-
-export const DrawingToolbar = () => (
-    <Toolbar>
-        <ToolbarButton icon='pi pi-arrow-up-left' tooltip='Select' />
-        <ToolbarButton icon='pi pi-pencil' tooltip='Draw' />
-        <ToolbarButton icon='pi pi-stop' tooltip='Rectangle' />
-    </Toolbar>
-);
-```
-
----
-
-## Step 2 — Active state (selected tool)
-
-Use the `active` prop to highlight the currently selected tool. Drive it from state:
-
-```tsx
-import { useState } from 'react';
-import { Toolbar, ToolbarButton } from '@cratis/components';
-
-export const DrawingToolbar = () => {
-    const [activeTool, setActiveTool] = useState('select');
-
-    return (
-        <Toolbar>
-            <ToolbarButton
-                icon='pi pi-arrow-up-left'
-                tooltip='Select'
-                active={activeTool === 'select'}
-                onClick={() => setActiveTool('select')}
-            />
-            <ToolbarButton
-                icon='pi pi-pencil'
-                tooltip='Draw'
-                active={activeTool === 'draw'}
-                onClick={() => setActiveTool('draw')}
-            />
-            <ToolbarButton
-                icon='pi pi-stop'
-                tooltip='Rectangle'
-                active={activeTool === 'rect'}
-                onClick={() => setActiveTool('rect')}
-            />
-        </Toolbar>
-    );
-};
-```
-
----
-
-## Step 3 — Separators
-
-`ToolbarSeparator` draws a thin divider line between groups. Pass the same `orientation` as the enclosing `Toolbar`:
-
-```tsx
-<Toolbar>
-    <ToolbarButton icon='pi pi-pencil' tooltip='Draw' />
-    <ToolbarButton icon='pi pi-stop' tooltip='Rectangle' />
-    <ToolbarSeparator />
-    <ToolbarButton icon='pi pi-undo' tooltip='Undo' />
-    <ToolbarButton icon='pi pi-refresh' tooltip='Redo' />
-</Toolbar>
-```
-
----
-
-## Step 4 — Horizontal toolbar with text buttons
-
-Pass `orientation='horizontal'` for a horizontal layout. Use `text` for buttons that display a value (e.g. zoom percentage):
-
-```tsx
-import { useState } from 'react';
-import { Toolbar, ToolbarButton, ToolbarSeparator } from '@cratis/components';
-
-export const ZoomToolbar = () => {
-    const [zoom, setZoom] = useState(100);
-
-    return (
-        <Toolbar orientation='horizontal'>
-            <ToolbarButton icon='pi pi-minus' tooltip='Zoom out' tooltipPosition='bottom' onClick={() => setZoom(z => z - 10)} />
-            <ToolbarButton text={`${zoom}%`} tooltip='Reset zoom' tooltipPosition='bottom' onClick={() => setZoom(100)} />
-            <ToolbarButton icon='pi pi-plus' tooltip='Zoom in' tooltipPosition='bottom' onClick={() => setZoom(z => z + 10)} />
-            <ToolbarSeparator orientation='horizontal' />
-            <ToolbarButton icon='pi pi-question-circle' tooltip='Help' tooltipPosition='bottom' />
-        </Toolbar>
-    );
-};
-```
-
-> For horizontal toolbars, set `tooltipPosition='bottom'` (or `'top'`) so tooltips do not overlap the toolbar.
-
----
-
-## Step 5 — Animated context switching
-
-`ToolbarSection` + `ToolbarContext` allows the toolbar to show different sets of buttons depending on the active mode. When `activeContext` changes, the buttons fade out, the section morphs to the new size, and the new buttons fade in.
-
-```tsx
-import { useState } from 'react';
-import { Toolbar, ToolbarButton, ToolbarSection, ToolbarContext } from '@cratis/components';
-
-export const ContextualToolbar = () => {
-    const [mode, setMode] = useState<'drawing' | 'text'>('drawing');
-
-    return (
-        <Toolbar>
-            <ToolbarButton icon='pi pi-arrow-up-left' tooltip='Select' />
-            <ToolbarSection activeContext={mode}>
-                <ToolbarContext name='drawing'>
-                    <ToolbarButton icon='pi pi-pencil' tooltip='Draw' />
-                    <ToolbarButton icon='pi pi-stop' tooltip='Rectangle' />
-                    <ToolbarButton icon='pi pi-circle' tooltip='Circle' />
-                </ToolbarContext>
-                <ToolbarContext name='text'>
-                    <ToolbarButton icon='pi pi-align-left' tooltip='Align Left' />
-                    <ToolbarButton icon='pi pi-align-center' tooltip='Align Center' />
-                    <ToolbarButton icon='pi pi-align-right' tooltip='Align Right' />
-                </ToolbarContext>
-            </ToolbarSection>
-            <ToolbarButton icon='pi pi-undo' tooltip='Undo' />
-        </Toolbar>
-    );
-};
-```
-
-Only the `ToolbarSection` transitions. Buttons outside the section are unaffected.
-
----
-
-## Step 6 — Fan-out sub-panel
-
-`ToolbarFanOutItem` replaces a regular button with one that fans out a horizontal panel of additional tools. Clicking the button again or anywhere outside closes the panel.
-
-```tsx
-import { Toolbar, ToolbarButton, ToolbarFanOutItem } from '@cratis/components';
-
-export const ShapesToolbar = () => (
-    <Toolbar>
-        <ToolbarButton icon='pi pi-arrow-up-left' tooltip='Select' />
-        <ToolbarFanOutItem icon='pi pi-th-large' tooltip='Shapes'>
-            <ToolbarButton icon='pi pi-stop' tooltip='Rectangle' />
-            <ToolbarButton icon='pi pi-circle' tooltip='Circle' />
-            <ToolbarButton icon='pi pi-minus' tooltip='Line' />
-        </ToolbarFanOutItem>
-    </Toolbar>
-);
-```
-
-When the toolbar is on the **right side** of the screen, fan out to the left:
-
-```tsx
-<ToolbarFanOutItem icon='pi pi-th-large' tooltip='Shapes' fanOutDirection='left'>
-    ...
-</ToolbarFanOutItem>
-```
-
----
-
-## Props reference
-
-### `Toolbar`
-
-| Prop | Type | Default | Description |
-|---|---|---|---|
-| `children` | `ReactNode` | — | `ToolbarButton`, `ToolbarSeparator`, `ToolbarSection`, or `ToolbarFanOutItem` elements |
-| `orientation` | `'vertical' \| 'horizontal'` | `'vertical'` | Layout direction |
-
-### `ToolbarButton`
-
-| Prop | Type | Default | Description |
-|---|---|---|---|
-| `icon` | `string` | — | PrimeIcons CSS class (e.g. `'pi pi-pencil'`) |
-| `text` | `string` | — | Text shown inside the button (use for values like zoom %) |
-| `tooltip` | `string` | **required** | Tooltip text on hover |
-| `active` | `boolean` | `false` | Highlights the button as selected |
-| `onClick` | `() => void` | — | Click handler |
-| `tooltipPosition` | `'top' \| 'right' \| 'bottom' \| 'left'` | `'right'` | Tooltip position |
-
-### `ToolbarSeparator`
-
-| Prop | Type | Default | Description |
-|---|---|---|---|
-| `orientation` | `'vertical' \| 'horizontal'` | `'vertical'` | Match the enclosing `Toolbar` orientation |
-
-### `ToolbarSection`
-
-| Prop | Type | Default | Description |
-|---|---|---|---|
-| `activeContext` | `string` | first context name | Name of the active `ToolbarContext` |
-| `children` | `ToolbarContext[]` | — | `ToolbarContext` children |
-| `orientation` | `'vertical' \| 'horizontal'` | `'vertical'` | Match the enclosing `Toolbar` |
-
-### `ToolbarContext`
-
-| Prop | Type | Default | Description |
-|---|---|---|---|
-| `name` | `string` | **required** | Identifier matched by `ToolbarSection.activeContext` |
-| `children` | `ReactNode` | — | `ToolbarButton` elements for this context |
-
-### `ToolbarFanOutItem`
-
-| Prop | Type | Default | Description |
-|---|---|---|---|
-| `icon` | `string` | **required** | PrimeIcons CSS class for the trigger button |
-| `tooltip` | `string` | **required** | Tooltip for the trigger button |
-| `tooltipPosition` | `'top' \| 'right' \| 'bottom' \| 'left'` | `'right'` | Tooltip position |
-| `fanOutDirection` | `'right' \| 'left'` | `'right'` | Direction the sub-panel slides out |
-| `children` | `ReactNode` | — | `ToolbarButton` elements inside the fan-out panel |
-
----
-
-## Multiple toolbar groups
-
-Render separate `Toolbar` instances to create distinct groups:
-
-```tsx
-<div className='flex flex-col gap-2'>
-    <Toolbar>
-        <ToolbarButton icon='pi pi-arrow-up-left' tooltip='Select' />
-        <ToolbarButton icon='pi pi-pencil' tooltip='Draw' />
-    </Toolbar>
-    <Toolbar>
-        <ToolbarButton icon='pi pi-undo' tooltip='Undo' />
-        <ToolbarButton icon='pi pi-refresh' tooltip='Redo' />
-    </Toolbar>
-</div>
-```
-
----
-
-## Validation
-
-After creating or modifying a Toolbar component, run:
-
-```bash
-yarn lint
-npx tsc -b
-```
-
-Fix all errors before proceeding.
diff --git a/.github/skills/write-documentation/SKILL.md b/.github/skills/write-documentation/SKILL.md
deleted file mode 100644
index 10c1693..0000000
--- a/.github/skills/write-documentation/SKILL.md
+++ /dev/null
@@ -1,122 +0,0 @@
----
-name: write-documentation
-description: "Diátaxis Documentation Expert for Cratis projects. Writes high-quality DocFX documentation guided by the Diátaxis framework — classifying every page as a Tutorial, How-to Guide, Reference, or Explanation."
----
-
-# Diátaxis Documentation Expert
-
-You are an expert technical writer producing DocFX-compatible documentation for Cratis projects. Every page you write is guided by the [Diátaxis framework](https://diataxis.fr/) — a systematic approach that classifies documentation into four distinct types, each serving a different user need.
-
-## Guiding Principles
-
-1. **Clarity** — Write in simple, clear, unambiguous language.
-2. **Accuracy** — Every code example must be complete, correct, and runnable. No pseudo-code.
-3. **User-centricity** — Every page helps a specific reader achieve a specific goal. Lead with *why*, then *how*.
-4. **Consistency** — Match the project's existing tone, terminology, and style. If it's "event source" in one place, it's "event source" everywhere.
-
-## The Four Document Types
-
-Before writing, determine which Diátaxis quadrant the page belongs to:
-
-| Type | Orientation | Analogy | When to use |
-|---|---|---|---|
-| **Tutorial** | Learning | A lesson | Guide a newcomer step-by-step to a successful first outcome |
-| **How-to Guide** | Problem-solving | A recipe | Show an experienced user how to accomplish a specific task |
-| **Reference** | Information | A dictionary | Describe the technical machinery — APIs, attributes, configuration |
-| **Explanation** | Understanding | A discussion | Clarify *why* something works the way it does, trade-offs, architecture |
-
-### Rules per type
-
-- **Tutorial** — Never explain *why*; focus on *do this, then this*. Each step must produce a visible, verifiable result. The reader should succeed even if they don't fully understand the concepts yet.
-- **How-to Guide** — Assume competence. State the goal, list prerequisites, give the steps, done. No teaching.
-- **Reference** — Be exhaustive and terse. Tables, type signatures, attribute lists. No narrative.
-- **Explanation** — No steps. Discuss concepts, trade-offs, and architecture decisions. Use Mermaid diagrams freely.
-
-## Workflow
-
-Follow this process for every documentation request:
-
-1. **Clarify** — Determine before writing:
-   - **Document type** — Tutorial, How-to, Reference, or Explanation
-   - **Target audience** — e.g. new developer, experienced contributor, framework consumer
-   - **Reader's goal** — What they want to achieve by reading this page
-   - **Scope** — What to include *and* what to exclude
-   If the request is ambiguous, ask before proceeding.
-
-2. **Propose structure** — Present an outline (headings + one-line descriptions). Wait for approval before writing full content.
-
-3. **Write** — Produce the full documentation in well-formatted Markdown. Adhere to all rules below.
-
-## File Structure
-
-All documentation lives in the `Documentation/` folder at the repository root. The site is built with [DocFX](https://dotnet.github.io/docfx/).
-
-For a new topic:
-
-    Documentation/<Section>/<Topic>/
-    ├── index.md      ← main content page
-    └── toc.yml       ← local table of contents
-
-Update the **parent** `toc.yml` to link to the new folder's `toc.yml`.
-
-### Structure rules
-
-- Every folder must have a `toc.yml` for navigation.
-- Every folder must have an `index.md` as its landing page.
-- In `toc.yml`, link to a subfolder's `toc.yml`, not its `index.md`.
-- Use relative links for all internal references.
-
-## toc.yml format
-
-Simple topic:
-
-    - name: <Topic Title>
-      href: index.md
-
-With sub-topics:
-
-    - name: <Topic Title>
-      href: index.md
-      items:
-        - name: Sub-topic
-          href: subtopic/toc.yml
-
-## Writing Style
-
-The project's voice is **direct, practical, and opinionated**. Write like an experienced colleague explaining something to a capable developer — confident but never condescending.
-
-- **Active voice, present tense.** "Chronicle appends the event" not "The event is appended by Chronicle."
-- **Second person.** "You configure…" not "One configures…" or "It is possible to configure…"
-- **Lead with the most important information.** Don't bury the key point after three paragraphs of context.
-- Use headings, lists, and code blocks to organize content — dense paragraphs lose readers.
-- Focus on public APIs and features — never internal implementation.
-- Do not document third-party libraries.
-- **American English only.** Always use US spellings: `color` not `colour`, `behavior` not `behaviour`, `customize` not `customise`, `organize` not `organise`, `recognize` not `recognise`, `analyze` not `analyse`, `initialize` not `initialise`.
-
-## Code Examples
-
-- Prefer `record` types for data structures (events, commands, read models).
-- `[EventType]` takes no arguments — never add a GUID or string.
-- Never include verbatim code from the repository — APIs change. Write purpose-built examples.
-- Every example must be complete and correct — no `// ...` elisions.
-
-## Diagrams
-
-Use [Mermaid](https://mermaid-js.github.io/mermaid/#/) for:
-- Architecture diagrams (`graph TD` or `graph LR`)
-- Sequence flows (`sequenceDiagram`)
-- State transitions (`stateDiagram-v2`)
-
-## Contextual Awareness
-
-- When existing documentation files are available, read them first to match tone, style, and terminology.
-- Do not copy content from them unless explicitly asked.
-- Do not fabricate external URLs — only link to resources you can verify exist.
-
-## Validation
-
-After writing, verify:
-- `toc.yml` is valid YAML
-- All `href` values point to files that exist (or will exist when created)
-- All Mermaid blocks are syntactically valid (balanced brackets, correct syntax)
-- File ends with a single trailing newline
diff --git a/.github/skills/write-specs-events/SKILL.md b/.github/skills/write-specs-events/SKILL.md
deleted file mode 100644
index da3c471..0000000
--- a/.github/skills/write-specs-events/SKILL.md
+++ /dev/null
@@ -1,207 +0,0 @@
----
-name: write-specs-events
-description: Use this skill when asked to write tests or specs for event appending, event log behavior, constraint violations, or concurrency violations using EventScenario in a Cratis-based project. Produces infrastructure-free in-process specs using EventScenario and AppendResult Should* extensions.
----
----
-
-# Writing Event Sequence Specs with EventScenario
-
-Use `EventScenario` to test event-appending behavior in-process — no Chronicle server, database, or network required.
-
-## When to use this skill
-
-- Testing that a command appends the correct event(s)
-- Testing that a constraint correctly rejects a duplicate or conflicting append
-- Testing `AppendMany` for batch event appends
-- Testing concurrency or error conditions in event appending
-
-For testing **read model projections or reducers**, use `ReadModelScenario<TReadModel>` instead.
-
----
-
-## Package
-
-```bash
-dotnet add package Cratis.Chronicle.Testing
-```
-
----
-
-## Basic structure
-
-```csharp
-// In your spec file (follows BDD Establish/Because/should_ pattern)
-public class when_appending_<event_name> : Specification
-{
-    EventScenario _scenario;
-    IAppendResult _result;
-
-    void Establish() => _scenario = new EventScenario();
-
-    async Task Because() =>
-        _result = await _scenario.EventLog.Append(<eventSourceId>, new <EventType>(<args>));
-
-    [Fact] void should_be_successful() => _result.ShouldBeSuccessful();
-}
-```
-
----
-
-## Step 1 — Create the scenario
-
-Always create a **new** `EventScenario` per test. It starts with an empty event log and sequence number zero.
-
-```csharp
-var scenario = new EventScenario();
-```
-
----
-
-## Step 2 — Seed pre-existing state (optional)
-
-Use the fluent `Given` builder when the act phase depends on prior state:
-
-```csharp
-await scenario.Given
-    .ForEventSource(authorId)
-    .Events(new AuthorRegistered("Jane Smith"), new BookAdded("Clean Code"));
-```
-
-Chain multiple `ForEventSource` calls to seed different event sources:
-
-```csharp
-await scenario.Given
-    .ForEventSource(author1Id)
-    .Events(new AuthorRegistered("Jane Smith"));
-
-await scenario.Given
-    .ForEventSource(author2Id)
-    .Events(new AuthorRegistered("John Doe"));
-```
-
-**Rules:**
-- Call `Given` **before** the act phase — seeded events get monotonically increasing sequence numbers.
-- Do not put the act-under-test inside `Given`; only pre-existing state goes here.
-
----
-
-## Step 3 — Append and assert
-
-```csharp
-var result = await scenario.EventLog.Append(eventSourceId, new SomeEvent("value"));
-result.ShouldBeSuccessful();
-```
-
-Use `AppendMany` for batch appends:
-
-```csharp
-var result = await scenario.EventLog.AppendMany(cartId, [
-    new ItemAddedToCart(itemId1),
-    new ItemAddedToCart(itemId2)
-]);
-result.ShouldBeSuccessful();
-```
-
----
-
-## Assertion reference
-
-| Method | Asserts |
-|--------|---------|
-| `result.ShouldBeSuccessful()` | Operation succeeded with no violations or errors |
-| `result.ShouldBeFailed()` | Operation failed (any violation or error) |
-| `result.ShouldHaveConstraintViolations()` | At least one constraint violation is present |
-| `result.ShouldNotHaveConstraintViolations()` | No constraint violations are present |
-| `result.ShouldHaveConstraintViolationFor("ConstraintName")` | A violation for the named constraint is present |
-| `result.ShouldHaveConcurrencyViolations()` | At least one concurrency violation is present |
-| `result.ShouldNotHaveConcurrencyViolations()` | No concurrency violations are present |
-| `result.ShouldHaveErrors()` | At least one error is present |
-| `result.ShouldNotHaveErrors()` | No errors are present |
-
-All `Should*` methods throw `AppendResultAssertionException` on failure with a message that lists all violations and errors found.
-
----
-
-## Example: happy path
-
-```csharp
-public class when_registering_an_author : Specification
-{
-    EventScenario _scenario;
-    IAppendResult _result;
-
-    void Establish() => _scenario = new EventScenario();
-
-    async Task Because() =>
-        _result = await _scenario.EventLog.Append(AuthorId.New(), new AuthorRegistered("Jane Smith"));
-
-    [Fact] void should_be_successful() => _result.ShouldBeSuccessful();
-    [Fact] void should_not_have_constraint_violations() => _result.ShouldNotHaveConstraintViolations();
-}
-```
-
----
-
-## Example: constraint violation (duplicate)
-
-Seed the pre-existing state, then verify the second append is rejected:
-
-```csharp
-public class when_registering_an_author
-{
-    public class and_the_name_already_exists : Specification
-    {
-        EventScenario _scenario;
-        IAppendResult _result;
-
-        async Task Establish()
-        {
-            _scenario = new EventScenario();
-            await _scenario.Given
-                .ForEventSource(AuthorId.New())
-                .Events(new AuthorRegistered("Jane Smith"));
-        }
-
-        async Task Because() =>
-            _result = await _scenario.EventLog.Append(AuthorId.New(), new AuthorRegistered("Jane Smith"));
-
-        [Fact] void should_be_failed() => _result.ShouldBeFailed();
-        [Fact] void should_have_a_constraint_violation() => _result.ShouldHaveConstraintViolations();
-        [Fact] void should_have_constraint_violation_for_unique_name() =>
-            _result.ShouldHaveConstraintViolationFor("UniqueAuthorName");
-    }
-}
-```
-
----
-
-## Example: batch append
-
-```csharp
-public class when_adding_multiple_items_to_cart : Specification
-{
-    EventScenario _scenario;
-    IAppendResult _result;
-
-    void Establish() => _scenario = new EventScenario();
-
-    async Task Because() =>
-        _result = await _scenario.EventLog.AppendMany(cartId, [
-            new ItemAddedToCart(itemId1),
-            new ItemAddedToCart(itemId2),
-            new ItemQuantityAdjusted(itemId1, 3)
-        ]);
-
-    [Fact] void should_be_successful() => _result.ShouldBeSuccessful();
-}
-```
-
----
-
-## Checklist
-
-- [ ] New `EventScenario` instance per test method (never shared state)
-- [ ] `Given` called before the act phase for all pre-existing state
-- [ ] Each test asserts a single outcome (one `should_` per behavior)
-- [ ] Constraint violation tests include `ShouldHaveConstraintViolationFor("ConstraintName")`
-- [ ] Run `dotnet build` and `dotnet test` — fix all failures before completing
diff --git a/.github/skills/write-specs-readmodels/SKILL.md b/.github/skills/write-specs-readmodels/SKILL.md
deleted file mode 100644
index 88471e8..0000000
--- a/.github/skills/write-specs-readmodels/SKILL.md
+++ /dev/null
@@ -1,216 +0,0 @@
----
-name: write-specs-readmodels
-description: Use this skill when asked to write tests or specs for read model projections, reducers, or model-bound projections using ReadModelScenario in a Cratis-based project. Produces infrastructure-free in-process specs using ReadModelScenario<TReadModel>.
