Sharpen KGraph onboarding copy

README.md

@@ -2,7 +2,7 @@
 
 <img src="media/logo.svg" alt="KGraph Atom Core Logo" width="140" height="140">
 
-# KGraph: Persistent repository intelligence for AI coding tools.
+# KGraph: Local repo memory for AI coding tools.
 
 <strong>atoms · evidence · context packs</strong>
 
@@ -35,10 +35,20 @@
 
 </div>
 
-KGraph gives Codex, GitHub Copilot, Cursor, Claude Code, Gemini CLI, Windsurf, and Cline a local knowledge layer for your repo: file maps, symbols, imports, relationships, and durable knowledge atoms from previous AI sessions. The goal is simple: your assistant should not spend every session re-learning the same codebase.
+KGraph gives AI coding tools a local repo memory: file maps, symbols, imports, relationships, and durable knowledge atoms stored under `.kgraph/`. The goal is simple: stop re-learning the same codebase every session.
 
 The CLI presents this as **Atom Core**: lightweight local atoms plus deterministic repo maps, context packs, and session history that remain inspectable under `.kgraph/`.
 
+## Try It in 30 Seconds
+
+```bash
+npm install -g @kentwynn/kgraph@latest
+kgraph init
+kgraph "auth token refresh"
+```
+
+That gets you a local scan, a focused context response, and durable memory under `.kgraph/` for the next session.
+
 ## The Workflow
 
 Use KGraph with one setup command and one normal daily command:
@@ -148,19 +158,21 @@ From the root of a repository:
 # 1. Create the local KGraph workspace and run the first scan
 kgraph init
 
-# 2. Optional: accept detected AI tool recommendations during init,
-# or add integrations later when you want KGraph-managed instructions
-kgraph integrate add codex copilot cursor claude-code gemini windsurf cline
-
-# 3. Run the normal workflow for a topic
+# 2. Ask for focused context on a real repo topic
 kgraph "auth token refresh"
 
-# 4. Verify the setup and use doctor as the quality gate
+# 3. Optional: verify the workspace and saved intelligence
 kgraph doctor
 ```
 
 `kgraph init` scans once, prints repo language coverage, detects likely local AI tools, and recommends matching integrations. Integrations are still optional: they only write local instruction files so tools know when to run KGraph. They do not start background agents or call AI providers.
 
+If you already know exactly which integrations you want, you can add them later:
+
+```bash
+kgraph integrate add codex copilot cursor claude-code gemini windsurf cline
+```
+
 After useful AI work, assistants save durable runtime-capture notes into `.kgraph/inbox/`. These notes are not project documentation; they are KGraph input files that the next `kgraph` run processes automatically. You can also process them directly with `kgraph update`.
 
 Normal agent flow is intentionally small. For coding context, agents use the

docs/wiki/Home.md

@@ -5,7 +5,7 @@ KGraph is a free, local-first repository intelligence layer for AI coding tools.
 The official npm package is `@kentwynn/kgraph`.
 The official repository is `github.com/kentwynn/KGraph`.
 
-KGraph helps coding agents reuse repo structure, symbols, relationships, and durable knowledge atoms instead of rediscovering the same codebase every session.
+KGraph gives AI coding tools a local repo memory: file maps, symbols, relationships, and durable knowledge atoms stored under `.kgraph/`.
 
 ## Start Here
 
@@ -25,18 +25,12 @@ KGraph helps coding agents reuse repo structure, symbols, relationships, and dur
 - No cloud service requirement
 - No source-code upload
 
-## Quick Install
+## Quick Start
 
 ```bash
 npm install -g @kentwynn/kgraph@latest
-kgraph --version
-```
-
-Or run without installing:
-
-```bash
-npx @kentwynn/kgraph@latest init
-npx @kentwynn/kgraph@latest "auth token refresh"
+kgraph init
+kgraph "auth token refresh"
 ```
 
 KGraph requires Node.js 20 or newer.