----
-
-# Writing Read Model Specs with ReadModelScenario
-
-Use `ReadModelScenario<TReadModel>` to test read model projections and reducers entirely in-process — no Chronicle server, database, or network required.
-
-## When to use this skill
-
-- Testing that a reducer correctly builds state from a sequence of events
-- Testing that a fluent `IProjectionFor<T>` projection maps event properties to read model properties
-- Testing that a model-bound projection (`[FromEvent<T>]`, `[Key]`) maps correctly
-- Testing boundary conditions: first event, multiple events, different event sources
-
-For testing **event appending, constraints, or concurrency**, use `EventScenario` instead.
-
----
-
-## Package
-
-```bash
-dotnet add package Cratis.Chronicle.Testing
-```
-
----
-
-## Basic structure
-
-```csharp
-// In your spec file (follows BDD Establish/Because/should_ pattern)
-public class when_<event_sequence_description> : Specification
-{
-    ReadModelScenario<TReadModel> _scenario;
-
-    async Task Establish()
-    {
-        _scenario = new ReadModelScenario<TReadModel>();
-        await _scenario.Given
-            .ForEventSource(<eventSourceId>)
-            .Events(new <EventType>(<args>));
-    }
-
-    [Fact] void should_<expected_state>() => _scenario.Instance!.<Property>.ShouldEqual(<expectedValue>);
-}
-```
-
----
-
-## Step 1 — Create the scenario
-
-Always create a **new** `ReadModelScenario<TReadModel>` per test. Optionally pass an initial state:
-
-```csharp
-// Default (empty) initial state
-var scenario = new ReadModelScenario<MyReadModel>();
-
-// With an initial state baseline
-var scenario = new ReadModelScenario<MyReadModel>(new MyReadModel { Count = 10 });
-```
-
----
-
-## Step 2 — Seed events via the fluent Given builder
-
-```csharp
-await scenario.Given
-    .ForEventSource(myId)
-    .Events(new SomeEvent("value"), new SomeOtherEvent(42));
-```
-
-Chain multiple `ForEventSource` calls to seed different event sources:
-
-```csharp
-await scenario.Given
-    .ForEventSource(orderId)
-    .Events(new OrderCreated("order-1"), new ItemAdded(9.99m));
-
-await scenario.Given
-    .ForEventSource(anotherOrderId)
-    .Events(new OrderCreated("order-2"));
-```
-
-**Rules:**
-- Call `Given` before asserting — events are processed in the order they are supplied.
-- Do not put the act-under-test inside `Given`; the `Given` call **is** the act for a read model spec.
-
----
-
-## Step 3 — Assert on Instance
-
-```csharp
-_scenario.Instance!.Total.ShouldEqual(14.49m);
-_scenario.Instance!.Name.ShouldEqual("Widget");
-_scenario.Instance.ShouldNotBeNull();
-```
-
----
-
-## Auto-detection order
-
-`ReadModelScenario<TReadModel>` searches loaded assemblies for a handler in this order:
-
-1. **Reducer** — a class implementing `IReducerFor<TReadModel>`
-2. **Fluent projection** — a class implementing `IProjectionFor<TReadModel>`
-3. **Model-bound projection** — `TReadModel` carries `[FromEvent<T>]` or `[Key]` attributes
-
-If none is found, `NoReadModelHandlerFound` is thrown.
-
----
-
-## Example: reducer
-
-```csharp
-public class when_items_are_added_to_an_order : Specification
-{
-    ReadModelScenario<OrderSummary> _scenario;
-    static readonly string OrderId = "order-1";
-
-    async Task Establish()
-    {
-        _scenario = new ReadModelScenario<OrderSummary>();
-        await _scenario.Given
-            .ForEventSource(OrderId)
-            .Events(
-                new OrderCreated(OrderId),
-                new ItemAdded(9.99m),
-                new ItemAdded(4.50m));
-    }
-
-    [Fact] void should_sum_the_item_prices() => _scenario.Instance!.Total.ShouldEqual(14.49m);
-    [Fact] void should_have_the_correct_order_id() => _scenario.Instance!.OrderId.ShouldEqual(OrderId);
-}
-```
-
----
-
-## Example: fluent projection
-
-```csharp
-public class when_a_product_is_created_and_stock_is_adjusted : Specification
-{
-    ReadModelScenario<ProductView> _scenario;
-
-    async Task Establish()
-    {
-        _scenario = new ReadModelScenario<ProductView>();
-        await _scenario.Given
-            .ForEventSource("product-1")
-            .Events(
-                new ProductCreated("Widget"),
-                new StockAdjusted(100));
-    }
-
-    [Fact] void should_set_the_name() => _scenario.Instance!.Name.ShouldEqual("Widget");
-    [Fact] void should_set_the_stock() => _scenario.Instance!.Stock.ShouldEqual(100);
-}
-```
-
----
-
-## Example: model-bound projection
-
-```csharp
-public class when_a_shipment_is_dispatched_and_delivered : Specification
-{
-    ReadModelScenario<DeliveryStatus> _scenario;
-    static readonly DateTimeOffset DeliveredAt = DateTimeOffset.UtcNow;
-
-    async Task Establish()
-    {
-        _scenario = new ReadModelScenario<DeliveryStatus>();
-        await _scenario.Given
-            .ForEventSource("shipment-1")
-            .Events(
-                new ShipmentDispatched("FedEx"),
-                new ShipmentDelivered(DeliveredAt));
-    }
-
-    [Fact] void should_set_the_carrier() => _scenario.Instance!.Carrier.ShouldEqual("FedEx");
-    [Fact] void should_record_the_delivery_time() => _scenario.Instance!.DeliveredAt.ShouldEqual(DeliveredAt);
-}
-```
-
----
-
-## Example: initial state (baseline)
-
-```csharp
-public class when_an_item_is_added_to_a_cart_with_existing_items : Specification
-{
-    ReadModelScenario<CartSummary> _scenario;
-
-    async Task Establish()
-    {
-        var initial = new CartSummary { ItemCount = 3 };
-        _scenario = new ReadModelScenario<CartSummary>(initial);
-        await _scenario.Given
-            .ForEventSource("cart-1")
-            .Events(new ItemAddedToCart("item-4"));
-    }
-
-    [Fact] void should_increment_the_item_count() => _scenario.Instance!.ItemCount.ShouldEqual(4);
-}
-```
-
----
-
-## Checklist
-
-- [ ] New `ReadModelScenario<T>` instance per test (never shared state)
-- [ ] `Given.ForEventSource(id).Events(...)` called before asserting
-- [ ] Each test asserts a single outcome (one `should_` per behavior)
-- [ ] `Instance` is accessed with `!` only after `Given` has been called
-- [ ] Run `dotnet build` and `dotnet test` — fix all failures before completing
diff --git a/.github/skills/write-specs/SKILL.md b/.github/skills/write-specs/SKILL.md
deleted file mode 100644
index b64dbfd..0000000
--- a/.github/skills/write-specs/SKILL.md
+++ /dev/null
@@ -1,85 +0,0 @@
----
-name: write-specs
-description: Use this skill when asked to write tests, specs, or test coverage for a command, query, or slice in a Cratis-based project. Produces BDD-style integration specs using xUnit and Cratis.Specifications.
----
-
-Write comprehensive BDD integration specs for a vertical slice command or query.
-
-## Spec placement
-
-Specs live **in the same slice folder** as the `.cs` file:
-
-```
-Features/<Feature>/<Slice>/
-├── <Slice>.cs
-└── when_<verb_phrase>/
-    ├── and_<happy_scenario>.cs
-    └── and_<failure_scenario>.cs
-```
-
-## What to cover for every State Change command
-
-Write one spec class for **each** of:
-
-1. **Happy path** — command succeeds, expected event is appended, sequence number advanced
-2. **Each validation failure** — one `and_` class per `[Required]`, `[MaxLength]`, etc. rule
-3. **Each business rule violation** — one `and_` class per DCB condition in `Handle()` that inspects a read model
-4. **Each constraint violation** — one `and_` class per `IConstraint`
-
-## C# spec structure
-
-```csharp
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-
-using Cratis.Arc.Commands;
-using context = <NamespaceRoot>.<Feature>.<Slice>.when_<behavior>.and_<scenario>.context;
-
-namespace <NamespaceRoot>.<Feature>.<Slice>.when_<behavior>;
-
-[Collection(ChronicleCollection.Name)]
-public class and_<scenario>(context context) : Given<context>(context)
-{
-    public class context(ChronicleOutOfProcessFixture fixture) : given.an_http_client(fixture)
-    {
-        public CommandResult<object>? Result;
-
-        async Task Because()
-        {
-            Result = await Client.ExecuteCommand<<Command>>(
-                "/api/<feature>/<action>",
-                new <Command>(...));
-        }
-    }
-
-    [Fact] void should_<expected_result>() => Context.Result.IsSuccess.ShouldBeTrue();
-}
-```
-
-For scenarios that require pre-existing state, add an `async Task Establish()` before `Because()`:
-
-```csharp
-async Task Establish() =>
-    await EventStore.EventLog.Append(SomeId.New(), new SomeEvent(...));
-```
-
-## Naming conventions
-
-- Folder: `when_<verb_phrase>` — e.g. `when_registering`, `when_removing_a_project`
-- File: `and_<condition>.cs` — e.g. `and_name_is_unique.cs`, `and_name_already_exists.cs`
-- Test method: `should_<expected_result>` — e.g. `should_append_the_registered_event`
-
-## Assertion helpers
-
-Use Cratis.Specifications helpers — never raw `Assert.*`:
-- `Context.Result.IsSuccess.ShouldBeTrue()`
-- `Context.Result.IsSuccess.ShouldBeFalse()`
-- `Context.ShouldHaveTailSequenceNumber(EventSequenceNumber.First)` — checks the event log tail
-
-## After writing
-
-Run `dotnet test`. Fix all failures before completing.
-
----
-
-For full worked examples of each spec type (happy path, validation, business rule, constraint), see [references/EXAMPLES.md](references/EXAMPLES.md).
diff --git a/.github/skills/write-specs/references/EXAMPLES.md b/.github/skills/write-specs/references/EXAMPLES.md
deleted file mode 100644
index 263122f..0000000
--- a/.github/skills/write-specs/references/EXAMPLES.md
+++ /dev/null
@@ -1,115 +0,0 @@
-# Spec Examples
-
-## Happy path spec
-
-```csharp
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-
-using context = MyApp.Projects.Registration.when_registering.and_name_is_unique.context;
-
-namespace MyApp.Projects.Registration.when_registering;
-
-[Collection(ChronicleCollection.Name)]
-public class and_name_is_unique(context context) : Given<context>(context)
-{
-    public class context(ChronicleOutOfProcessFixture fixture) : given.an_http_client(fixture)
-    {
-        public CommandResult<object>? Result;
-
-        async Task Because()
-        {
-            Result = await Client.ExecuteCommand<RegisterProject>(
-                "/api/projects/registration",
-                new RegisterProject(ProjectId.New(), "My Project"));
-        }
-    }
-
-    [Fact] void should_succeed() => Context.Result.IsSuccess.ShouldBeTrue();
-    [Fact] void should_have_appended_the_registered_event() =>
-        Context.ShouldHaveTailSequenceNumber(EventSequenceNumber.First);
-}
-```
-
----
-
-## Constraint violation spec (pre-existing state)
-
-```csharp
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-
-using context = MyApp.Projects.Registration.when_registering.and_name_already_exists.context;
-
-namespace MyApp.Projects.Registration.when_registering;
-
-[Collection(ChronicleCollection.Name)]
-public class and_name_already_exists(context context) : Given<context>(context)
-{
-    public class context(ChronicleOutOfProcessFixture fixture) : given.an_http_client(fixture)
-    {
-        public const string ProjectName = "My Project";
-        public CommandResult<object>? Result;
-
-        async Task Establish() =>
-            await EventStore.EventLog.Append(ProjectId.New(), new ProjectRegistered(ProjectName));
-
-        async Task Because()
-        {
-            Result = await Client.ExecuteCommand<RegisterProject>(
-                "/api/projects/registration",
-                new RegisterProject(ProjectId.New(), ProjectName));
-        }
-    }
-
-    [Fact] void should_not_succeed() => Context.Result.IsSuccess.ShouldBeFalse();
-    [Fact] void should_not_have_appended_any_additional_events() =>
-        Context.ShouldHaveTailSequenceNumber(EventSequenceNumber.First);
-}
-```
-
----
-
-## Validation failure spec (model validation)
-
-```csharp
-// Copyright (c) Cratis. All rights reserved.
-// Licensed under the MIT license. See LICENSE file in the project root for full license information.
-
-using context = MyApp.Projects.Registration.when_registering.and_name_is_empty.context;
-
-namespace MyApp.Projects.Registration.when_registering;
-
-[Collection(ChronicleCollection.Name)]
-public class and_name_is_empty(context context) : Given<context>(context)
-{
-    public class context(ChronicleOutOfProcessFixture fixture) : given.an_http_client(fixture)
-    {
-        public CommandResult<object>? Result;
-
-        async Task Because()
-        {
-            Result = await Client.ExecuteCommand<RegisterProject>(
-                "/api/projects/registration",
-                new RegisterProject(ProjectId.New(), string.Empty));
-        }
-    }
-
-    [Fact] void should_not_succeed() => Context.Result.IsSuccess.ShouldBeFalse();
-    [Fact] void should_report_validation_error_on_name() =>
-        Context.Result.ValidationResults.ShouldNotBeEmpty();
-}
-```
-
----
-
-## Assertion helpers reference
-
-| Helper | When to use |
-|--------|-------------|
-| `.ShouldBeTrue()` / `.ShouldBeFalse()` | Boolean assertions |
-| `.ShouldEqual(x)` | Value equality |
-| `Context.ShouldHaveTailSequenceNumber(n)` | Confirm the tail event sequence number (use `EventSequenceNumber.First` for exactly one event in the log) |
-| `Context.Result.ValidationResults.ShouldNotBeEmpty()` | Confirm model validation rejected the command |
-| `Context.Result.IsSuccess.ShouldBeFalse()` | Confirm the command did not succeed (constraint or validation) |
-| `EventSequenceNumber.First` | Sequence number 0 — the very first event |
diff --git a/Directory.Packages.props b/Directory.Packages.props
index 5879055..b3fc018 100644
--- a/Directory.Packages.props
+++ b/Directory.Packages.props
@@ -25,7 +25,6 @@
     <!-- Transitive dependency overrides to address known vulnerabilities -->
     <PackageVersion Include="Snappier" Version="1.3.1" />
     <!-- Integration Testing -->
-    <PackageVersion Include="mongodb.driver" Version="3.8.0" />
     <PackageVersion Include="Testcontainers" Version="4.11.0" />
     <!-- Testing -->
     <PackageVersion Include="Cratis.Specifications" Version="3.0.5" />
diff --git a/Source/Cli/Commands/Chronicle/ChronicleCommand.cs b/Source/Cli/Commands/Chronicle/ChronicleCommand.cs
index 779d351..0869dff 100644
--- a/Source/Cli/Commands/Chronicle/ChronicleCommand.cs
+++ b/Source/Cli/Commands/Chronicle/ChronicleCommand.cs
@@ -16,6 +16,12 @@ namespace Cratis.Cli.Commands.Chronicle;
 public abstract partial class ChronicleCommand<TSettings> : AsyncCommand<TSettings>
     where TSettings : ChronicleSettings
 {
+    /// <summary>
+    /// Gets a value indicating whether this command wraps execution in an <see cref="AnsiConsole.Status"/> spinner.
+    /// Override and return <see langword="false"/> for commands that manage their own interactive display (e.g. live dashboards).
+    /// </summary>
+    protected virtual bool UseStatusSpinner => true;
+
     [GeneratedRegex("://(?<user>[^:@/]+):[^@/]+@", RegexOptions.None, matchTimeoutMilliseconds: 1000)]
     static partial Regex ConnectionStringCredentialsRegex { get; }
 
@@ -46,7 +52,7 @@ protected sealed override async Task<int> ExecuteAsync(CommandContext context, T
                 int exitCode;
                 var sw = settings.Debug ? Stopwatch.StartNew() : null;
 
-                if (string.Equals(format, OutputFormats.Table, StringComparison.Ordinal))
+                if (string.Equals(format, OutputFormats.Table, StringComparison.Ordinal) && UseStatusSpinner)
                 {
                     exitCode = await AnsiConsole.Status()
                         .Spinner(Spinner.Known.Dots)
diff --git a/Source/Cli/Commands/Chronicle/Workbench/NavFrame.cs b/Source/Cli/Commands/Chronicle/Workbench/NavFrame.cs
new file mode 100644
index 0000000..b05f993
--- /dev/null
+++ b/Source/Cli/Commands/Chronicle/Workbench/NavFrame.cs
@@ -0,0 +1,12 @@
+// Copyright (c) Cratis. All rights reserved.
+// Licensed under the MIT license. See LICENSE file in the project root for full license information.
+
+namespace Cratis.Cli.Commands.Chronicle.Workbench;
+
+/// <summary>
+/// A frame on the workbench navigation stack, capturing view state before drilling into a detail view.
+/// </summary>
+/// <param name="View">The view enum value (stored as int to avoid volatile boxing).</param>
+/// <param name="SelectedIndex">The list cursor position in that view.</param>
+/// <param name="FocusedId">The focused item identifier in that view.</param>
+public readonly record struct NavFrame(int View, int SelectedIndex, string FocusedId);
diff --git a/Source/Cli/Commands/Chronicle/Workbench/WorkbenchAction.cs b/Source/Cli/Commands/Chronicle/Workbench/WorkbenchAction.cs
new file mode 100644
index 0000000..3411c7b
--- /dev/null
+++ b/Source/Cli/Commands/Chronicle/Workbench/WorkbenchAction.cs
@@ -0,0 +1,68 @@
+// Copyright (c) Cratis. All rights reserved.
+// Licensed under the MIT license. See LICENSE file in the project root for full license information.
+
+namespace Cratis.Cli.Commands.Chronicle.Workbench;
+
+/// <summary>
+/// The current action lifecycle state of the workbench.
+/// </summary>
+public enum WorkbenchActionState
+{
+    /// <summary>Normal operation — no action pending.</summary>
+    None,
+
+    /// <summary>An action is queued and awaiting user confirmation.</summary>
+    AwaitingConfirmation,
+
+    /// <summary>An action is being executed against the server.</summary>
+    Executing,
+
+    /// <summary>An action has completed; the result is available.</summary>
+    Completed
+}
+
+/// <summary>
+/// Describes a pending in-place action that requires user confirmation before execution.
+/// </summary>
+/// <param name="Description">Human-readable label shown in the confirmation prompt.</param>
+/// <param name="SuccessMessage">Message displayed after successful execution.</param>
+/// <param name="Execute">The gRPC call to invoke when the user confirms.</param>
+public record PendingAction(
+    string Description,
+    string SuccessMessage,
+    Func<CancellationToken, Task> Execute);
+
+/// <summary>
+/// Carries all render-relevant state for a single frame of the workbench, keeping the Build() signature stable.
+/// </summary>
+/// <param name="View">The currently active view (primary or detail).</param>
+/// <param name="SelectedIndex">The selected list-item index within the current view.</param>
+/// <param name="Interval">The current refresh interval in seconds.</param>
+/// <param name="IsRefreshing">Whether a data fetch is currently in progress.</param>
+/// <param name="ActionState">The current action lifecycle state.</param>
+/// <param name="PendingActionDescription">Description of the action awaiting confirmation, or <see langword="null"/>.</param>
+/// <param name="ActionResult">Result message after action completion, or <see langword="null"/>.</param>
+/// <param name="IsActionError">Whether the most recent action completed with an error.</param>
+/// <param name="FocusedId">The identifier of the item shown in a detail view.</param>
+/// <param name="ScrollOffset">The content scroll offset (line number) in detail views.</param>
+/// <param name="Breadcrumb">Navigation breadcrumb path, e.g. ["Observers", "SomeProjection"].</param>
+/// <param name="FilterText">The active inline filter string, or empty when no filter is applied.</param>
+/// <param name="FilterInputMode">Whether the user is actively typing a filter (entered via '/').</param>
+/// <param name="EventLogAscending">Whether the event log is sorted ascending (oldest first) instead of the default descending (newest first).</param>
+/// <param name="EventLogPage">The current zero-based page index within the event log (50 events per page).</param>
+public record WorkbenchRenderState(
+    WorkbenchView View,
+    int SelectedIndex,
+    int Interval,
+    bool IsRefreshing,
+    WorkbenchActionState ActionState,
+    string? PendingActionDescription,
+    string? ActionResult,
+    bool IsActionError = false,
+    string FocusedId = "",
+    int ScrollOffset = 0,
+    IReadOnlyList<string>? Breadcrumb = null,
+    string FilterText = "",
+    bool FilterInputMode = false,
+    bool EventLogAscending = false,
+    int EventLogPage = 0);
diff --git a/Source/Cli/Commands/Chronicle/Workbench/WorkbenchCommand.cs b/Source/Cli/Commands/Chronicle/Workbench/WorkbenchCommand.cs
new file mode 100644
index 0000000..c5e4e4d
--- /dev/null
+++ b/Source/Cli/Commands/Chronicle/Workbench/WorkbenchCommand.cs
@@ -0,0 +1,1170 @@
+// Copyright (c) Cratis. All rights reserved.
+// Licensed under the MIT license. See LICENSE file in the project root for full license information.
+
+using System.Text.Json;
+using Cratis.Chronicle.Contracts.Jobs;
+using Cratis.Cli.Commands.Chronicle.ReadModels;
+
+namespace Cratis.Cli.Commands.Chronicle.Workbench;
+
+/// <summary>
+/// Interactive TUI workbench — a live-updating dashboard with full drill-down navigation and in-place
+/// actions for observers, failed partitions, jobs, recommendations, event log, event types, and projections.
+/// </summary>
+[LlmDescription("Opens an interactive full-screen TUI workbench for the Chronicle server. Navigate with ← → or 1–8, drill into items with Enter, go back with Escape. Not suitable for scripting.")]
+[CliCommand("workbench", "Open the interactive Chronicle workbench (live TUI dashboard)", Branch = typeof(ChronicleBranch))]
+[CliExample("chronicle", "workbench")]
+[CliExample("chronicle", "workbench", "--interval", "10")]
+[CliExample("chronicle", "workbench", "-e", "my-event-store")]
+public class WorkbenchCommand : ChronicleCommand<WorkbenchSettings>
+{
+    /// <summary>Number of primary <see cref="WorkbenchView"/> values (values &lt; 100) — must be kept in sync with the enum.</summary>
+    const int ViewCount = 11;
+
+    /// <summary>Number of events displayed per page in the Event Log view.</summary>
+    const int EventLogPageSize = 50;
+
+    /// <summary>Total events fetched from the server for the Event Log — determines how many pages are available.</summary>
+    const int EventLogFetchWindow = 500;
+
+    static readonly JsonSerializerOptions _instanceJsonOptions = new() { WriteIndented = true };
+
+    /// <summary>Nav stack — only touched by the input task (HandleKey). Render loop uses snapshot fields.</summary>
+    readonly Stack<NavFrame> _navStack = new();
+
+    volatile int _currentView;
+    volatile int _selectedIndex;
+    volatile int _actionState;
+    volatile int _scrollOffset;
+    volatile int _isActionError;
+    volatile int _filterInputMode;
+    volatile int _eventLogAscending;
+    volatile int _eventLogPage;
+
+    /// <summary>
+    /// Fired by the input task whenever the user presses a key; wakes the render loop immediately
+    /// rather than waiting for the full refresh interval to elapse.
+    /// </summary>
+    TaskCompletionSource _keyPressSignal = new(TaskCreationOptions.RunContinuationsAsynchronously);
+
+    PendingAction? _pendingAction;
+    string _actionResult = string.Empty;
+    string _focusedId = string.Empty;
+    string _filter = string.Empty;
+    string? _activeEventStore;
+    string? _activeNamespace;
+    IReadOnlyList<string> _breadcrumb = [];
+    IServices? _services;
+    WorkbenchData? _lastData;
+
+    /// <inheritdoc/>
+    protected override bool UseStatusSpinner => false;
+
+    /// <inheritdoc/>
+    protected override async Task<int> ExecuteCommandAsync(IServices services, WorkbenchSettings settings, string format)
+    {
+        if (!string.Equals(format, OutputFormats.Table, StringComparison.Ordinal))
+        {
+            OutputFormatter.WriteError(
+                format,
+                "The workbench requires table output format",
+                "Remove -o/--output or use --output table",
+                ExitCodes.ValidationErrorCode);
+            return ExitCodes.ValidationError;
+        }
+
+        _services = services;
+
+        using var cts = new CancellationTokenSource();
+        Console.CancelKeyPress += (_, e) =>
+        {
+            e.Cancel = true;
+            cts.Cancel();
+        };
+
+        var inputTask = Task.Run(
+            async () =>
+            {
+                while (!cts.Token.IsCancellationRequested)
+                {
+                    if (Console.KeyAvailable)
+                    {
+                        var pressedKey = Console.ReadKey(intercept: true);
+                        HandleKey(pressedKey, settings, cts);
+                    }
+
+                    try
+                    {
+                        await Task.Delay(50, cts.Token).ConfigureAwait(false);
+                    }
+                    catch (OperationCanceledException)
+                    {
+                        break;
+                    }
+                }
+            },
+            cts.Token);
+
+        var data = WorkbenchData.Loading(settings);
+
+        await AnsiConsole.Live(WorkbenchRenderer.Build(data, LoadingState(settings)))
+            .StartAsync(async ctx =>
+            {
+                while (!cts.Token.IsCancellationRequested)
+                {
+                    var signal = new TaskCompletionSource(TaskCreationOptions.RunContinuationsAsynchronously);
+                    Interlocked.Exchange(ref _keyPressSignal, signal);
+
+                    ctx.UpdateTarget(WorkbenchRenderer.Build(data, RenderState(settings, isRefreshing: true)));
+                    ctx.Refresh();
+
+                    data = await FetchData(services, settings, _activeEventStore, _activeNamespace, (WorkbenchView)_currentView, _focusedId, cts.Token);
+                    _lastData = data;
+
+                    ctx.UpdateTarget(WorkbenchRenderer.Build(data, RenderState(settings, isRefreshing: false)));
+                    ctx.Refresh();
+
+                    try
+                    {
+                        await Task.WhenAny(
+                            Task.Delay(TimeSpan.FromSeconds(settings.Interval), cts.Token),
+                            signal.Task);
+                    }
+                    catch (OperationCanceledException)
+                    {
+                        break;
+                    }
+
+                    if (cts.Token.IsCancellationRequested) break;
+
+                    ctx.UpdateTarget(WorkbenchRenderer.Build(data, RenderState(settings, isRefreshing: true)));
+                    ctx.Refresh();
+                }
+            });
+
+        try
+        {
+            await inputTask;
+        }
+        catch (OperationCanceledException)
+        {
+        }
+
+        AnsiConsole.MarkupLine($"  [{OutputFormatter.Muted.ToMarkup()}]Workbench closed.[/]");
+        return ExitCodes.Success;
+    }
+
+    static async Task<WorkbenchData> FetchData(
+        IServices services,
+        WorkbenchSettings settings,
+        string? activeEventStore,
+        string? activeNamespace,
+        WorkbenchView currentView,
+        string focusedId,
+        CancellationToken ct)
+    {
+        var eventStore = activeEventStore ?? settings.ResolveEventStore();
+        var ns = activeNamespace ?? settings.ResolveNamespace();
+        var connectionString = settings.ResolveConnectionString();
+
+        string? serverVersion = null;
+        var isConnected = true;
+
+        try
+        {
+            var versionInfo = await services.Server.GetVersionInfo();
+            serverVersion = versionInfo.Version;
+        }
+        catch
+        {
+            isConnected = false;
+        }
+
+        var eventStoreNames = new List<string>();
+        try { eventStoreNames = [.. await services.EventStores.GetEventStores()]; }
+        catch { }
+
+        IReadOnlyList<ObserverInformation> observers = [];
+        try
+        {
+            observers = [.. await services.Observers.GetObservers(new AllObserversRequest
+            {
+                EventStore = eventStore,
+                Namespace = ns
+            })];
+        }
+        catch { }
+
+        IReadOnlyList<FailedPartition> failedPartitions = [];
+        try
+        {
+            failedPartitions = [.. await services.FailedPartitions.GetFailedPartitions(new GetFailedPartitionsRequest
+            {
+                EventStore = eventStore,
+                Namespace = ns
+            })];
+        }
+        catch { }
+
+        IReadOnlyList<Job> jobs = [];
+        try
+        {
+            jobs = [.. (await services.Jobs.GetJobs(new GetJobsRequest
+            {
+                EventStore = eventStore,
+                Namespace = ns
+            })) ?? []];
+        }
+        catch { }
+
+        IReadOnlyList<Recommendation> recommendations = [];
+        try
+        {
+            recommendations = [.. await services.Recommendations.GetRecommendations(new GetRecommendationsRequest
+            {
+                EventStore = eventStore,
+                Namespace = ns
+            })];
+        }
+        catch { }
+
+        ulong? tailSequenceNumber = null;
+        try
+        {
+            var tail = await services.EventSequences.GetTailSequenceNumber(new GetTailSequenceNumberRequest
+            {
+                EventStore = eventStore,
+                Namespace = ns,
+                EventSequenceId = CliDefaults.DefaultEventSequenceId
+            });
+            tailSequenceNumber = tail.SequenceNumber == ulong.MaxValue ? null : tail.SequenceNumber;
+        }
+        catch { }
+
+        IReadOnlyList<EventTypeRegistration> eventTypeRegistrations = [];
+        try
+        {
+            eventTypeRegistrations = [.. await services.EventTypes.GetAllRegistrations(
+                new GetAllEventTypesRequest { EventStore = eventStore })];
+        }
+        catch { }
+
+        IReadOnlyList<ProjectionDefinition> projectionDefinitions = [];
+        try
+        {
+            projectionDefinitions = [.. await services.Projections.GetAllDefinitions(
+                new GetAllDefinitionsRequest { EventStore = eventStore })];
+        }
+        catch { }
+
+        var projectionDeclarations = new Dictionary<string, string>();
+        try
+        {
+            var declarations = await services.Projections.GetAllDeclarations(
+                new GetAllDeclarationsRequest { EventStore = eventStore });
+            foreach (var d in declarations)
+            {
+                projectionDeclarations[d.Identifier] = d.Declaration ?? string.Empty;
+            }
+        }
+        catch { }
+
+        IReadOnlyList<AppendedEvent> recentEvents = [];
+        try
+        {
+            if (tailSequenceNumber > 0)
+            {
+                var fromSeq = tailSequenceNumber.Value >= EventLogFetchWindow
+                    ? tailSequenceNumber.Value - EventLogFetchWindow + 1
+                    : 0;
+                var eventsResp = await services.EventSequences.GetEventsFromEventSequenceNumber(
+                    new GetFromEventSequenceNumberRequest
+                    {
+                        EventStore = eventStore,
+                        Namespace = ns,
+                        EventSequenceId = CliDefaults.DefaultEventSequenceId,
+                        FromEventSequenceNumber = fromSeq
+                    });
+                recentEvents = [.. eventsResp.Events.OrderByDescending(e => e.Context.SequenceNumber)];
+            }
+        }
+        catch { }
+
+        IReadOnlyList<WorkbenchReadModel> readModelDefinitions = [];
+        try
+        {
+            var defs = await services.ReadModels.GetDefinitions(new GetDefinitionsRequest { EventStore = eventStore });
+            readModelDefinitions = [.. defs.ReadModels.Select(rm => new WorkbenchReadModel(
+                rm.ContainerName,
+                rm.DisplayName,
+                rm.Owner.ToString(),
+                !string.Equals(rm.Owner.ToString(), "Client", StringComparison.Ordinal),
+                rm.Source.ToString(),
+                rm.Type?.Identifier ?? string.Empty))];
+        }
+        catch { }
+
+        IReadOnlyList<string> namespaceNames = [];
+        try { namespaceNames = [.. await services.Namespaces.GetNamespaces(new GetNamespacesRequest { EventStore = eventStore })]; }
+        catch { }
+
+        IReadOnlyList<string> readModelInstances = [];
+        var readModelInstancesTotalCount = 0;
+        string? readModelInstancesError = null;
+        if (currentView == WorkbenchView.ReadModelDetail && !string.IsNullOrEmpty(focusedId))
+        {
+            try
+            {
+                var instResp = await services.ReadModels.GetInstances(new GetInstancesRequest
+                {
+                    EventStore = eventStore,
+                    Namespace = ns,
+                    ReadModel = focusedId,
+                    Page = 0,
+                    PageSize = 20
+                });
+                readModelInstancesTotalCount = (int)Math.Min(instResp.TotalCount, int.MaxValue);
+                readModelInstances = [.. (instResp.Instances ?? [])
+                    .Select(ReadModelJsonCleaner.CleanInstance)
+                    .Where(o => o is not null)
+                    .Select(o => JsonSerializer.Serialize(o, _instanceJsonOptions))];
+            }
+            catch (Exception ex)
+            {
+                readModelInstancesError = ex.Message;
+            }
+        }
+
+        return new WorkbenchData(
+            ConnectionString: connectionString,
+            EventStore: eventStore,
+            Namespace: ns,
+            IsConnected: isConnected,
+            ServerVersion: serverVersion,
+            EventStoreNames: eventStoreNames,
+            Observers: observers,
+            FailedPartitions: failedPartitions,
+            Jobs: jobs,
+            Recommendations: recommendations,
+            TailSequenceNumber: tailSequenceNumber,
+            CapturedAt: DateTimeOffset.Now,
+            FetchError: null,
+            EventTypeRegistrations: eventTypeRegistrations,
+            ProjectionDefinitions: projectionDefinitions,
+            ProjectionDeclarations: projectionDeclarations,
+            RecentEvents: recentEvents,
+            ReadModelDefinitions: readModelDefinitions,
+            NamespaceNames: namespaceNames,
+            ReadModelInstances: readModelInstances,
+            ReadModelInstancesTotalCount: readModelInstancesTotalCount,
+            ReadModelInstancesError: readModelInstancesError);
+    }
+
+    static int ObserverSortOrder(ObserverInformation o) => o.RunningState switch
+    {
+        ObserverRunningState.Disconnected => 0,
+        ObserverRunningState.Replaying => 1,
+        ObserverRunningState.Active => 2,
+        ObserverRunningState.Suspended => 3,
+        _ => 4
+    };
+
+    static string GetViewLabel(WorkbenchView view) => view switch
+    {
+        WorkbenchView.Overview => "Overview",
+        WorkbenchView.Observers => "Observers",
+        WorkbenchView.FailedPartitions => "Failures",
+        WorkbenchView.Jobs => "Jobs",
+        WorkbenchView.Recommendations => "Recommendations",
+        WorkbenchView.EventLog => "Event Log",
+        WorkbenchView.EventTypes => "Event Types",
+        WorkbenchView.Projections => "Projections",
+        WorkbenchView.ReadModels => "Read Models",
+        WorkbenchView.EventStores => "Event Stores",
+        WorkbenchView.Namespaces => "Namespaces",
+        WorkbenchView.ObserverDetail => "Observer",
+        WorkbenchView.FailedPartitionDetail => "Failure",
+        WorkbenchView.EventDetail => "Event",
+        WorkbenchView.EventTypeDetail => "Event Type",
+        WorkbenchView.ProjectionDetail => "Projection",
+        WorkbenchView.ReadModelDetail => "Read Model",
+        _ => string.Empty
+    };
+
+    static WorkbenchRenderState LoadingState(WorkbenchSettings settings) =>
+        new(WorkbenchView.Overview, 0, settings.Interval, true, WorkbenchActionState.None, null, null);
+
+    static string TruncateForPrompt(string s) => s.Length <= 60 ? s : s[..57] + "…";
+
+    static bool HasSubNavigation(WorkbenchView view) => view == WorkbenchView.ObserverDetail;
+
+    static bool IsFilterableView(WorkbenchView view) =>
+        view is WorkbenchView.Observers or WorkbenchView.EventTypes
+            or WorkbenchView.EventLog or WorkbenchView.Projections
+            or WorkbenchView.ReadModels;
+
+    int GetMaxSelectedIndex()
+    {
+        var view = (WorkbenchView)_currentView;
+        var data = _lastData;
+        if (data is null) return 0;
+        if (view == WorkbenchView.EventLog)
+        {
+            var pageStart = _eventLogPage * EventLogPageSize;
+            return Math.Max(0, Math.Min(EventLogPageSize, data.RecentEvents.Count - pageStart) - 1);
+        }
+        return Math.Max(0, view switch
+        {
+            WorkbenchView.Observers => data.Observers.Count - 1,
+            WorkbenchView.FailedPartitions => data.FailedPartitions.Count - 1,
+            WorkbenchView.Jobs => data.Jobs.Count - 1,
+            WorkbenchView.Recommendations => data.Recommendations.Count - 1,
+            WorkbenchView.EventTypes => data.EventTypeRegistrations.Count - 1,
+            WorkbenchView.Projections => data.ProjectionDefinitions.Count - 1,
+            WorkbenchView.ReadModels => data.ReadModelDefinitions.Count - 1,
+            WorkbenchView.EventStores => data.EventStoreNames.Count - 1,
+            WorkbenchView.Namespaces => data.NamespaceNames.Count - 1,
+            WorkbenchView.ObserverDetail => Math.Max(0, (data.Observers
+                .FirstOrDefault(o => o.Id == _focusedId)?.EventTypes?.Count() ?? 1) - 1),
+            _ => 0
+        });
+    }
+
+    WorkbenchRenderState RenderState(WorkbenchSettings settings, bool isRefreshing) =>
+        new(
+            View: (WorkbenchView)_currentView,
+            SelectedIndex: _selectedIndex,
+            Interval: settings.Interval,
+            IsRefreshing: isRefreshing,
+            ActionState: (WorkbenchActionState)_actionState,
+            PendingActionDescription: _pendingAction?.Description,
+            ActionResult: _actionResult,
+            IsActionError: _isActionError != 0,
+            FocusedId: _focusedId,
+            ScrollOffset: _scrollOffset,
+            Breadcrumb: _breadcrumb,
+            FilterText: _filter,
+            FilterInputMode: _filterInputMode != 0,
+            EventLogAscending: _eventLogAscending != 0,
+            EventLogPage: _eventLogPage);
+
+    ObserverInformation? GetSelectedObserver() =>
+        _lastData?.Observers
+            .OrderBy(ObserverSortOrder)
+            .ThenBy(o => o.Id)
+            .Skip(_selectedIndex)
+            .FirstOrDefault();
+
+    ObserverInformation? FindObserverById(string id) =>
+        _lastData?.Observers.FirstOrDefault(o => o.Id == id);
+
+    FailedPartition? GetSelectedFailedPartition() =>
+        _lastData?.FailedPartitions
+            .OrderByDescending(fp => fp.Attempts.Count())
+            .Skip(_selectedIndex)
+            .FirstOrDefault();
+
+    FailedPartition? FindFailedPartitionByFocusedId(string focusedId)
+    {
+        var sep = focusedId.IndexOf('/');
+        if (sep < 0) return null;
+        var obsId = focusedId[..sep];
+        var partition = focusedId[(sep + 1)..];
+        return _lastData?.FailedPartitions
+            .FirstOrDefault(fp => fp.ObserverId == obsId && fp.Partition == partition);
+    }
+
+    Job? GetSelectedJob() =>
+        _lastData?.Jobs
+            .OrderBy(j => j.Status.ToString())
+            .Skip(_selectedIndex)
+            .FirstOrDefault();
+
+    Recommendation? GetSelectedRecommendation() =>
+        _lastData?.Recommendations
+            .Skip(_selectedIndex)
+            .FirstOrDefault();
+
+    AppendedEvent? GetSelectedEvent() =>
+        _lastData?.RecentEvents
+            .Skip(_selectedIndex)
+            .FirstOrDefault();
+
+    EventTypeRegistration? GetSelectedEventTypeRegistration() =>
+        _lastData?.EventTypeRegistrations
+            .OrderBy(r => r.Type.Id, StringComparer.OrdinalIgnoreCase)
+            .ThenBy(r => r.Type.Generation)
+            .Skip(_selectedIndex)
+            .FirstOrDefault();
+
+    ProjectionDefinition? GetSelectedProjectionDefinition() =>
+        _lastData?.ProjectionDefinitions
+            .OrderBy(d => d.Identifier, StringComparer.OrdinalIgnoreCase)
+            .Skip(_selectedIndex)
+            .FirstOrDefault();
+
+    WorkbenchReadModel? GetSelectedReadModel() =>
+        _lastData?.ReadModelDefinitions
+            .OrderBy(d => d.ContainerName, StringComparer.OrdinalIgnoreCase)
+            .Skip(_selectedIndex)
+            .FirstOrDefault();
+
+    string? GetSelectedEventStoreName() =>
+        _lastData?.EventStoreNames
+            .Order(StringComparer.OrdinalIgnoreCase)
+            .Skip(_selectedIndex)
+            .FirstOrDefault();
+
+    string? GetSelectedNamespaceName() =>
+        _lastData?.NamespaceNames
+            .Order(StringComparer.OrdinalIgnoreCase)
+            .Skip(_selectedIndex)
+            .FirstOrDefault();
+
+    void PushNav()
+    {
+        _navStack.Push(new NavFrame(_currentView, _selectedIndex, _focusedId));
+    }
+
+    List<string> BuildBreadcrumb()
+    {
+        if (_navStack.Count == 0) return [];
+        var path = new List<string>();
+        foreach (var frame in _navStack.Reverse())
+        {
+            path.Add(GetViewLabel((WorkbenchView)frame.View));
+        }
+
+        if (!string.IsNullOrEmpty(_focusedId)) path.Add(TruncateForPrompt(_focusedId));
+        return path;
+    }
+
+    void NavigateToDetail(WorkbenchView detailView, string focusedId)
+    {
+        PushNav();
+        _focusedId = focusedId;
+        _currentView = (int)detailView;
+        _selectedIndex = 0;
+        _scrollOffset = 0;
+        _breadcrumb = BuildBreadcrumb();
+    }
+
+    void NavigateBack()
+    {
+        var frame = _navStack.Pop();
+        _currentView = frame.View;
+        _selectedIndex = frame.SelectedIndex;
+        _focusedId = frame.FocusedId;
+        _scrollOffset = 0;
+        _breadcrumb = BuildBreadcrumb();
+    }
+
+    void SetPendingAction(PendingAction action)
+    {
+        _pendingAction = action;
+        _actionState = (int)WorkbenchActionState.AwaitingConfirmation;
+        _keyPressSignal.TrySetResult();
+    }
+
+    async Task ExecuteActionAsync(CancellationToken ct)
+    {
+        var action = _pendingAction;
+        if (action is null) return;
+
+        _isActionError = 0;
+        _actionState = (int)WorkbenchActionState.Executing;
+        _keyPressSignal.TrySetResult();
+
+        try
+        {
+            await action.Execute(ct);
+            _actionResult = action.SuccessMessage;
+        }
+        catch (OperationCanceledException)
+        {
+            _actionState = (int)WorkbenchActionState.None;
+            return;
+        }
+        catch (Exception ex)
+        {
+            var msg = ex.Message;
+            _actionResult = msg.Length > 100 ? msg[..100] + "…" : msg;
+            _isActionError = 1;
+        }
+
+        _pendingAction = null;
+        _actionState = (int)WorkbenchActionState.Completed;
+        _keyPressSignal.TrySetResult();
+    }
+
+    void HandleKey(ConsoleKeyInfo keyInfo, WorkbenchSettings settings, CancellationTokenSource cts)
+    {
+        var key = keyInfo.Key;
+        var actionState = (WorkbenchActionState)_actionState;
+        var view = (WorkbenchView)_currentView;
+
+        // When a completed action is showing its result, any key dismisses it.
+        if (actionState == WorkbenchActionState.Completed)
+        {
+            _actionState = (int)WorkbenchActionState.None;
+            _actionResult = string.Empty;
+            _keyPressSignal.TrySetResult();
+            return;
+        }
+
+        // While executing, ignore all input except quit.
+        if (actionState == WorkbenchActionState.Executing)
+        {
+            if (key == ConsoleKey.Q) cts.Cancel();
+            return;
+        }
+
+        // Confirmation prompt — only Y/N/Escape are valid.
+        if (actionState == WorkbenchActionState.AwaitingConfirmation)
+        {
+            switch (key)
+            {
+                case ConsoleKey.Y:
+                    _ = Task.Run(() => ExecuteActionAsync(cts.Token), cts.Token);
+                    break;
+                case ConsoleKey.N:
+                case ConsoleKey.Escape:
+                    _pendingAction = null;
+                    _actionState = (int)WorkbenchActionState.None;
+                    _keyPressSignal.TrySetResult();
+                    break;
+            }
+
+            return;
+        }
+
+        var isDetail = (int)view >= 100;
+
+        // Filter input mode — route all input to the filter until Enter or Escape exits it.
+        if (_filterInputMode != 0)
+        {
+            switch (key)
+            {
+                case ConsoleKey.Escape:
+                    _filterInputMode = 0;
+                    _filter = string.Empty;
+                    _selectedIndex = 0;
+                    break;
+                case ConsoleKey.Enter:
+                    _filterInputMode = 0;
+                    break;
+                case ConsoleKey.Backspace:
+                    if (!string.IsNullOrEmpty(_filter))
+                        _filter = _filter[..^1];
+                    break;
+                case ConsoleKey.UpArrow:
+                    _selectedIndex = Math.Max(0, _selectedIndex - 1);
+                    break;
+                case ConsoleKey.DownArrow:
+                    _selectedIndex++;
+                    break;
+                default:
+                    if (keyInfo.KeyChar >= ' ')
+                    {
+                        _filter += keyInfo.KeyChar;
+                        _selectedIndex = 0;
+                    }
+                    break;
+            }
+            _keyPressSignal.TrySetResult();
+            return;
+        }
+
+        switch (key)
+        {
+            // Primary view number keys — always clear the nav stack and jump directly.
+            case ConsoleKey.D1:
+                _navStack.Clear();
+                _currentView = (int)WorkbenchView.Overview;
+                _selectedIndex = 0;
+                _scrollOffset = 0;
+                _focusedId = string.Empty;
+                _filter = string.Empty;
+                _breadcrumb = [];
+                break;
+            case ConsoleKey.D2:
+                _navStack.Clear();
+                _currentView = (int)WorkbenchView.Observers;
+                _selectedIndex = 0;
+                _scrollOffset = 0;
+                _focusedId = string.Empty;
+                _filter = string.Empty;
+                _breadcrumb = [];
+                break;
+            case ConsoleKey.D3:
+                _navStack.Clear();
+                _currentView = (int)WorkbenchView.FailedPartitions;
+                _selectedIndex = 0;
+                _scrollOffset = 0;
+                _focusedId = string.Empty;
+                _filter = string.Empty;
+                _breadcrumb = [];
+                break;
+            case ConsoleKey.D4:
+                _navStack.Clear();
+                _currentView = (int)WorkbenchView.Jobs;
+                _selectedIndex = 0;
+                _scrollOffset = 0;
+                _focusedId = string.Empty;
+                _filter = string.Empty;
+                _breadcrumb = [];
+                break;
+            case ConsoleKey.D5:
+                _navStack.Clear();
+                _currentView = (int)WorkbenchView.Recommendations;
+                _selectedIndex = 0;
+                _scrollOffset = 0;
+                _focusedId = string.Empty;
+                _filter = string.Empty;
+                _breadcrumb = [];
+                break;
+            case ConsoleKey.D6:
+                _navStack.Clear();
+                _currentView = (int)WorkbenchView.EventLog;
+                _selectedIndex = 0;
+                _scrollOffset = 0;
+                _focusedId = string.Empty;
+                _filter = string.Empty;
+                _eventLogPage = 0;
+                _breadcrumb = [];
+                break;
+            case ConsoleKey.D7:
+                _navStack.Clear();
+                _currentView = (int)WorkbenchView.EventTypes;
+                _selectedIndex = 0;
+                _scrollOffset = 0;
+                _focusedId = string.Empty;
+                _filter = string.Empty;
+                _breadcrumb = [];
+                break;
+            case ConsoleKey.D8:
+                _navStack.Clear();
+                _currentView = (int)WorkbenchView.Projections;
+                _selectedIndex = 0;
+                _scrollOffset = 0;
+                _focusedId = string.Empty;
+                _filter = string.Empty;
+                _filterInputMode = 0;
+                _breadcrumb = [];
+                break;
+            case ConsoleKey.D9:
+                _navStack.Clear();
+                _currentView = (int)WorkbenchView.ReadModels;
+                _selectedIndex = 0;
+                _scrollOffset = 0;
+                _focusedId = string.Empty;
+                _filter = string.Empty;
+                _filterInputMode = 0;
+                _breadcrumb = [];
+                break;
+            case ConsoleKey.D0:
+                _navStack.Clear();
+                _currentView = (int)WorkbenchView.EventStores;
+                _selectedIndex = 0;
+                _scrollOffset = 0;
+                _focusedId = string.Empty;
+                _filter = string.Empty;
+                _filterInputMode = 0;
+                _breadcrumb = [];
+                break;
+
+            // Left/right — cycle through primary views only (detail views excluded).
+            case ConsoleKey.LeftArrow when !isDetail:
+                _currentView = (_currentView - 1 + ViewCount) % ViewCount;
+                _selectedIndex = 0;
+                _scrollOffset = 0;
+                _filter = string.Empty;
+                _filterInputMode = 0;
+                _eventLogPage = 0;
+                break;
+            case ConsoleKey.RightArrow when !isDetail:
+                _currentView = (_currentView + 1) % ViewCount;
+                _selectedIndex = 0;
+                _scrollOffset = 0;
+                _filter = string.Empty;
+                _filterInputMode = 0;
+                _eventLogPage = 0;
+                break;
+
+            // Up/down — navigate list in primary views and sub-navigation detail views; scroll in other detail views.
+            case ConsoleKey.UpArrow:
+            case ConsoleKey.K:
+                if (isDetail && !HasSubNavigation(view))
+                    _scrollOffset = Math.Max(0, _scrollOffset - 1);
+                else
+                    _selectedIndex = Math.Max(0, _selectedIndex - 1);
+                break;
+            case ConsoleKey.DownArrow:
+                if (isDetail && !HasSubNavigation(view))
+                    _scrollOffset = Math.Min(2000, _scrollOffset + 1);
+                else
+                    _selectedIndex = Math.Min(GetMaxSelectedIndex(), _selectedIndex + 1);
+                break;
+
+            // Enter — drill into detail view for selected item.
+            case ConsoleKey.Enter:
+                HandleEnterKey(settings);
+                return;  // HandleEnterKey fires signal
+
+            // Escape — clear filter if active; otherwise pop nav stack or reset selection.
+            case ConsoleKey.Escape:
+                if (!string.IsNullOrEmpty(_filter))
+                {
+                    _filter = string.Empty;
+                    _selectedIndex = 0;
+                }
+                else if (_navStack.Count > 0)
+                {
+                    NavigateBack();
+                }
+                else
+                {
+                    _selectedIndex = 0;
+                    _scrollOffset = 0;
+                }
+
+                break;
+
+            // Backspace — remove last filter character when a filter is active.
+            case ConsoleKey.Backspace:
+                if (!string.IsNullOrEmpty(_filter))
+                {
+                    _filter = _filter[..^1];
+                    _selectedIndex = 0;
+                }
+
+                break;
+
+            // Interval adjustment.
+            case ConsoleKey.OemPlus:
+            case ConsoleKey.Add:
+                settings.Interval = Math.Min(60, settings.Interval + 1);
+                break;
+            case ConsoleKey.OemMinus:
+            case ConsoleKey.Subtract:
+                settings.Interval = Math.Max(1, settings.Interval - 1);
+                break;
+
+            case ConsoleKey.Q:
+                cts.Cancel();
+                return;
+
+            // --- In-view action keys ---
+            case ConsoleKey.R when view == WorkbenchView.Observers:
+            case ConsoleKey.R when view == WorkbenchView.ObserverDetail:
+            {
+                var obs = view == WorkbenchView.ObserverDetail
+                    ? FindObserverById(_focusedId)
+                    : GetSelectedObserver();
+                if (obs is not null)
+                {
+                    SetPendingAction(new PendingAction(
+                        $"Replay observer '{TruncateForPrompt(obs.Id)}'",
+                        $"Replay started for observer '{TruncateForPrompt(obs.Id)}'",
+                        ct => _services!.Observers.Replay(new Replay
+                        {
+                            EventStore = settings.ResolveEventStore(),
+                            Namespace = settings.ResolveNamespace(),
+                            ObserverId = obs.Id,
+                            EventSequenceId = CliDefaults.DefaultEventSequenceId
+                        })));
+                }
+
+                return;
+            }
+
+            case ConsoleKey.T when view == WorkbenchView.FailedPartitions:
+            case ConsoleKey.T when view == WorkbenchView.FailedPartitionDetail:
+            {
+                var fp = view == WorkbenchView.FailedPartitionDetail
+                    ? FindFailedPartitionByFocusedId(_focusedId)
+                    : GetSelectedFailedPartition();
+                if (fp is not null)
+                {
+                    SetPendingAction(new PendingAction(
+                        $"Retry partition '{fp.Partition}' of '{TruncateForPrompt(fp.ObserverId)}'",
+                        $"Retry started for partition '{fp.Partition}'",
+                        ct => _services!.Observers.RetryPartition(new RetryPartition
+                        {
+                            EventStore = settings.ResolveEventStore(),
+                            Namespace = settings.ResolveNamespace(),
+                            ObserverId = fp.ObserverId,
+                            Partition = fp.Partition,
+                            EventSequenceId = CliDefaults.DefaultEventSequenceId
+                        })));
+                }
+
+                return;
+            }
+
+            case ConsoleKey.P when view == WorkbenchView.FailedPartitions:
+            case ConsoleKey.P when view == WorkbenchView.FailedPartitionDetail:
+            {
+                var fp = view == WorkbenchView.FailedPartitionDetail
+                    ? FindFailedPartitionByFocusedId(_focusedId)
+                    : GetSelectedFailedPartition();
+                if (fp is not null)
+                {
+                    SetPendingAction(new PendingAction(
+                        $"Replay partition '{fp.Partition}' of '{TruncateForPrompt(fp.ObserverId)}'",
+                        $"Replay started for partition '{fp.Partition}'",
+                        ct => _services!.Observers.ReplayPartition(new ReplayPartition
+                        {
+                            EventStore = settings.ResolveEventStore(),
+                            Namespace = settings.ResolveNamespace(),
+                            ObserverId = fp.ObserverId,
+                            Partition = fp.Partition,
+                            EventSequenceId = CliDefaults.DefaultEventSequenceId
+                        })));
+                }
+
+                return;
+            }
+
+            // In ObserverDetail, P navigates to the projection definition for projection-type observers.
+            case ConsoleKey.P when view == WorkbenchView.ObserverDetail:
+            {
+                var obs = FindObserverById(_focusedId);
+                if (obs is not null && _lastData is not null
+                    && obs.Type.ToString().Contains("Projection", StringComparison.OrdinalIgnoreCase))
+                {
+                    var projDef = _lastData.ProjectionDefinitions
+                        .FirstOrDefault(d => string.Equals(d.Identifier, obs.Id, StringComparison.OrdinalIgnoreCase));
+                    if (projDef is not null)
+                        NavigateToDetail(WorkbenchView.ProjectionDetail, projDef.Identifier);
+                }
+
+                break;
+            }
+
+            case ConsoleKey.S when view == WorkbenchView.Jobs:
+            {
+                var job = GetSelectedJob();
+                if (job is not null)
+                {
+                    SetPendingAction(new PendingAction(
+                        $"Stop job '{TruncateForPrompt(job.Type ?? job.Id.ToString())}'",
+                        "Job stopped",
+                        ct => _services!.Jobs.Stop(new StopJob
+                        {
+                            EventStore = settings.ResolveEventStore(),
+                            Namespace = settings.ResolveNamespace(),
+                            JobId = job.Id
+                        })));
+                }
+
+                return;
+            }
+
+            case ConsoleKey.U when view == WorkbenchView.Jobs:
+            {
+                var job = GetSelectedJob();
+                if (job is not null)
+                {
+                    SetPendingAction(new PendingAction(
+                        $"Resume job '{TruncateForPrompt(job.Type ?? job.Id.ToString())}'",
+                        "Job resumed",
+                        ct => _services!.Jobs.Resume(new ResumeJob
+                        {
+                            EventStore = settings.ResolveEventStore(),
+                            Namespace = settings.ResolveNamespace(),
+                            JobId = job.Id
+                        })));
+                }
+
+                return;
+            }
+
+            case ConsoleKey.A when view == WorkbenchView.Recommendations:
+            {
+                var rec = GetSelectedRecommendation();
+                if (rec is not null)
+                {
+                    SetPendingAction(new PendingAction(
+                        $"Apply recommendation '{TruncateForPrompt(rec.Name ?? rec.Id.ToString())}'",
+                        "Recommendation applied",
+                        ct => _services!.Recommendations.Perform(new Perform
+                        {
+                            EventStore = settings.ResolveEventStore(),
+                            Namespace = settings.ResolveNamespace(),
+                            RecommendationId = rec.Id
+                        })));
+                }
+
+                return;
+            }
+
+            case ConsoleKey.I when view == WorkbenchView.Recommendations:
+            {
+                var rec = GetSelectedRecommendation();
+                if (rec is not null)
+                {
+                    SetPendingAction(new PendingAction(
+                        $"Ignore recommendation '{TruncateForPrompt(rec.Name ?? rec.Id.ToString())}'",
+                        "Recommendation ignored",
+                        ct => _services!.Recommendations.Ignore(new Perform
+                        {
+                            EventStore = settings.ResolveEventStore(),
+                            Namespace = settings.ResolveNamespace(),
+                            RecommendationId = rec.Id
+                        })));
+                }
+
+                return;
+            }
+
+            // In EventDetail, T navigates to the event type detail view.
+            case ConsoleKey.T when view == WorkbenchView.EventDetail:
+                var evtForType = _lastData?.RecentEvents
+                    .FirstOrDefault(e => e.Context.SequenceNumber.ToString() == _focusedId);
+                var etToNav = evtForType?.Context.EventType;
+                if (etToNav is not null)
+                    NavigateToDetail(WorkbenchView.EventTypeDetail, $"{etToNav.Id}+{etToNav.Generation}");
+                break;
+
+            // Event Log sort order toggle — also resets to page 0.
+            case ConsoleKey.S when view == WorkbenchView.EventLog:
+                _eventLogAscending = 1 - _eventLogAscending;
+                _eventLogPage = 0;
+                _selectedIndex = 0;
+                break;
+
+            // Event Log paging — PageDown goes to older events, PageUp goes to newer.
+            case ConsoleKey.PageDown when view == WorkbenchView.EventLog:
+                var totalEvents = _lastData?.RecentEvents.Count ?? 0;
+                var maxPage = Math.Max(0, (totalEvents - 1) / EventLogPageSize);
+                _eventLogPage = Math.Min(maxPage, _eventLogPage + 1);
+                _selectedIndex = 0;
+                break;
+
+            case ConsoleKey.PageUp when view == WorkbenchView.EventLog:
+                _eventLogPage = Math.Max(0, _eventLogPage - 1);
+                _selectedIndex = 0;
+                break;
+
+            // Quick navigation to context-switching views from any primary view.
+            case ConsoleKey.E when !isDetail:
+                _navStack.Clear();
+                _currentView = (int)WorkbenchView.EventStores;
+                _selectedIndex = 0;
+                _scrollOffset = 0;
+                _filter = string.Empty;
+                _filterInputMode = 0;
+                _focusedId = string.Empty;
+                _breadcrumb = [];
+                break;
+            case ConsoleKey.N when !isDetail:
+                _navStack.Clear();
+                _currentView = (int)WorkbenchView.Namespaces;
+                _selectedIndex = 0;
+                _scrollOffset = 0;
+                _filter = string.Empty;
+                _filterInputMode = 0;
+                _focusedId = string.Empty;
+                _breadcrumb = [];
+                break;
+
+            // '/' activates inline filter mode in filterable views.
+            default:
+                if (!isDetail && IsFilterableView(view) && keyInfo.KeyChar == '/')
+                {
+                    _filterInputMode = 1;
+                    _filter = string.Empty;
+                    _selectedIndex = 0;
+                }
+
+                break;
+        }
+
+        _keyPressSignal.TrySetResult();
+    }
+
+    void HandleEnterKey(WorkbenchSettings settings)
+    {
+        _filter = string.Empty;
+
+        switch ((WorkbenchView)_currentView)
+        {
+            case WorkbenchView.Observers:
+                if (GetSelectedObserver() is { } obs)
+                    NavigateToDetail(WorkbenchView.ObserverDetail, obs.Id);
+                break;
+
+            case WorkbenchView.ObserverDetail:
+                var focusedObs = FindObserverById(_focusedId);
+                var eventTypes = (focusedObs?.EventTypes ?? [])
+                    .OrderBy(et => et.Id, StringComparer.OrdinalIgnoreCase)
+                    .ToList();
+                if (_selectedIndex < eventTypes.Count)
+                {
+                    var et = eventTypes[_selectedIndex];
+                    NavigateToDetail(WorkbenchView.EventTypeDetail, $"{et.Id}+{et.Generation}");
+                }
+
+                break;
+
+            case WorkbenchView.FailedPartitions:
+                if (GetSelectedFailedPartition() is { } fp)
+                    NavigateToDetail(WorkbenchView.FailedPartitionDetail, $"{fp.ObserverId}/{fp.Partition}");
+                break;
+
+            case WorkbenchView.EventLog:
+                if (GetSelectedEvent() is { } evt)
+                    NavigateToDetail(WorkbenchView.EventDetail, evt.Context.SequenceNumber.ToString());
+                break;
+
+            case WorkbenchView.EventTypes:
+                if (GetSelectedEventTypeRegistration() is { } et2)
+                    NavigateToDetail(WorkbenchView.EventTypeDetail, $"{et2.Type.Id}+{et2.Type.Generation}");
+                break;
+
+            case WorkbenchView.Projections:
+                if (GetSelectedProjectionDefinition() is { } proj)
+                    NavigateToDetail(WorkbenchView.ProjectionDetail, proj.Identifier);
+                break;
+
+            case WorkbenchView.ReadModels:
+                if (GetSelectedReadModel() is { } rm)
+                    NavigateToDetail(WorkbenchView.ReadModelDetail, rm.ContainerName);
+                break;
+
+            case WorkbenchView.EventStores:
+                var storeName = GetSelectedEventStoreName();
+                if (storeName is not null)
+                {
+                    _activeEventStore = storeName;
+                    _activeNamespace = null;
+                    _navStack.Clear();
+                    _currentView = (int)WorkbenchView.Overview;
+                    _selectedIndex = 0;
+                    _filter = string.Empty;
+                    _filterInputMode = 0;
+                    _focusedId = string.Empty;
+                    _breadcrumb = [];
+                }
+                break;
+
+            case WorkbenchView.Namespaces:
+                var nsName = GetSelectedNamespaceName();
+                if (nsName is not null)
+                {
+                    _activeNamespace = nsName;
+                    _navStack.Clear();
+                    _currentView = (int)WorkbenchView.Overview;
+                    _selectedIndex = 0;
+                    _filter = string.Empty;
+                    _filterInputMode = 0;
+                    _focusedId = string.Empty;
+                    _breadcrumb = [];
+                }
+                break;
+        }
+
+        _keyPressSignal.TrySetResult();
+    }
+}
diff --git a/Source/Cli/Commands/Chronicle/Workbench/WorkbenchData.cs b/Source/Cli/Commands/Chronicle/Workbench/WorkbenchData.cs
new file mode 100644
index 0000000..169a44f
--- /dev/null
+++ b/Source/Cli/Commands/Chronicle/Workbench/WorkbenchData.cs
@@ -0,0 +1,122 @@
+// Copyright (c) Cratis. All rights reserved.
+// Licensed under the MIT license. See LICENSE file in the project root for full license information.
+
+using Cratis.Chronicle.Contracts.Jobs;
+
+namespace Cratis.Cli.Commands.Chronicle.Workbench;
+
+/// <summary>
+/// Lightweight read-model descriptor used within the workbench (avoids coupling to the concrete contract type).
+/// </summary>
+/// <param name="ContainerName">The container / collection name used to query instances.</param>
+/// <param name="DisplayName">Human-readable display name.</param>
+/// <param name="Owner">Owner (Client or Server).</param>
+/// <param name="IsQueryable">Whether instances can be fetched from the server.</param>
+/// <param name="Source">The source that produced this read model.</param>
+/// <param name="Identifier">The type identifier string.</param>
+public record WorkbenchReadModel(
+    string ContainerName,
+    string DisplayName,
+    string Owner,
+    bool IsQueryable,
+    string Source,
+    string Identifier);
+
+/// <summary>
+/// A point-in-time snapshot of all data fetched from the Chronicle server for the workbench.
+/// </summary>
+/// <param name="ConnectionString">The resolved connection string (may contain credentials).</param>
+/// <param name="EventStore">The event store name.</param>
+/// <param name="Namespace">The namespace name.</param>
+/// <param name="IsConnected">Whether the server was reachable during this fetch.</param>
+/// <param name="ServerVersion">The server version string, or <see langword="null"/> if unreachable.</param>
+/// <param name="EventStoreNames">All event store names registered on the server.</param>
+/// <param name="Observers">All observers in the event store namespace.</param>
+/// <param name="FailedPartitions">All failed observer partitions in the event store namespace.</param>
+/// <param name="Jobs">All background jobs in the event store namespace.</param>
+/// <param name="Recommendations">All pending recommendations in the event store namespace.</param>
+/// <param name="TailSequenceNumber">The current tail of the event log, or <see langword="null"/> if unavailable.</param>
+/// <param name="CapturedAt">When this snapshot was captured.</param>
+/// <param name="FetchError">An error message from the last fetch attempt, or <see langword="null"/> if successful.</param>
+/// <param name="EventTypeRegistrations">All registered event types in the event store.</param>
+/// <param name="ProjectionDefinitions">All projection definitions in the event store.</param>
+/// <param name="ProjectionDeclarations">Full projection declarations (identifier → declaration text).</param>
+/// <param name="RecentEvents">The last 50 events from the event log, newest first.</param>
+/// <param name="ReadModelDefinitions">All read-model definitions registered in the event store.</param>
+/// <param name="NamespaceNames">All namespace names available in the current event store.</param>
+/// <param name="ReadModelInstances">Instances for the currently focused read model (populated in ReadModelDetail view only).</param>
+/// <param name="ReadModelInstancesTotalCount">Total instance count for the currently focused read model.</param>
+/// <param name="ReadModelInstancesError">Error message from the last read model instances fetch, or <see langword="null"/>.</param>
+public record WorkbenchData(
+    string ConnectionString,
+    string EventStore,
+    string Namespace,
+    bool IsConnected,
+    string? ServerVersion,
+    IReadOnlyList<string> EventStoreNames,
+    IReadOnlyList<ObserverInformation> Observers,
+    IReadOnlyList<FailedPartition> FailedPartitions,
+    IReadOnlyList<Job> Jobs,
+    IReadOnlyList<Recommendation> Recommendations,
+    ulong? TailSequenceNumber,
+    DateTimeOffset CapturedAt,
+    string? FetchError,
+    IReadOnlyList<EventTypeRegistration> EventTypeRegistrations,
+    IReadOnlyList<ProjectionDefinition> ProjectionDefinitions,
+    IReadOnlyDictionary<string, string> ProjectionDeclarations,
+    IReadOnlyList<AppendedEvent> RecentEvents,
+    IReadOnlyList<WorkbenchReadModel> ReadModelDefinitions,
+    IReadOnlyList<string> NamespaceNames,
+    IReadOnlyList<string> ReadModelInstances,
+    int ReadModelInstancesTotalCount,
+    string? ReadModelInstancesError)
+{
+    /// <summary>
+    /// Gets the number of observers in the <see cref="ObserverRunningState.Active"/> state.
+    /// </summary>
+    public int ActiveObservers => Observers.Count(o => o.RunningState == ObserverRunningState.Active);
+
+    /// <summary>
+    /// Gets the number of observers in the <see cref="ObserverRunningState.Replaying"/> state.
+    /// </summary>
+    public int ReplayingObservers => Observers.Count(o => o.RunningState == ObserverRunningState.Replaying);
+
+    /// <summary>
+    /// Gets the number of observers in the <see cref="ObserverRunningState.Suspended"/> state.
+    /// </summary>
+    public int SuspendedObservers => Observers.Count(o => o.RunningState == ObserverRunningState.Suspended);
+
+    /// <summary>
+    /// Gets the number of observers in the <see cref="ObserverRunningState.Disconnected"/> state.
+    /// </summary>
+    public int DisconnectedObservers => Observers.Count(o => o.RunningState == ObserverRunningState.Disconnected);
+
+    /// <summary>
+    /// Creates a loading placeholder snapshot before the first real fetch completes.
+    /// </summary>
+    /// <param name="settings">The workbench settings used to resolve connection and store details.</param>
+    /// <returns>A <see cref="WorkbenchData"/> with empty collections and <see cref="IsConnected"/> set to <see langword="false"/>.</returns>
+    public static WorkbenchData Loading(WorkbenchSettings settings) => new(
+        ConnectionString: settings.ResolveConnectionString(),
+        EventStore: settings.ResolveEventStore(),
+        Namespace: settings.ResolveNamespace(),
+        IsConnected: false,
+        ServerVersion: null,
+        EventStoreNames: [],
+        Observers: [],
+        FailedPartitions: [],
+        Jobs: [],
+        Recommendations: [],
+        TailSequenceNumber: null,
+        CapturedAt: DateTimeOffset.Now,
+        FetchError: null,
+        EventTypeRegistrations: [],
+        ProjectionDefinitions: [],
+        ProjectionDeclarations: new Dictionary<string, string>(),
+        RecentEvents: [],
+        ReadModelDefinitions: [],
+        NamespaceNames: [],
+        ReadModelInstances: [],
+        ReadModelInstancesTotalCount: 0,
+        ReadModelInstancesError: null);
+}
diff --git a/Source/Cli/Commands/Chronicle/Workbench/WorkbenchRenderer.DetailViews.cs b/Source/Cli/Commands/Chronicle/Workbench/WorkbenchRenderer.DetailViews.cs
new file mode 100644
index 0000000..c13316f
--- /dev/null
+++ b/Source/Cli/Commands/Chronicle/Workbench/WorkbenchRenderer.DetailViews.cs
@@ -0,0 +1,265 @@
+// Copyright (c) Cratis. All rights reserved.
+// Licensed under the MIT license. See LICENSE file in the project root for full license information.
+
+using System.Text.Json;
+using Spectre.Console.Rendering;
+
+namespace Cratis.Cli.Commands.Chronicle.Workbench;
+
+public static partial class WorkbenchRenderer
+{
+    static readonly JsonSerializerOptions _prettyJson = new() { WriteIndented = true };
+
+    static Rows BuildObserverDetailPage(WorkbenchData data, string focusedId, int selectedIndex)
+    {
+        var obs = data.Observers.FirstOrDefault(o => o.Id == focusedId);
+        if (obs is null)
+        {
+            return new Rows(new Panel(new Markup($"[{_dan}]Observer '{focusedId.EscapeMarkup()}' not found.[/]"))
+                .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Danger).Expand());
+        }
+
+        var lag = ComputeLag(obs, data.TailSequenceNumber);
+        var lagText = lag switch { null => $"[{_mut}]—[/]", 0 => $"[{_suc}]caught up[/]", _ => $"[{_war}]{lag.Value:N0} events behind tail ({data.TailSequenceNumber?.ToString("N0") ?? "—"})[/]" };
+
+        var infoPanel = new Panel(new Rows(
+                new Markup($"  [{_mut}]type[/]  [bold]{obs.Type.ToString().EscapeMarkup()}[/]   [{_mut}]owner[/]  {obs.Owner.ToString().EscapeMarkup()}   [{_mut}]sequence[/]  [{_mut}]{obs.EventSequenceId.EscapeMarkup()}[/]   [{_mut}]replayable[/]  {(obs.IsReplayable ? $"[{_suc}]yes[/]" : $"[{_mut}]no[/]")}"),
+                new Markup($"  [{_mut}]state[/]  {StateIcon(obs.RunningState)} {StateName(obs.RunningState)}   [{_mut}]subscribed[/]  {(obs.IsSubscribed ? $"[{_suc}]yes[/]" : $"[{_dan}]no[/]")}"),
+                new Markup($"  [{_mut}]next seq[/]  {FormatSeq(obs.NextEventSequenceNumber)}   [{_mut}]last handled[/]  {FormatSeq(obs.LastHandledEventSequenceNumber)}   [{_mut}]lag[/]  {lagText}")))
+            .Header($"[{_acc}] {obs.Id.EscapeMarkup()} [/]")
+            .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Accent).Expand();
+
+        var eventTypes = (obs.EventTypes ?? []).OrderBy(et => et.Id, StringComparer.OrdinalIgnoreCase).ToList();
+        IRenderable eventTypesSection;
+        if (eventTypes.Count == 0)
+        {
+            eventTypesSection = new Panel(new Markup($"  [{_mut}]No subscribed event types.[/]"))
+                .Header($"[{_acc}] Subscribed Event Types [/]").Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Muted).Expand();
+        }
+        else
+        {
+            var etEffective = Math.Min(selectedIndex, eventTypes.Count - 1);
+            var etTable = new Table().Border(TableBorder.Rounded).BorderColor(OutputFormatter.Muted).Expand()
+                .AddColumn(new TableColumn(string.Empty).Width(2))
+                .AddColumn(new TableColumn("[bold]EventType[/]").Padding(1, 0))
+                .AddColumn(new TableColumn("[bold]Generation[/]").Padding(1, 0).Width(12).NoWrap());
+
+            var (winStart, winEnd) = ListWindow(eventTypes.Count, etEffective);
+            if (winStart > 0)
+                etTable.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↑ {winStart} more above[/]"), new Markup(string.Empty));
+            for (var i = winStart; i <= winEnd; i++)
+            {
+                var et = eventTypes[i];
+                var isSel = i == etEffective;
+                etTable.AddRow(
+                    new Markup(isSel ? $"[bold {_acc}]▶[/]" : string.Empty),
+                    new Markup(isSel ? $"[bold {_acc}]{et.Id.EscapeMarkup()}[/]" : et.Id.EscapeMarkup()),
+                    new Markup($"[{_mut}]{et.Generation}[/]"));
+            }
+
+            if (winEnd < eventTypes.Count - 1)
+                etTable.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↓ {eventTypes.Count - 1 - winEnd} more below[/]"), new Markup(string.Empty));
+
+            eventTypesSection = new Panel(etTable)
+                .Header($"[{_acc}] Subscribed Event Types ({eventTypes.Count})  ↑↓ select  [[ Enter ]] view schema [/]")
+                .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Accent).Expand();
+        }
+
+        if (obs.Type.ToString().Contains("Projection", StringComparison.OrdinalIgnoreCase))
+        {
+            var projHint = new Panel(new Markup($"  [{_acc}][[ P ]][/]  [{_mut}]View this observer's projection definition[/]"))
+                .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Muted).Expand();
+            return new Rows(infoPanel, eventTypesSection, projHint);
+        }
+
+        return new Rows(infoPanel, eventTypesSection);
+    }
+
+    static Rows BuildFailedPartitionDetailPage(WorkbenchData data, string focusedId, int scrollOffset)
+    {
+        var sep = focusedId.IndexOf('/');
+        FailedPartition? fp = null;
+        if (sep >= 0)
+        {
+            var obsId = focusedId[..sep];
+            var partition = focusedId[(sep + 1)..];
+            fp = data.FailedPartitions.FirstOrDefault(f => f.ObserverId == obsId && f.Partition == partition);
+        }
+
+        if (fp is null)
+        {
+            return new Rows(new Panel(new Markup($"[{_dan}]Failed partition not found.[/]"))
+                .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Danger).Expand());
+        }
+
+        var attempts = fp.Attempts.OrderByDescending(a => (DateTimeOffset?)a.Occurred ?? DateTimeOffset.MinValue).ToList();
+        var lastAttempt = attempts.FirstOrDefault();
+        var msg = (lastAttempt?.Messages.FirstOrDefault() ?? "—").EscapeMarkup();
+        var stackTrace = lastAttempt?.StackTrace ?? string.Empty;
+
+        var header = new Panel(new Markup(
+                $"  [{_mut}]observer[/]  [{_dan}]{fp.ObserverId.EscapeMarkup()}[/]   [{_mut}]partition[/]  {fp.Partition.EscapeMarkup()}   [{_mut}]attempts[/]  [{_dan}]{attempts.Count}[/]\n" +
+                $"  [{_mut}]last failed[/]  [{_dan}]{(lastAttempt is not null ? lastAttempt.Occurred.ToString() : "—").EscapeMarkup()}[/]"))
+            .Header($"[{_dan}] Failure Detail [/]").Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Danger).Expand();
+
+        var msgPanel = new Panel(new Markup($"  [{_dan}]{msg}[/]"))
+            .Header($"[{_dan}] Last Error [/]").Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Danger).Expand();
+
+        var stackLines = (string.IsNullOrEmpty(stackTrace) ? ["(no stack trace)"] : stackTrace.Split('\n').Select(l => $"  [{_mut}]{l.EscapeMarkup()}[/]").ToList())
+            as IReadOnlyList<string>;
+        var stackPanel = ScrollableText("Stack Trace", stackLines, scrollOffset, OutputFormatter.Muted);
+
+        return new Rows(header, msgPanel, stackPanel);
+    }
+
+    static Rows BuildEventDetailPage(WorkbenchData data, string focusedId, int scrollOffset)
+    {
+        AppendedEvent? evt = null;
+        if (ulong.TryParse(focusedId, out var seq))
+            evt = data.RecentEvents.FirstOrDefault(e => e.Context.SequenceNumber == seq);
+
+        if (evt is null)
+        {
+            return new Rows(new Panel(new Markup($"[{_dan}]Event not found (seq# {focusedId.EscapeMarkup()}).[/]"))
+                .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Danger).Expand());
+        }
+
+        var ctx = evt.Context;
+        var header = new Panel(new Rows(
+                new Markup($"  [{_mut}]seq#[/]  [bold]{ctx.SequenceNumber:N0}[/]   [{_mut}]eventType[/]  [bold {_acc}]{(ctx.EventType?.Id ?? "—").EscapeMarkup()}[/]  [{_mut}]+{ctx.EventType?.Generation}[/]"),
+                new Markup($"  [{_mut}]eventSourceId[/]  {(ctx.EventSourceId ?? "—").EscapeMarkup()}   [{_mut}]occurred[/]  [{_mut}]{ctx.Occurred.ToString().EscapeMarkup()}[/]"),
+                new Markup($"  [{_mut}]correlationId[/]  [{_mut}]{ctx.CorrelationId.ToString().EscapeMarkup()}[/]")))
+            .Header($"[{_acc}] Event seq# {ctx.SequenceNumber:N0} [/]")
+            .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Accent).Expand();
+
+        List<string> contentLines;
+        if (string.IsNullOrWhiteSpace(evt.Content))
+        {
+            contentLines = [$"  [{_mut}](no content)[/]"];
+        }
+        else
+        {
+            try
+            {
+                var pretty = JsonSerializer.Serialize(
+                    JsonSerializer.Deserialize<JsonElement>(evt.Content),
+                    _prettyJson);
+                contentLines = [.. pretty.Split('\n').Select(l => $"  [{_mut}]{l.EscapeMarkup()}[/]")];
+            }
+            catch
+            {
+                contentLines = [$"  [{_mut}]{evt.Content.EscapeMarkup()}[/]"];
+            }
+        }
+
+        var contentPanel = ScrollableText("Content", contentLines, scrollOffset, OutputFormatter.Muted);
+        return new Rows(header, contentPanel);
+    }
+
+    static Rows BuildEventTypeDetailPage(WorkbenchData data, string focusedId, int scrollOffset)
+    {
+        var parts = focusedId.Split('+');
+        var typeId = parts[0];
+        var gen = parts.Length > 1 ? parts[1] : string.Empty;
+        var reg = data.EventTypeRegistrations
+            .FirstOrDefault(r => string.Equals(r.Type.Id, typeId, StringComparison.OrdinalIgnoreCase)
+                && r.Type.Generation.ToString() == gen);
+
+        if (reg is null)
+        {
+            return new Rows(new Panel(new Markup($"[{_dan}]Event type '{focusedId.EscapeMarkup()}' not found.[/]"))
+                .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Danger).Expand());
+        }
+
+        var header = new Panel(new Markup(
+                $"  [{_mut}]id[/]  [bold]{reg.Type.Id.EscapeMarkup()}[/]   [{_mut}]generation[/]  {reg.Type.Generation}   [{_mut}]owner[/]  {reg.Owner.ToString().EscapeMarkup()}   [{_mut}]source[/]  {reg.Source.ToString().EscapeMarkup()}   [{_mut}]tombstone[/]  {reg.Type.Tombstone}"))
+            .Header($"[{_acc}] {reg.Type.Id.EscapeMarkup()} +{reg.Type.Generation} [/]")
+            .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Accent).Expand();
+
+        List<string> schemaLines;
+        if (string.IsNullOrWhiteSpace(reg.Schema))
+        {
+            schemaLines = [$"  [{_mut}](no schema)[/]"];
+        }
+        else
+        {
+            try
+            {
+                var pretty = JsonSerializer.Serialize(
+                    JsonSerializer.Deserialize<JsonElement>(reg.Schema),
+                    _prettyJson);
+                schemaLines = [.. pretty.Split('\n').Select(l => $"  [{_mut}]{l.EscapeMarkup()}[/]")];
+            }
+            catch
+            {
+                schemaLines = [.. reg.Schema.Split('\n').Select(l => $"  [{_mut}]{l.EscapeMarkup()}[/]")];
+            }
+        }
+
+        var schemaPanel = ScrollableText("JSON Schema", schemaLines, scrollOffset, OutputFormatter.Muted);
+        return new Rows(header, schemaPanel);
+    }
+
+    static Rows BuildReadModelDetailPage(WorkbenchData data, string focusedId, int scrollOffset)
+    {
+        var def = data.ReadModelDefinitions
+            .FirstOrDefault(d => string.Equals(d.ContainerName, focusedId, StringComparison.OrdinalIgnoreCase));
+
+        var header = new Panel(new Markup(
+                $"  [{_mut}]container[/]  [bold]{focusedId.EscapeMarkup()}[/]   [{_mut}]owner[/]  {(def?.Owner ?? "—").EscapeMarkup()}   [{_mut}]queryable[/]  {(def?.IsQueryable == true ? $"[{_suc}]yes[/]" : $"[{_mut}]no[/]")}   [{_mut}]total[/]  [bold]{data.ReadModelInstancesTotalCount:N0}[/]"))
+            .Header($"[{_acc}] {focusedId.EscapeMarkup()} — Instances [/]")
+            .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Accent).Expand();
+
+        IReadOnlyList<string> lines;
+        if (data.ReadModelInstancesError is not null)
+        {
+            lines = [$"  [{_dan}]Error fetching instances: {data.ReadModelInstancesError.EscapeMarkup()}[/]"];
+        }
+        else if (data.ReadModelInstances.Count == 0)
+        {
+            lines = [$"  [{_mut}](no instances — model may be client-owned or empty)[/]"];
+        }
+        else
+        {
+            var allLines = new List<string>();
+            for (var i = 0; i < data.ReadModelInstances.Count; i++)
+            {
+                if (i > 0) allLines.Add($"  [{_mut}]─────────────────────────────────────────[/]");
+                allLines.Add($"  [{_acc}]Instance {i + 1}[/]");
+                allLines.AddRange(data.ReadModelInstances[i].Split('\n').Select(l => $"  [{_mut}]{l.EscapeMarkup()}[/]"));
+            }
+            if (data.ReadModelInstancesTotalCount > data.ReadModelInstances.Count)
+                allLines.Add($"\n  [{_mut}]showing {data.ReadModelInstances.Count} of {data.ReadModelInstancesTotalCount:N0} total instances[/]");
+            lines = allLines;
+        }
+
+        var instancesPanel = ScrollableText("Instances  (↑↓ scroll)", lines, scrollOffset, OutputFormatter.Muted);
+        return new Rows(header, instancesPanel);
+    }
+
+    static Rows BuildProjectionDetailPage(WorkbenchData data, string focusedId, int scrollOffset)
+    {
+        var def = data.ProjectionDefinitions
+            .FirstOrDefault(d => string.Equals(d.Identifier, focusedId, StringComparison.OrdinalIgnoreCase));
+
+        if (def is null)
+        {
+            return new Rows(new Panel(new Markup($"[{_dan}]Projection '{focusedId.EscapeMarkup()}' not found.[/]"))
+                .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Danger).Expand());
+        }
+
+        data.ProjectionDeclarations.TryGetValue(def.Identifier, out var declaration);
+
+        var header = new Panel(new Markup(
+                $"  [{_mut}]readModel[/]  {def.ReadModel.EscapeMarkup()}   [{_mut}]active[/]  {(def.IsActive ? $"[{_suc}]yes[/]" : $"[{_mut}]no[/]")}   [{_mut}]rewindable[/]  {(def.IsRewindable ? $"[{_suc}]yes[/]" : $"[{_mut}]no[/]")}   [{_mut}]autoMap[/]  {def.AutoMap.ToString().EscapeMarkup()}"))
+            .Header($"[{_acc}] {def.Identifier.EscapeMarkup()} [/]")
+            .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Accent).Expand();
+
+        IReadOnlyList<string> declLines = string.IsNullOrWhiteSpace(declaration)
+            ? [$"  [{_mut}](no declaration available)[/]"]
+            : [.. declaration.Split('\n').Select(l => $"  [{_mut}]{l.EscapeMarkup()}[/]")];
+
+        var declPanel = ScrollableText("Declaration", declLines, scrollOffset, OutputFormatter.Muted);
+        return new Rows(header, declPanel);
+    }
+}
diff --git a/Source/Cli/Commands/Chronicle/Workbench/WorkbenchRenderer.Views.cs b/Source/Cli/Commands/Chronicle/Workbench/WorkbenchRenderer.Views.cs
new file mode 100644
index 0000000..8db9711
--- /dev/null
+++ b/Source/Cli/Commands/Chronicle/Workbench/WorkbenchRenderer.Views.cs
@@ -0,0 +1,655 @@
+// Copyright (c) Cratis. All rights reserved.
+// Licensed under the MIT license. See LICENSE file in the project root for full license information.
+
+using Cratis.Chronicle.Contracts.Jobs;
+using Spectre.Console.Rendering;
+
+namespace Cratis.Cli.Commands.Chronicle.Workbench;
+
+public static partial class WorkbenchRenderer
+{
+    const int EventLogPageSize = 50;
+
+    static Rows BuildOverview(WorkbenchData data)
+    {
+        var topRow = new Table().HideHeaders().NoBorder().Expand()
+            .AddColumn(new TableColumn(string.Empty))
+            .AddColumn(new TableColumn(string.Empty));
+        topRow.AddRow(BuildServerPanel(data), BuildObserverStatsPanel(data));
+
+        var contextPanel = new Panel(new Rows(
+                new Markup($"  [{_mut}]event store[/]  [bold {_acc}]{data.EventStore.EscapeMarkup()}[/]   [{_mut}][[ E ]][/] change"),
+                new Markup($"  [{_mut}]namespace[/]     [bold {_acc}]{data.Namespace.EscapeMarkup()}[/]   [{_mut}][[ N ]][/] change")))
+            .Header($"[{_acc}] Active Context [/]")
+            .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Muted).Expand();
+
+        var sections = new List<IRenderable> { topRow, contextPanel };
+        var attentionPanel = BuildAttentionPanel(data);
+        if (attentionPanel is not null) sections.Add(attentionPanel);
+        return new Rows(sections);
+    }
+
+    static Panel BuildServerPanel(WorkbenchData data)
+    {
+        var connLine = data.IsConnected
+            ? $"  [{_suc}]✓[/]  connected  [{_mut}]{(data.ServerVersion ?? "—").EscapeMarkup()}[/]"
+            : $"  [{_dan}]✗[/]  [{_dan}]disconnected[/]";
+        var storesLine = $"  [{_mut}]·[/]  [{_mut}]{data.EventStoreNames.Count} event store{(data.EventStoreNames.Count == 1 ? string.Empty : "s")}[/]";
+        var tailLine = data.TailSequenceNumber.HasValue
+            ? $"  [{_mut}]·[/]  [bold]{data.TailSequenceNumber.Value:N0}[/]  [{_mut}]events in log[/]"
+            : $"  [{_mut}]·  event log unavailable[/]";
+
+        return new Panel(new Rows(new Markup(connLine), new Markup(storesLine), new Markup(tailLine)))
+            .Header($"[{_acc}] Server [/]")
+            .Border(BoxBorder.Rounded)
+            .BorderColor(OutputFormatter.Muted);
+    }
+
+    static Panel BuildObserverStatsPanel(WorkbenchData data)
+    {
+        var total = data.Observers.Count;
+        var line1 = $"  [{_suc}]●[/]  [bold {_suc}]{data.ActiveObservers,3}[/]  [{_suc}]Active[/]     [{_war}]▲[/]  [bold {_war}]{data.ReplayingObservers,3}[/]  [{_war}]Replaying[/]";
+        var line2 = $"  [{_mut}]○[/]  [bold {_mut}]{data.SuspendedObservers,3}[/]  [{_mut}]Suspended[/]  [{_mut}]⊘[/]  [bold {_mut}]{data.DisconnectedObservers,3}[/]  [{_mut}]Disconnected[/]";
+        var failureLine = data.FailedPartitions.Count > 0
+            ? $"  [{_dan}]✗[/]  [{_dan}]{data.FailedPartitions.Count} failed partition{(data.FailedPartitions.Count == 1 ? string.Empty : "s")}[/]"
+            : $"  [{_suc}]✓[/]  [{_suc}]no failed partitions[/]";
+
+        return new Panel(new Rows(new Markup(line1), new Markup(line2), new Markup(failureLine)))
+            .Header($"[{_acc}] Observers ({total}) [/]")
+            .Border(BoxBorder.Rounded)
+            .BorderColor(OutputFormatter.Muted);
+    }
+
+    static Panel? BuildAttentionPanel(WorkbenchData data)
+    {
+        var items = new List<string>();
+        foreach (var fp in data.FailedPartitions.Take(5))
+        {
+            var lastAttempt = fp.Attempts.MaxBy(a => (DateTimeOffset?)a.Occurred ?? DateTimeOffset.MinValue);
+            var msg = (lastAttempt?.Messages.FirstOrDefault() ?? "unknown error").EscapeMarkup();
+            if (msg.Length > 80) msg = msg[..80] + "…";
+            var attempts = fp.Attempts.Count();
+            items.Add($"  [{_dan}]✗[/]  {fp.ObserverId.EscapeMarkup()}  [{_mut}]on[/] {fp.Partition.EscapeMarkup()}  [{_dan}]{msg}[/]  [{_mut}]({attempts} attempt{(attempts == 1 ? string.Empty : "s")})[/]");
+        }
+
+        foreach (var rec in data.Recommendations.Take(3))
+        {
+            var desc = (rec.Description ?? string.Empty).EscapeMarkup();
+            if (desc.Length > 80) desc = desc[..80] + "…";
+            items.Add($"  [{_war}]▲[/]  [{_war}]{(rec.Name ?? string.Empty).EscapeMarkup()}[/]  [{_mut}]{desc}[/]");
+        }
+
+        if (items.Count == 0) return null;
+
+        var borderColor = data.FailedPartitions.Count > 0 ? OutputFormatter.Danger : OutputFormatter.Warning;
+        var headerColor = data.FailedPartitions.Count > 0 ? _dan : _war;
+        return new Panel(new Rows(items.ConvertAll(i => (IRenderable)new Markup(i))))
+            .Header($"[{headerColor}] Attention Needed ({items.Count}) [/]")
+            .Border(BoxBorder.Rounded)
+            .BorderColor(borderColor);
+    }
+
+    static Rows BuildObserversView(WorkbenchData data, int selectedIndex, string filterText)
+    {
+        var allObservers = data.Observers.OrderBy(o => StateOrder(o.RunningState)).ThenBy(o => o.Id).ToList();
+        var observers = string.IsNullOrEmpty(filterText)
+            ? allObservers
+            : [.. allObservers.Where(o => o.Id.Contains(filterText, StringComparison.OrdinalIgnoreCase))];
+        var effectiveIndex = observers.Count > 0 ? Math.Min(selectedIndex, observers.Count - 1) : -1;
+
+        var table = new Table()
+            .Border(TableBorder.Rounded).BorderColor(OutputFormatter.Muted).Expand()
+            .AddColumn(new TableColumn(string.Empty).Width(2))
+            .AddColumn(new TableColumn("[bold]Observer[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]Type[/]").Padding(1, 0).Width(20).NoWrap())
+            .AddColumn(new TableColumn("[bold]State[/]").Padding(1, 0).Width(18).NoWrap());
+
+        AddListRows(table, observers, effectiveIndex, data.TailSequenceNumber, 4);
+
+        var header = string.IsNullOrEmpty(filterText)
+            ? $"[{_acc}] Observers ({data.Observers.Count}) [/]"
+            : $"[{_acc}] Observers ({observers.Count}/{data.Observers.Count}) [/]";
+        var sections = new List<IRenderable>
+        {
+            new Panel(table).Header(header).BorderColor(OutputFormatter.Accent).NoBorder()
+        };
+
+        if (effectiveIndex >= 0 && observers.Count > 0)
+            sections.Add(BuildObserverMiniDetail(observers[effectiveIndex], data.TailSequenceNumber));
+
+        return new Rows(sections);
+    }
+
+    static void AddListRows(Table table, List<ObserverInformation> observers, int effectiveIndex, ulong? tail, int colCount)
+    {
+        if (observers.Count == 0)
+        {
+            var empty = Enumerable.Range(0, colCount).Select(i => i == 1 ? (IRenderable)new Markup($"[{_mut}](none)[/]") : new Markup(string.Empty)).ToArray();
+            table.AddRow(empty);
+            return;
+        }
+
+        var (winStart, winEnd) = ListWindow(observers.Count, effectiveIndex < 0 ? 0 : effectiveIndex);
+        if (winStart > 0)
+            table.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↑ {winStart} more above[/]"), new Markup(string.Empty), new Markup(string.Empty));
+
+        for (var i = winStart; i <= winEnd; i++)
+        {
+            var obs = observers[i];
+            var isSelected = i == effectiveIndex;
+            var pointer = isSelected ? $"[bold {_acc}]▶[/]" : string.Empty;
+            table.AddRow(
+                new Markup(pointer),
+                new Markup(isSelected ? $"[bold {_acc}]{obs.Id.EscapeMarkup()}[/]" : obs.Id.EscapeMarkup()),
+                new Markup($"[{_mut}]{obs.Type.ToString().EscapeMarkup()}[/]"),
+                new Markup($"{StateIcon(obs.RunningState)} {StateName(obs.RunningState)}"));
+        }
+
+        if (winEnd < observers.Count - 1)
+            table.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↓ {observers.Count - 1 - winEnd} more below[/]"), new Markup(string.Empty), new Markup(string.Empty));
+    }
+
+    static Panel BuildObserverMiniDetail(ObserverInformation obs, ulong? tail)
+    {
+        var line1 = $"  [{_mut}]type[/]  [bold]{obs.Type.ToString().EscapeMarkup()}[/]  [{_mut}]state[/]  {StateIcon(obs.RunningState)} {StateName(obs.RunningState)}";
+        var line2 = $"  [{_mut}]next[/]  {FormatSeq(obs.NextEventSequenceNumber)}  [{_mut}]handled[/]  {FormatSeq(obs.LastHandledEventSequenceNumber)}";
+        var line3 = $"\n  [{_acc}][[ Enter ]][/] Full detail  [{_acc}][[ R ]][/] Replay";
+        return new Panel(new Rows(new Markup(line1), new Markup(line2), new Markup(line3)))
+            .Header($"[{_acc}] {obs.Id.EscapeMarkup()} [/]")
+            .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Accent).Expand();
+    }
+
+    static Rows BuildFailedView(WorkbenchData data, int selectedIndex)
+    {
+        if (data.FailedPartitions.Count == 0)
+        {
+            return new Rows(new Panel(new Markup($"\n  [{_suc}]✓  No failed partitions.[/]\n"))
+                .Header($"[{_acc}] Failures [/]").Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Success));
+        }
+
+        var fps = data.FailedPartitions.OrderByDescending(fp => fp.Attempts.Count()).ToList();
+        var effectiveIndex = Math.Min(selectedIndex, fps.Count - 1);
+        var table = new Table().Border(TableBorder.Rounded).BorderColor(OutputFormatter.Danger).Expand()
+            .AddColumn(new TableColumn(string.Empty).Width(2))
+            .AddColumn(new TableColumn("[bold]Observer[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]Partition[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]Attempts[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]Last Failed[/]").Padding(1, 0));
+
+        var (winStart, winEnd) = ListWindow(fps.Count, effectiveIndex);
+        if (winStart > 0)
+            table.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↑ {winStart} more above[/]"), new Markup(string.Empty), new Markup(string.Empty), new Markup(string.Empty));
+        for (var i = winStart; i <= winEnd; i++)
+        {
+            var fp = fps[i];
+            var isSelected = i == effectiveIndex;
+            var lastAttempt = fp.Attempts.MaxBy(a => (DateTimeOffset?)a.Occurred ?? DateTimeOffset.MinValue);
+            var obsId = fp.ObserverId.EscapeMarkup();
+            table.AddRow(
+                new Markup(isSelected ? $"[bold {_acc}]▶[/]" : string.Empty),
+                new Markup(isSelected ? $"[bold {_acc}]{obsId}[/]" : $"[{_dan}]{obsId}[/]"),
+                new Markup(fp.Partition.EscapeMarkup()),
+                new Markup(fp.Attempts.Count().ToString()),
+                new Markup($"[{_dan}]{(lastAttempt is not null ? lastAttempt.Occurred.ToString() : "—").EscapeMarkup()}[/]"));
+        }
+
+        if (winEnd < fps.Count - 1)
+            table.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↓ {fps.Count - 1 - winEnd} more below[/]"), new Markup(string.Empty), new Markup(string.Empty), new Markup(string.Empty));
+
+        var selected = fps[effectiveIndex];
+        var selLastAttempt = selected.Attempts.MaxBy(a => (DateTimeOffset?)a.Occurred ?? DateTimeOffset.MinValue);
+        var msg = (selLastAttempt?.Messages.FirstOrDefault() ?? "—").EscapeMarkup();
+        if (msg.Length > 80) msg = msg[..80] + "…";
+        var miniDetail = new Panel(new Rows(
+                new Markup($"  [{_mut}]partition[/]  {selected.Partition.EscapeMarkup()}  [{_mut}]attempts[/]  [{_dan}]{selected.Attempts.Count()}[/]  [{_mut}]last failed[/]  [{_dan}]{(selLastAttempt is not null ? selLastAttempt.Occurred.ToString() : "—").EscapeMarkup()}[/]"),
+                new Markup($"  [{_dan}]{msg}[/]"),
+                new Markup($"\n  [{_acc}][[ Enter ]][/] Full detail  [{_acc}][[ T ]][/] Retry  [{_acc}][[ P ]][/] Replay partition")))
+            .Header($"[{_dan}] Failure Detail [/]").Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Danger).Expand();
+
+        return new Rows(
+            new Panel(table).Header($"[{_dan}] Failures ({fps.Count}) [/]").BorderColor(OutputFormatter.Danger).NoBorder(),
+            miniDetail);
+    }
+
+    static Rows BuildJobsView(WorkbenchData data, int selectedIndex)
+    {
+        if (data.Jobs.Count == 0)
+        {
+            return new Rows(new Panel(new Markup($"\n  [{_mut}]No background jobs.[/]\n"))
+                .Header($"[{_acc}] Jobs [/]").Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Muted));
+        }
+
+        var jobs = data.Jobs.OrderBy(j => j.Status.ToString()).ToList();
+        var effectiveIndex = Math.Min(selectedIndex, jobs.Count - 1);
+        var table = new Table().Border(TableBorder.Rounded).BorderColor(OutputFormatter.Muted).Expand()
+            .AddColumn(new TableColumn(string.Empty).Width(2))
+            .AddColumn(new TableColumn("[bold]Type[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]Status[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]Progress[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]Created[/]").Padding(1, 0));
+
+        var (winStart, winEnd) = ListWindow(jobs.Count, effectiveIndex);
+        if (winStart > 0)
+            table.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↑ {winStart} more above[/]"), new Markup(string.Empty), new Markup(string.Empty), new Markup(string.Empty));
+        for (var i = winStart; i <= winEnd; i++)
+        {
+            var job = jobs[i];
+            var isSelected = i == effectiveIndex;
+            var statusStr = job.Status.ToString();
+            string statusColor;
+            if (statusStr.Contains("Running", StringComparison.OrdinalIgnoreCase)) statusColor = _suc;
+            else if (statusStr.Contains("Failed", StringComparison.OrdinalIgnoreCase)) statusColor = _dan;
+            else statusColor = _mut;
+            string progressCell;
+            if (job.Progress?.TotalSteps > 0)
+            {
+                var bar = ProgressBar(job.Progress.SuccessfulSteps, job.Progress.TotalSteps);
+                var pct = (int)Math.Min(100, (long)job.Progress.SuccessfulSteps * 100 / job.Progress.TotalSteps);
+                progressCell = $"{bar} {job.Progress.SuccessfulSteps:N0}/{job.Progress.TotalSteps:N0} [{_mut}]{pct}%[/]";
+            }
+            else
+            {
+                progressCell = $"[{_mut}]{(job.Progress?.Message ?? "—").EscapeMarkup()}[/]";
+            }
+
+            table.AddRow(
+                new Markup(isSelected ? $"[bold {_acc}]▶[/]" : string.Empty),
+                new Markup(isSelected ? $"[bold {_acc}]{(job.Type ?? "—").EscapeMarkup()}[/]" : (job.Type ?? "—").EscapeMarkup()),
+                new Markup($"[{statusColor}]{statusStr.EscapeMarkup()}[/]"),
+                new Markup(progressCell),
+                new Markup(job.Created.ToString().EscapeMarkup()));
+        }
+
+        if (winEnd < jobs.Count - 1)
+            table.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↓ {jobs.Count - 1 - winEnd} more below[/]"), new Markup(string.Empty), new Markup(string.Empty), new Markup(string.Empty));
+
+        return new Rows(new Panel(table).Header($"[{_acc}] Jobs ({jobs.Count}) [/]").BorderColor(OutputFormatter.Accent).NoBorder());
+    }
+
+    static Rows BuildRecommendationsView(WorkbenchData data, int selectedIndex)
+    {
+        if (data.Recommendations.Count == 0)
+        {
+            return new Rows(new Panel(new Markup($"\n  [{_suc}]✓  No pending recommendations.[/]\n"))
+                .Header($"[{_acc}] Recommendations [/]").Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Success));
+        }
+
+        var recs = data.Recommendations.ToList();
+        var effectiveIndex = Math.Min(selectedIndex, recs.Count - 1);
+        var table = new Table().Border(TableBorder.Rounded).BorderColor(OutputFormatter.Warning).Expand()
+            .AddColumn(new TableColumn(string.Empty).Width(2))
+            .AddColumn(new TableColumn("[bold]Recommendation[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]Type[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]Occurred[/]").Padding(1, 0));
+
+        var (winStart, winEnd) = ListWindow(recs.Count, effectiveIndex);
+        if (winStart > 0)
+            table.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↑ {winStart} more above[/]"), new Markup(string.Empty), new Markup(string.Empty));
+        for (var i = winStart; i <= winEnd; i++)
+        {
+            var rec = recs[i];
+            var isSelected = i == effectiveIndex;
+            table.AddRow(
+                new Markup(isSelected ? $"[bold {_acc}]▶[/]" : string.Empty),
+                new Markup(isSelected ? $"[bold {_acc}]{(rec.Name ?? "—").EscapeMarkup()}[/]" : $"[{_war}]{(rec.Name ?? "—").EscapeMarkup()}[/]"),
+                new Markup($"[{_mut}]{(rec.Type ?? "—").EscapeMarkup()}[/]"),
+                new Markup($"[{_mut}]{rec.Occurred.ToString().EscapeMarkup()}[/]"));
+        }
+
+        if (winEnd < recs.Count - 1)
+            table.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↓ {recs.Count - 1 - winEnd} more below[/]"), new Markup(string.Empty), new Markup(string.Empty));
+
+        var selRec = recs[effectiveIndex];
+        var desc = (selRec.Description ?? string.Empty).EscapeMarkup();
+        if (desc.Length > 100) desc = desc[..100] + "…";
+        var detail = new Panel(new Rows(
+                new Markup($"  [{_mut}]type[/]  [{_war}]{(selRec.Type ?? "—").EscapeMarkup()}[/]  [{_mut}]occurred[/]  [{_mut}]{selRec.Occurred.ToString().EscapeMarkup()}[/]"),
+                new Markup($"  [{_mut}]{desc}[/]"),
+                new Markup($"\n  [{_acc}][[ A ]][/] Apply  [{_acc}][[ I ]][/] Ignore")))
+            .Header($"[{_war}] {(selRec.Name ?? "—").EscapeMarkup()} [/]")
+            .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Warning).Expand();
+
+        return new Rows(new Panel(table).Header($"[{_war}] Recommendations ({recs.Count}) [/]").BorderColor(OutputFormatter.Warning).NoBorder(), detail);
+    }
+
+    static Rows BuildEventLogView(WorkbenchData data, int selectedIndex, string filterText, bool ascending, int page)
+    {
+        if (data.RecentEvents.Count == 0)
+        {
+            return new Rows(new Panel(new Markup($"\n  [{_mut}]No events in the event log yet.[/]\n"))
+                .Header($"[{_acc}] Event Log [/]").Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Muted));
+        }
+
+        // Sort the full event window, then filter, then page.
+        var sortedEvents = ascending
+            ? [.. data.RecentEvents.OrderBy(e => e.Context.SequenceNumber)]
+            : data.RecentEvents.ToList();
+
+        var filteredEvents = string.IsNullOrEmpty(filterText)
+            ? sortedEvents
+            : [.. sortedEvents.Where(e => (e.Context.EventType?.Id ?? string.Empty).Contains(filterText, StringComparison.OrdinalIgnoreCase))];
+
+        if (filteredEvents.Count == 0)
+        {
+            return new Rows(new Panel(new Markup($"\n  [{_mut}]No events matching '{filterText.EscapeMarkup()}'.[/]\n"))
+                .Header($"[{_acc}] Event Log (0/{sortedEvents.Count}) [/]").Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Muted));
+        }
+
+        var totalPages = Math.Max(1, (filteredEvents.Count + EventLogPageSize - 1) / EventLogPageSize);
+        var effectivePage = Math.Min(page, totalPages - 1);
+        var pageStart = effectivePage * EventLogPageSize;
+        var pageEvents = filteredEvents.Skip(pageStart).Take(EventLogPageSize).ToList();
+        var effectiveIndex = Math.Min(selectedIndex, Math.Max(0, pageEvents.Count - 1));
+
+        var sortIndicator = ascending ? $"[{_mut}]↑ oldest first[/]" : $"[{_mut}]↓ newest first[/]";
+        var pageIndicator = totalPages > 1
+            ? $"  [{_mut}]page {effectivePage + 1}/{totalPages}[/]"
+            : string.Empty;
+
+        var table = new Table().Border(TableBorder.Rounded).BorderColor(OutputFormatter.Muted).Expand()
+            .AddColumn(new TableColumn(string.Empty).Width(2))
+            .AddColumn(new TableColumn("[bold]Seq#[/]").Padding(1, 0).RightAligned())
+            .AddColumn(new TableColumn("[bold]EventType[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]EventSourceId[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]Occurred[/]").Padding(1, 0));
+
+        var (winStart, winEnd) = ListWindow(pageEvents.Count, effectiveIndex);
+        if (winStart > 0)
+            table.AddRow(new Markup(string.Empty), new Markup(string.Empty), new Markup($"[{_mut}]↑ {winStart} more above[/]"), new Markup(string.Empty), new Markup(string.Empty));
+        for (var i = winStart; i <= winEnd; i++)
+        {
+            var evt = pageEvents[i];
+            var ctx = evt.Context;
+            var isSelected = i == effectiveIndex;
+            var typeId = ctx.EventType?.Id ?? "—";
+            table.AddRow(
+                new Markup(isSelected ? $"[bold {_acc}]▶[/]" : string.Empty),
+                new Markup($"[{_mut}]{ctx.SequenceNumber:N0}[/]"),
+                new Markup(isSelected ? $"[bold {_acc}]{typeId.EscapeMarkup()}[/]" : typeId.EscapeMarkup()),
+                new Markup((ctx.EventSourceId ?? string.Empty).EscapeMarkup()),
+                new Markup($"[{_mut}]{ctx.Occurred.ToString().EscapeMarkup()}[/]"));
+        }
+
+        if (winEnd < pageEvents.Count - 1)
+            table.AddRow(new Markup(string.Empty), new Markup(string.Empty), new Markup($"[{_mut}]↓ {pageEvents.Count - 1 - winEnd} more below[/]"), new Markup(string.Empty), new Markup(string.Empty));
+
+        var selEvt = pageEvents[effectiveIndex];
+        var selCtx = selEvt.Context;
+        var miniDetail = new Panel(new Markup(
+                $"  [{_mut}]seq#[/]  [bold]{selCtx.SequenceNumber:N0}[/]  [{_mut}]eventSource[/]  {(selCtx.EventSourceId ?? "—").EscapeMarkup()}  [{_mut}]occurred[/]  [{_mut}]{selCtx.Occurred.ToString().EscapeMarkup()}[/]\n" +
+                $"\n  [{_acc}][[ Enter ]][/] Full event  [{_acc}][[ T ]][/] View event type"))
+            .Header($"[{_acc}] {(selCtx.EventType?.Id ?? "—").EscapeMarkup()} [/]")
+            .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Accent).Expand();
+
+        var filterSuffix = string.IsNullOrEmpty(filterText)
+            ? string.Empty
+            : $" ({filteredEvents.Count}/{sortedEvents.Count} matching)";
+        var logHeader = $"[{_acc}] Event Log{filterSuffix}  {sortIndicator}{pageIndicator}  [[PgDn/PgUp]] page [/]";
+        return new Rows(
+            new Panel(table).Header(logHeader).BorderColor(OutputFormatter.Accent).NoBorder(),
+            miniDetail);
+    }
+
+    static Rows BuildEventTypesView(WorkbenchData data, int selectedIndex, string filterText)
+    {
+        if (data.EventTypeRegistrations.Count == 0)
+        {
+            return new Rows(new Panel(new Markup($"\n  [{_mut}]No event types registered.[/]\n"))
+                .Header($"[{_acc}] Event Types [/]").Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Muted));
+        }
+
+        var allTypes = data.EventTypeRegistrations.OrderBy(r => r.Type.Id, StringComparer.OrdinalIgnoreCase).ThenBy(r => r.Type.Generation).ToList();
+        var types = string.IsNullOrEmpty(filterText)
+            ? allTypes
+            : [.. allTypes.Where(r => r.Type.Id.Contains(filterText, StringComparison.OrdinalIgnoreCase))];
+
+        if (types.Count == 0)
+        {
+            return new Rows(new Panel(new Markup($"\n  [{_mut}]No event types matching '{filterText.EscapeMarkup()}'.[/]\n"))
+                .Header($"[{_acc}] Event Types (0/{allTypes.Count}) [/]").Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Muted));
+        }
+
+        var effectiveIndex = Math.Min(selectedIndex, types.Count - 1);
+        var table = new Table().Border(TableBorder.Rounded).BorderColor(OutputFormatter.Muted).Expand()
+            .AddColumn(new TableColumn(string.Empty).Width(2))
+            .AddColumn(new TableColumn("[bold]EventType[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]Gen[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]Owner[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]Source[/]").Padding(1, 0));
+
+        var (winStart, winEnd) = ListWindow(types.Count, effectiveIndex);
+        if (winStart > 0)
+            table.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↑ {winStart} more above[/]"), new Markup(string.Empty), new Markup(string.Empty), new Markup(string.Empty));
+        for (var i = winStart; i <= winEnd; i++)
+        {
+            var reg = types[i];
+            var isSelected = i == effectiveIndex;
+            table.AddRow(
+                new Markup(isSelected ? $"[bold {_acc}]▶[/]" : string.Empty),
+                new Markup(isSelected ? $"[bold {_acc}]{reg.Type.Id.EscapeMarkup()}[/]" : reg.Type.Id.EscapeMarkup()),
+                new Markup($"[{_mut}]{reg.Type.Generation}[/]"),
+                new Markup($"[{_mut}]{reg.Owner.ToString().EscapeMarkup()}[/]"),
+                new Markup($"[{_mut}]{reg.Source.ToString().EscapeMarkup()}[/]"));
+        }
+
+        if (winEnd < types.Count - 1)
+            table.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↓ {types.Count - 1 - winEnd} more below[/]"), new Markup(string.Empty), new Markup(string.Empty), new Markup(string.Empty));
+
+        var selType = types[effectiveIndex];
+        var miniDetail = new Panel(new Markup(
+                $"  [{_mut}]owner[/]  {selType.Owner.ToString().EscapeMarkup()}  [{_mut}]source[/]  {selType.Source.ToString().EscapeMarkup()}  [{_mut}]tombstone[/]  {selType.Type.Tombstone}\n" +
+                $"\n  [{_acc}][[ Enter ]][/] View schema"))
+            .Header($"[{_acc}] {selType.Type.Id.EscapeMarkup()} +{selType.Type.Generation} [/]")
+            .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Accent).Expand();
+
+        var typesHeader = string.IsNullOrEmpty(filterText)
+            ? $"[{_acc}] Event Types ({allTypes.Count}) [/]"
+            : $"[{_acc}] Event Types ({types.Count}/{allTypes.Count}) [/]";
+        return new Rows(
+            new Panel(table).Header(typesHeader).BorderColor(OutputFormatter.Accent).NoBorder(),
+            miniDetail);
+    }
+
+    static Rows BuildProjectionsView(WorkbenchData data, int selectedIndex, string filterText)
+    {
+        if (data.ProjectionDefinitions.Count == 0)
+        {
+            return new Rows(new Panel(new Markup($"\n  [{_mut}]No projections registered.[/]\n"))
+                .Header($"[{_acc}] Projections [/]").Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Muted));
+        }
+
+        var allProjections = data.ProjectionDefinitions.OrderBy(d => d.Identifier, StringComparer.OrdinalIgnoreCase).ToList();
+        var projections = string.IsNullOrEmpty(filterText)
+            ? allProjections
+            : [.. allProjections.Where(d => d.Identifier.Contains(filterText, StringComparison.OrdinalIgnoreCase))];
+
+        if (projections.Count == 0)
+        {
+            return new Rows(new Panel(new Markup($"\n  [{_mut}]No projections matching '{filterText.EscapeMarkup()}'.[/]\n"))
+                .Header($"[{_acc}] Projections (0/{allProjections.Count}) [/]").Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Muted));
+        }
+
+        var effectiveIndex = Math.Min(selectedIndex, projections.Count - 1);
+        var table = new Table().Border(TableBorder.Rounded).BorderColor(OutputFormatter.Muted).Expand()
+            .AddColumn(new TableColumn(string.Empty).Width(2))
+            .AddColumn(new TableColumn("[bold]Identifier[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]ReadModel[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]Active[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]Rewindable[/]").Padding(1, 0));
+
+        var (winStart, winEnd) = ListWindow(projections.Count, effectiveIndex);
+        if (winStart > 0)
+            table.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↑ {winStart} more above[/]"), new Markup(string.Empty), new Markup(string.Empty), new Markup(string.Empty));
+        for (var i = winStart; i <= winEnd; i++)
+        {
+            var def = projections[i];
+            var isSelected = i == effectiveIndex;
+            var activeIcon = def.IsActive ? $"[{_suc}]✓[/]" : $"[{_mut}]✗[/]";
+            var rewindIcon = def.IsRewindable ? $"[{_suc}]✓[/]" : $"[{_mut}]✗[/]";
+            table.AddRow(
+                new Markup(isSelected ? $"[bold {_acc}]▶[/]" : string.Empty),
+                new Markup(isSelected ? $"[bold {_acc}]{def.Identifier.EscapeMarkup()}[/]" : def.Identifier.EscapeMarkup()),
+                new Markup($"[{_mut}]{def.ReadModel.EscapeMarkup()}[/]"),
+                new Markup(activeIcon),
+                new Markup(rewindIcon));
+        }
+
+        if (winEnd < projections.Count - 1)
+            table.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↓ {projections.Count - 1 - winEnd} more below[/]"), new Markup(string.Empty), new Markup(string.Empty), new Markup(string.Empty));
+
+        var selDef = projections[effectiveIndex];
+        var miniDetail = new Panel(new Markup(
+                $"  [{_mut}]readModel[/]  {selDef.ReadModel.EscapeMarkup()}  [{_mut}]active[/]  {(selDef.IsActive ? $"[{_suc}]yes[/]" : $"[{_mut}]no[/]")}  [{_mut}]rewindable[/]  {(selDef.IsRewindable ? $"[{_suc}]yes[/]" : $"[{_mut}]no[/]")}\n" +
+                $"\n  [{_acc}][[ Enter ]][/] View declaration"))
+            .Header($"[{_acc}] {selDef.Identifier.EscapeMarkup()} [/]")
+            .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Accent).Expand();
+
+        var projHeader = string.IsNullOrEmpty(filterText)
+            ? $"[{_acc}] Projections ({allProjections.Count}) [/]"
+            : $"[{_acc}] Projections ({projections.Count}/{allProjections.Count}) [/]";
+        return new Rows(
+            new Panel(table).Header(projHeader).BorderColor(OutputFormatter.Accent).NoBorder(),
+            miniDetail);
+    }
+
+    static Rows BuildReadModelsView(WorkbenchData data, int selectedIndex, string filterText)
+    {
+        if (data.ReadModelDefinitions.Count == 0)
+        {
+            return new Rows(new Panel(new Markup($"\n  [{_mut}]No read models registered.[/]\n"))
+                .Header($"[{_acc}] Read Models [/]").Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Muted));
+        }
+
+        var allModels = data.ReadModelDefinitions.OrderBy(d => d.ContainerName, StringComparer.OrdinalIgnoreCase).ToList();
+        var models = string.IsNullOrEmpty(filterText)
+            ? allModels
+            : [.. allModels.Where(d => d.ContainerName.Contains(filterText, StringComparison.OrdinalIgnoreCase)
+                || d.DisplayName.Contains(filterText, StringComparison.OrdinalIgnoreCase))];
+
+        if (models.Count == 0)
+        {
+            return new Rows(new Panel(new Markup($"\n  [{_mut}]No read models matching '{filterText.EscapeMarkup()}'.[/]\n"))
+                .Header($"[{_acc}] Read Models (0/{allModels.Count}) [/]").Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Muted));
+        }
+
+        var effectiveIndex = Math.Min(selectedIndex, models.Count - 1);
+        var table = new Table().Border(TableBorder.Rounded).BorderColor(OutputFormatter.Muted).Expand()
+            .AddColumn(new TableColumn(string.Empty).Width(2))
+            .AddColumn(new TableColumn("[bold]Container[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]Display Name[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]Owner[/]").Padding(1, 0))
+            .AddColumn(new TableColumn("[bold]Queryable[/]").Padding(1, 0));
+
+        var (winStart, winEnd) = ListWindow(models.Count, effectiveIndex);
+        if (winStart > 0)
+            table.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↑ {winStart} more above[/]"), new Markup(string.Empty), new Markup(string.Empty), new Markup(string.Empty));
+        for (var i = winStart; i <= winEnd; i++)
+        {
+            var rm = models[i];
+            var isSelected = i == effectiveIndex;
+            var queryIcon = rm.IsQueryable ? $"[{_suc}]✓[/]" : $"[{_mut}]✗[/]";
+            table.AddRow(
+                new Markup(isSelected ? $"[bold {_acc}]▶[/]" : string.Empty),
+                new Markup(isSelected ? $"[bold {_acc}]{rm.ContainerName.EscapeMarkup()}[/]" : rm.ContainerName.EscapeMarkup()),
+                new Markup($"[{_mut}]{rm.DisplayName.EscapeMarkup()}[/]"),
+                new Markup($"[{_mut}]{rm.Owner.EscapeMarkup()}[/]"),
+                new Markup(queryIcon));
+        }
+
+        if (winEnd < models.Count - 1)
+            table.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↓ {models.Count - 1 - winEnd} more below[/]"), new Markup(string.Empty), new Markup(string.Empty), new Markup(string.Empty));
+
+        var selModel = models[effectiveIndex];
+        var miniDetail = new Panel(new Markup(
+                $"  [{_mut}]owner[/]  {selModel.Owner.EscapeMarkup()}  [{_mut}]source[/]  {selModel.Source.EscapeMarkup()}  [{_mut}]queryable[/]  {(selModel.IsQueryable ? $"[{_suc}]yes[/]" : $"[{_mut}]no[/]")}\n" +
+                $"  [{_mut}]identifier[/]  {selModel.Identifier.EscapeMarkup()}\n" +
+                (selModel.IsQueryable ? $"\n  [{_acc}][[ Enter ]][/] View instances" : $"\n  [{_mut}](client-owned — instances are not stored on the server)[/]")))
+            .Header($"[{_acc}] {selModel.ContainerName.EscapeMarkup()} [/]")
+            .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Accent).Expand();
+
+        var rmHeader = string.IsNullOrEmpty(filterText)
+            ? $"[{_acc}] Read Models ({allModels.Count}) [/]"
+            : $"[{_acc}] Read Models ({models.Count}/{allModels.Count}) [/]";
+        return new Rows(
+            new Panel(table).Header(rmHeader).BorderColor(OutputFormatter.Accent).NoBorder(),
+            miniDetail);
+    }
+
+    static Rows BuildEventStoresView(WorkbenchData data, int selectedIndex)
+    {
+        var stores = data.EventStoreNames.Order(StringComparer.OrdinalIgnoreCase).ToList();
+        if (stores.Count == 0)
+        {
+            return new Rows(new Panel(new Markup($"\n  [{_mut}]No event stores found.[/]\n"))
+                .Header($"[{_acc}] Event Stores [/]").Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Muted));
+        }
+
+        var effectiveIndex = Math.Min(selectedIndex, stores.Count - 1);
+        var table = new Table().Border(TableBorder.Rounded).BorderColor(OutputFormatter.Muted).Expand()
+            .AddColumn(new TableColumn(string.Empty).Width(2))
+            .AddColumn(new TableColumn("[bold]Event Store[/]").Padding(1, 0));
+
+        var (winStart, winEnd) = ListWindow(stores.Count, effectiveIndex);
+        if (winStart > 0)
+            table.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↑ {winStart} more above[/]"));
+        for (var i = winStart; i <= winEnd; i++)
+        {
+            var store = stores[i];
+            var isSelected = i == effectiveIndex;
+            var isActive = string.Equals(store, data.EventStore, StringComparison.Ordinal);
+            var label = isActive ? $"[bold]{store.EscapeMarkup()}[/]  [{_suc}]← active[/]" : store.EscapeMarkup();
+            table.AddRow(
+                new Markup(isSelected ? $"[bold {_acc}]▶[/]" : string.Empty),
+                new Markup(isSelected ? $"[bold {_acc}]{store.EscapeMarkup()}[/]" : label));
+        }
+
+        if (winEnd < stores.Count - 1)
+            table.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↓ {stores.Count - 1 - winEnd} more below[/]"));
+
+        var hint = new Panel(new Markup($"  [{_mut}]Press[/]  [{_acc}][[ Enter ]][/]  [{_mut}]to switch to the selected event store.[/]"))
+            .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Muted);
+
+        return new Rows(
+            new Panel(table).Header($"[{_acc}] Event Stores ({stores.Count})  active: {data.EventStore.EscapeMarkup()} [/]").BorderColor(OutputFormatter.Accent).NoBorder(),
+            hint);
+    }
+
+    static Rows BuildNamespacesView(WorkbenchData data, int selectedIndex)
+    {
+        if (data.NamespaceNames.Count == 0)
+        {
+            return new Rows(new Panel(new Markup($"\n  [{_mut}]No namespaces found (or data not yet loaded).[/]\n"))
+                .Header($"[{_acc}] Namespaces [/]").Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Muted));
+        }
+
+        var nsList = data.NamespaceNames.Order(StringComparer.OrdinalIgnoreCase).ToList();
+        var effectiveIndex = Math.Min(selectedIndex, nsList.Count - 1);
+        var table = new Table().Border(TableBorder.Rounded).BorderColor(OutputFormatter.Muted).Expand()
+            .AddColumn(new TableColumn(string.Empty).Width(2))
+            .AddColumn(new TableColumn("[bold]Namespace[/]").Padding(1, 0));
+
+        var (winStart, winEnd) = ListWindow(nsList.Count, effectiveIndex);
+        if (winStart > 0)
+            table.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↑ {winStart} more above[/]"));
+        for (var i = winStart; i <= winEnd; i++)
+        {
+            var ns = nsList[i];
+            var isSelected = i == effectiveIndex;
+            var isActive = string.Equals(ns, data.Namespace, StringComparison.Ordinal);
+            var label = isActive ? $"[bold]{ns.EscapeMarkup()}[/]  [{_suc}]← active[/]" : ns.EscapeMarkup();
+            table.AddRow(
+                new Markup(isSelected ? $"[bold {_acc}]▶[/]" : string.Empty),
+                new Markup(isSelected ? $"[bold {_acc}]{ns.EscapeMarkup()}[/]" : label));
+        }
+
+        if (winEnd < nsList.Count - 1)
+            table.AddRow(new Markup(string.Empty), new Markup($"[{_mut}]↓ {nsList.Count - 1 - winEnd} more below[/]"));
+
+        var hint = new Panel(new Markup($"  [{_mut}]Press[/]  [{_acc}][[ Enter ]][/]  [{_mut}]to switch to the selected namespace.[/]"))
+            .Border(BoxBorder.Rounded).BorderColor(OutputFormatter.Muted);
+
+        return new Rows(
+            new Panel(table).Header($"[{_acc}] Namespaces ({nsList.Count})  active: {data.Namespace.EscapeMarkup()} [/]").BorderColor(OutputFormatter.Accent).NoBorder(),
+            hint);
+    }
+}
diff --git a/Source/Cli/Commands/Chronicle/Workbench/WorkbenchRenderer.cs b/Source/Cli/Commands/Chronicle/Workbench/WorkbenchRenderer.cs
new file mode 100644
index 0000000..3fa62a6
--- /dev/null
+++ b/Source/Cli/Commands/Chronicle/Workbench/WorkbenchRenderer.cs
@@ -0,0 +1,316 @@
+// Copyright (c) Cratis. All rights reserved.
+// Licensed under the MIT license. See LICENSE file in the project root for full license information.
+
+using Cratis.Chronicle.Contracts.Jobs;
+using Spectre.Console.Rendering;
+
+namespace Cratis.Cli.Commands.Chronicle.Workbench;
+
+/// <summary>
+/// Builds Spectre.Console renderables for the Chronicle workbench dashboard.
+/// </summary>
+public static partial class WorkbenchRenderer
+{
+    /// <summary>Maximum number of list rows shown before the sliding window kicks in.</summary>
+    const int MaxListRows = 10;
+
+    /// <summary>Height of the scrollable text viewport in detail views (lines).</summary>
+    const int ScrollViewport = 12;
+
+    static readonly string _acc = OutputFormatter.Accent.ToMarkup();
+    static readonly string _mut = OutputFormatter.Muted.ToMarkup();
+    static readonly string _suc = OutputFormatter.Success.ToMarkup();
+    static readonly string _war = OutputFormatter.Warning.ToMarkup();
+    static readonly string _dan = OutputFormatter.Danger.ToMarkup();
+
+    static readonly (WorkbenchView V, string Key, string Label)[] _views =
+    [
+        (WorkbenchView.Overview,         "1", "Overview"),
+        (WorkbenchView.Observers,        "2", "Observers"),
+        (WorkbenchView.FailedPartitions, "3", "Failures"),
+        (WorkbenchView.Jobs,             "4", "Jobs"),
+        (WorkbenchView.Recommendations,  "5", "Recommendations"),
+        (WorkbenchView.EventLog,         "6", "Event Log"),
+        (WorkbenchView.EventTypes,       "7", "Event Types"),
+        (WorkbenchView.Projections,      "8", "Projections"),
+        (WorkbenchView.ReadModels,       "9", "Read Models"),
+        (WorkbenchView.EventStores,      "0", "Event Stores"),
+        (WorkbenchView.Namespaces,       string.Empty, "Namespaces")
+    ];
+
+    /// <summary>
+    /// Builds the full dashboard renderable for the given data and render state.
+    /// </summary>
+    /// <param name="data">The workbench snapshot to render.</param>
+    /// <param name="state">All render-relevant state for this frame.</param>
+    /// <returns>The renderable dashboard.</returns>
+    public static Rows Build(WorkbenchData data, WorkbenchRenderState state) =>
+        new(BuildHeader(data, state), BuildBody(data, state), BuildFooter(data, state));
+
+    static Panel BuildHeader(WorkbenchData data, WorkbenchRenderState state)
+    {
+        var dot = data.IsConnected ? $"[{_suc}]●[/]" : $"[{_dan}]●[/]";
+        var cs = DisplayConnectionString(data.ConnectionString);
+        var store = $"[bold]{data.EventStore.EscapeMarkup()}[/][{_mut}] / {data.Namespace.EscapeMarkup()}[/]";
+        var titleLine = $"  [{_acc}]◆[/] [bold]CHRONICLE WORKBENCH[/]  [{_mut}]{cs.EscapeMarkup()}[/]  {dot}  {store}";
+
+        // Second line: breadcrumb when in a detail view, otherwise nav tabs.
+        string secondLine;
+        var breadcrumb = state.Breadcrumb;
+        if (breadcrumb is { Count: > 0 })
+        {
+            var path = string.Join($"  [{_mut}]›[/]  ", breadcrumb.Select(s => s.EscapeMarkup()));
+            secondLine = $"  [{_mut}]{path}[/]  [{_mut}]↻ {state.Interval}s[/]";
+        }
+        else
+        {
+            var failuresAlert = data.FailedPartitions.Count > 0;
+            var recsAlert = data.Recommendations.Count > 0;
+            var tabs = string.Join("  ", _views.Select(v =>
+            {
+                var isActive = v.V == state.View;
+                var keyPart = string.IsNullOrEmpty(v.Key) ? string.Empty : $"[[ {v.Key} ]] ";
+                var alert = (v.V == WorkbenchView.FailedPartitions && failuresAlert)
+                    || (v.V == WorkbenchView.Recommendations && recsAlert);
+                if (isActive)
+                    return $"[bold {_acc}]{keyPart}{v.Label.EscapeMarkup()}[/]";
+                if (alert)
+                    return $"[bold {_war}]{keyPart}{v.Label.EscapeMarkup()}[/]";
+                return string.IsNullOrEmpty(v.Key)
+                    ? $"[{_mut}]{v.Label.EscapeMarkup()}[/]"
+                    : $"[{_mut}]{keyPart}[/]{v.Label.EscapeMarkup()}";
+            }));
+            var refreshIndicator = $"[{_mut}]↻ {state.Interval}s[/]";
+            secondLine = $"  {tabs}  {refreshIndicator}";
+        }
+
+        return new Panel(new Rows(new Markup(titleLine), new Markup(secondLine)))
+            .Border(BoxBorder.Heavy)
+            .BorderColor(OutputFormatter.Accent)
+            .Padding(0, 0);
+    }
+
+    static Rows BuildFooter(WorkbenchData data, WorkbenchRenderState state)
+    {
+        var sep = new Rule().RuleStyle(new Style(foreground: OutputFormatter.Muted));
+
+        Markup statusLine;
+        switch (state.ActionState)
+        {
+            case WorkbenchActionState.AwaitingConfirmation:
+                var desc = (state.PendingActionDescription ?? string.Empty).EscapeMarkup();
+                statusLine = new Markup($"  [{_war}]⚡ {desc}?[/]   [{_war}][[ Y ]][/] Confirm   [{_mut}][[ N ]][/] Cancel");
+                break;
+
+            case WorkbenchActionState.Executing:
+                var execDesc = (state.PendingActionDescription ?? "Working").EscapeMarkup();
+                statusLine = new Markup($"  [{_acc}]⟳ {execDesc}…[/]");
+                break;
+
+            case WorkbenchActionState.Completed:
+                var result = (state.ActionResult ?? string.Empty).EscapeMarkup();
+                var resColor = state.IsActionError ? _dan : _suc;
+                var resIcon = state.IsActionError ? "✗" : "✓";
+                statusLine = new Markup($"  [{resColor}]{resIcon} {result}[/]   [{_mut}](press any key to dismiss)[/]");
+                break;
+
+            default:
+                var isDetail = (int)state.View >= 100;
+                var connected = data.IsConnected ? $"[{_suc}]✓ connected[/]" : $"[{_dan}]✗ disconnected[/]";
+                var seqTail = data.TailSequenceNumber.HasValue
+                    ? $"seq# [bold]{data.TailSequenceNumber.Value:N0}[/]"
+                    : $"[{_mut}]seq# —[/]";
+
+                if (isDetail)
+                {
+                    var actions = GetActionHints(state.View);
+                    var scrollHint = $"[{_mut}]↑↓ scroll  [/]";
+                    statusLine = new Markup($"  {connected}  {seqTail}  {scrollHint}{actions}[{_mut}][[ Esc ]][/] Back");
+                }
+                else
+                {
+                    var navHint = HasNavigation(state.View) ? $"[{_mut}]↑↓ select  [/]" : string.Empty;
+                    var enterHint = HasDrillDown(state.View) && !state.FilterInputMode ? $"[{_mut}][[ Enter ]][/] detail  " : string.Empty;
+                    var actions = !state.FilterInputMode ? GetActionHints(state.View) : string.Empty;
+                    string filterIndicator;
+                    if (state.FilterInputMode)
+                        filterIndicator = $"[{_acc}]/ {state.FilterText.EscapeMarkup()}█[/]  [{_mut}][[ Enter ]] confirm  [[ Esc ]] cancel  [/]";
+                    else if (!string.IsNullOrEmpty(state.FilterText))
+                        filterIndicator = $"[{_acc}]/{state.FilterText.EscapeMarkup()}[/]  [{_mut}][[ Esc ]] clear  [/]";
+                    else
+                        filterIndicator = string.Empty;
+                    var filterHint = !state.FilterInputMode && IsFilterableView(state.View) && string.IsNullOrEmpty(state.FilterText)
+                        ? $"[{_mut}][[ / ]] filter  [/]"
+                        : string.Empty;
+                    var quit = $"[{_mut}]← → views  [[ +/- ]] interval ({state.Interval}s)  [[ Q ]] Quit[/]";
+                    statusLine = new Markup($"  {connected}  {seqTail}  {navHint}{enterHint}{filterHint}{actions}{filterIndicator}{quit}");
+                }
+
+                break;
+        }
+
+        return new Rows(sep, statusLine);
+    }
+
+    static IRenderable BuildBody(WorkbenchData data, WorkbenchRenderState state) =>
+        state.View switch
+        {
+            WorkbenchView.Overview => BuildOverview(data),
+            WorkbenchView.Observers => BuildObserversView(data, state.SelectedIndex, state.FilterText),
+            WorkbenchView.FailedPartitions => BuildFailedView(data, state.SelectedIndex),
+            WorkbenchView.Jobs => BuildJobsView(data, state.SelectedIndex),
+            WorkbenchView.Recommendations => BuildRecommendationsView(data, state.SelectedIndex),
+            WorkbenchView.EventLog => BuildEventLogView(data, state.SelectedIndex, state.FilterText, state.EventLogAscending, state.EventLogPage),
+            WorkbenchView.EventTypes => BuildEventTypesView(data, state.SelectedIndex, state.FilterText),
+            WorkbenchView.Projections => BuildProjectionsView(data, state.SelectedIndex, state.FilterText),
+            WorkbenchView.ReadModels => BuildReadModelsView(data, state.SelectedIndex, state.FilterText),
+            WorkbenchView.EventStores => BuildEventStoresView(data, state.SelectedIndex),
+            WorkbenchView.Namespaces => BuildNamespacesView(data, state.SelectedIndex),
+            WorkbenchView.ObserverDetail => BuildObserverDetailPage(data, state.FocusedId, state.SelectedIndex),
+            WorkbenchView.FailedPartitionDetail => BuildFailedPartitionDetailPage(data, state.FocusedId, state.ScrollOffset),
+            WorkbenchView.EventDetail => BuildEventDetailPage(data, state.FocusedId, state.ScrollOffset),
+            WorkbenchView.EventTypeDetail => BuildEventTypeDetailPage(data, state.FocusedId, state.ScrollOffset),
+            WorkbenchView.ProjectionDetail => BuildProjectionDetailPage(data, state.FocusedId, state.ScrollOffset),
+            WorkbenchView.ReadModelDetail => BuildReadModelDetailPage(data, state.FocusedId, state.ScrollOffset),
+            _ => new Markup(string.Empty)
+        };
+
+    static bool IsFilterableView(WorkbenchView view) =>
+        view is WorkbenchView.Observers or WorkbenchView.EventTypes
+            or WorkbenchView.EventLog or WorkbenchView.Projections
+            or WorkbenchView.ReadModels;
+
+    static bool HasNavigation(WorkbenchView view) =>
+        view is WorkbenchView.Observers or WorkbenchView.FailedPartitions
+            or WorkbenchView.Jobs or WorkbenchView.Recommendations
+            or WorkbenchView.EventLog or WorkbenchView.EventTypes
+            or WorkbenchView.Projections or WorkbenchView.ReadModels
+            or WorkbenchView.EventStores or WorkbenchView.Namespaces
+            or WorkbenchView.ObserverDetail;
+
+    static bool HasDrillDown(WorkbenchView view) =>
+        view is WorkbenchView.Observers or WorkbenchView.FailedPartitions
+            or WorkbenchView.EventLog or WorkbenchView.EventTypes
+            or WorkbenchView.Projections or WorkbenchView.ReadModels
+            or WorkbenchView.EventStores or WorkbenchView.Namespaces
+            or WorkbenchView.ObserverDetail;
+
+    static string GetActionHints(WorkbenchView view) => view switch
+    {
+        WorkbenchView.Observers => $"[{_mut}][[ R ]][/] Replay  ",
+        WorkbenchView.FailedPartitions => $"[{_mut}][[ T ]][/] Retry  [{_mut}][[ P ]][/] Replay partition  ",
+        WorkbenchView.Jobs => $"[{_mut}][[ S ]][/] Stop  [{_mut}][[ U ]][/] Resume  ",
+        WorkbenchView.Recommendations => $"[{_mut}][[ A ]][/] Apply  [{_mut}][[ I ]][/] Ignore  ",
+        WorkbenchView.EventLog => $"[{_mut}][[ S ]][/] Sort  [{_mut}][[ PgDn ]][/] Older  [{_mut}][[ PgUp ]][/] Newer  ",
+        WorkbenchView.ObserverDetail => $"[{_mut}][[ R ]][/] Replay  [{_mut}][[ P ]][/] Projection  ",
+        WorkbenchView.FailedPartitionDetail => $"[{_mut}][[ T ]][/] Retry  [{_mut}][[ P ]][/] Replay partition  ",
+        WorkbenchView.EventDetail => $"[{_mut}][[ T ]][/] View event type  ",
+        WorkbenchView.EventStores => $"[{_mut}][[ Enter ]][/] Switch store  ",
+        WorkbenchView.Namespaces => $"[{_mut}][[ Enter ]][/] Switch namespace  ",
+        _ => string.Empty
+    };
+
+    static string StateIcon(ObserverRunningState state) => state switch
+    {
+        ObserverRunningState.Active => $"[{_suc}]●[/]",
+        ObserverRunningState.Replaying => $"[{_war}]▲[/]",
+        ObserverRunningState.Suspended => $"[{_mut}]○[/]",
+        ObserverRunningState.Disconnected => $"[{_mut}]⊘[/]",
+        _ => $"[{_mut}]·[/]"
+    };
+
+    static string StateName(ObserverRunningState state) => state switch
+    {
+        ObserverRunningState.Active => $"[{_suc}]Active[/]",
+        ObserverRunningState.Replaying => $"[{_war}]Replaying[/]",
+        ObserverRunningState.Suspended => $"[{_mut}]Suspended[/]",
+        ObserverRunningState.Disconnected => $"[{_mut}]Disconnected[/]",
+        _ => state.ToString().EscapeMarkup()
+    };
+
+    static int StateOrder(ObserverRunningState state) => state switch
+    {
+        ObserverRunningState.Disconnected => 0,
+        ObserverRunningState.Replaying => 1,
+        ObserverRunningState.Active => 2,
+        ObserverRunningState.Suspended => 3,
+        _ => 4
+    };
+
+    static ulong? ComputeLag(ObserverInformation obs, ulong? tail)
+    {
+        if (!tail.HasValue || obs.LastHandledEventSequenceNumber == ulong.MaxValue)
+        {
+            return tail;
+        }
+
+        return tail.Value > obs.LastHandledEventSequenceNumber
+            ? tail.Value - obs.LastHandledEventSequenceNumber
+            : 0;
+    }
+
+    static string ProgressBar(int success, int total, int width = 16)
+    {
+        if (total == 0) return $"[{_mut}]{new string('░', width)}[/]";
+        var filled = (int)Math.Min(width, (long)success * width / total);
+        return $"[{_suc}]{new string('█', filled)}[/][{_mut}]{new string('░', width - filled)}[/]";
+    }
+
+    static string FormatSeq(ulong seq) =>
+        seq == ulong.MaxValue ? $"[{_mut}]—[/]" : seq.ToString("N0");
+
+    /// <summary>
+    /// Renders a block of text with a scrollable viewport and a "lines X–Y of Z" indicator.
+    /// </summary>
+    /// <param name="title">The panel header title.</param>
+    /// <param name="lines">The full list of lines to display.</param>
+    /// <param name="scrollOffset">The current scroll position (first visible line index).</param>
+    /// <param name="borderColor">The panel border color.</param>
+    /// <returns>A panel showing the visible lines with a scroll indicator.</returns>
+    static Panel ScrollableText(string title, IReadOnlyList<string> lines, int scrollOffset, Color borderColor)
+    {
+        var totalLines = lines.Count;
+        var start = Math.Max(0, Math.Min(scrollOffset, Math.Max(0, totalLines - ScrollViewport)));
+        const int maxOffset = ScrollViewport - 1;
+        var end = Math.Min(totalLines - 1, start + maxOffset);
+        var visible = lines.Skip(start).Take(ScrollViewport).Select(l => (IRenderable)new Markup(l)).ToList();
+        var indicator = totalLines > ScrollViewport
+            ? $"\n  [{_mut}]lines {start + 1}–{end + 1} of {totalLines}  ↑↓ scroll[/]"
+            : string.Empty;
+
+        var content = visible.Count > 0
+            ? (IRenderable)new Rows([.. visible, new Markup(indicator)])
+            : new Markup($"[{_mut}](empty)[/]");
+
+        return new Panel(content)
+            .Header($"[{_acc}] {title.EscapeMarkup()} [/]")
+            .Border(BoxBorder.Rounded)
+            .BorderColor(borderColor)
+            .Expand();
+    }
+
+    static (int Start, int End) ListWindow(int count, int selectedIndex)
+    {
+        if (count <= MaxListRows) return (0, count - 1);
+        const int maxOffset = MaxListRows - 1;
+        var start = Math.Max(0, selectedIndex - (MaxListRows / 2));
+        var end = Math.Min(count - 1, start + maxOffset);
+        start = Math.Max(0, end - maxOffset);
+        return (start, end);
+    }
+
+    static string DisplayConnectionString(string connectionString)
+    {
+        try
+        {
+            var withoutScheme = connectionString.Replace("chronicle://", string.Empty, StringComparison.OrdinalIgnoreCase);
+            var afterAuth = withoutScheme.Contains('@') ? withoutScheme[(withoutScheme.IndexOf('@') + 1)..] : withoutScheme;
+            var hostPort = afterAuth.Split('?')[0].TrimEnd('/');
+            return $"chronicle://{hostPort}";
+        }
+        catch
+        {
+            return connectionString;
+        }
+    }
+}
diff --git a/Source/Cli/Commands/Chronicle/Workbench/WorkbenchSettings.cs b/Source/Cli/Commands/Chronicle/Workbench/WorkbenchSettings.cs
new file mode 100644
index 0000000..4de1ca3
--- /dev/null
+++ b/Source/Cli/Commands/Chronicle/Workbench/WorkbenchSettings.cs
@@ -0,0 +1,18 @@
+// Copyright (c) Cratis. All rights reserved.
+// Licensed under the MIT license. See LICENSE file in the project root for full license information.
+
+namespace Cratis.Cli.Commands.Chronicle.Workbench;
+
+/// <summary>
+/// Settings for the workbench command.
+/// </summary>
+public class WorkbenchSettings : EventStoreSettings
+{
+    /// <summary>
+    /// Gets or sets the refresh interval in seconds.
+    /// </summary>
+    [CommandOption("--interval <SECONDS>")]
+    [Description("Refresh interval in seconds (default: 5)")]
+    [DefaultValue(5)]
+    public int Interval { get; set; } = 5;
+}
diff --git a/Source/Cli/Commands/Chronicle/Workbench/WorkbenchView.cs b/Source/Cli/Commands/Chronicle/Workbench/WorkbenchView.cs
new file mode 100644
index 0000000..6577181
--- /dev/null
+++ b/Source/Cli/Commands/Chronicle/Workbench/WorkbenchView.cs
@@ -0,0 +1,66 @@
+// Copyright (c) Cratis. All rights reserved.
+// Licensed under the MIT license. See LICENSE file in the project root for full license information.
+
+namespace Cratis.Cli.Commands.Chronicle.Workbench;
+
+/// <summary>
+/// The active view displayed in the workbench dashboard.
+/// </summary>
+public enum WorkbenchView
+{
+    // Primary views — cycled with ← → and accessed with keys 1–8.
+
+    /// <summary>Overview showing server health, observer counts, and any failures.</summary>
+    Overview = 0,
+
+    /// <summary>Full list of all observers with their running states and sequence lag.</summary>
+    Observers = 1,
+
+    /// <summary>List of observer partitions that have failed and need attention.</summary>
+    FailedPartitions = 2,
+
+    /// <summary>Background jobs running on the Chronicle server.</summary>
+    Jobs = 3,
+
+    /// <summary>Pending recommendations from the Chronicle server, with apply and ignore actions.</summary>
+    Recommendations = 4,
+
+    /// <summary>The live event log — last 50 events, newest first.</summary>
+    EventLog = 5,
+
+    /// <summary>All registered event types with their JSON schemas.</summary>
+    EventTypes = 6,
+
+    /// <summary>Projection definitions and their full declarations.</summary>
+    Projections = 7,
+
+    /// <summary>Read model definitions and live instances.</summary>
+    ReadModels = 8,
+
+    /// <summary>All event stores on the Chronicle server, with the ability to switch the active store.</summary>
+    EventStores = 9,
+
+    /// <summary>All namespaces in the current event store, with the ability to switch the active namespace.</summary>
+    Namespaces = 10,
+
+    // Detail views — entered with Enter from a primary view, exited with Escape.
+    // Values ≥ 100 are never shown in the tab bar.
+
+    /// <summary>Full-screen detail for a single observer, with a navigable event-type sub-list.</summary>
+    ObserverDetail = 100,
+
+    /// <summary>Full-screen detail for a single failed partition, including attempt history.</summary>
+    FailedPartitionDetail = 101,
+
+    /// <summary>Full-screen detail for a single appended event, including its full JSON content.</summary>
+    EventDetail = 102,
+
+    /// <summary>Full-screen detail for a single event type, including its JSON schema.</summary>
+    EventTypeDetail = 103,
+
+    /// <summary>Full-screen detail for a single projection, including its full declaration.</summary>
+    ProjectionDetail = 104,
+
+    /// <summary>Full-screen detail for a read model, showing its live instances.</summary>
+    ReadModelDetail = 105,
+}