feat(v12.2.0): shadcn/improve folds, first /harvest run, CC v2.1.202 sync#124
Merged
Conversation
…sync Three-part batch, no new skills (library stays 37/40): shadcn/improve folds (4 mechanics, skill not adopted — audit surface already owned by adversarial-audit/nuclear-review/review): - plan stamps + reconcile pass (project, orchestrate) - expected-output done criteria + escape hatches (implementer briefing contract) - untrusted-diff rule (review-batch) - considered-and-rejected ledger (adversarial-audit, nuclear-review) First live /harvest run, from two 2026-07-06 programa transcripts (both independently exhibited the top procedure); all three folds trap-validated 4/4 with blind subagents: - verify subagent claims independently (orchestrate) - when CI goes red: reproduce locally, canonical bump scripts, read .conclusion explicitly (ship) - commits named in a bug report are hypotheses; blame first (fix) Claude Code v2.1.202 sync: docs-only — /review reverted to fast single-pass (MANUAL.md note), manifest bump. Dynamic-workflow-size setting is config-state, not a settings.json key; no schema surface.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What this does
Hardens how our agents plan, delegate, and ship — plans now carry the commit they were written against so drift is detectable, subagent "done" reports and CI watcher exit codes are no longer trusted without independent verification, and disproved audit findings get a ledger so they aren't re-litigated. No new skills; every change is a fold into an existing one.
Summary
shadcn/improve folds (skill itself not adopted — its audit surface is already owned by
/adversarial-audit+/nuclear-review+/review):project,orchestrate) — plans record their authoring SHA; reconcile re-runs done-criteria, refreshes drifted refs, retires obsoleted tasksagents/implementer.md, CLAUDE-FULL briefing contract)review-batch) — every hunk must trace to a task stepadversarial-audit,nuclear-review)First live
/harvestrun — extracted from two real programa sessions (2026-07-06), both independently exhibiting the top procedure; all folds validated with 4/4 blind trap prompts:orchestrate) — re-run briefed verification against the real artifact; absorb subagent capability gaps; fix-solo vs re-delegate rule; SendMessage-resume over respawnship) — reproduce the failing guard locally against the real artifact; canonical bump scripts; read.conclusionexplicitly, never the watcher's exit codefix) — commits named in a bug report are hypothesesClaude Code v2.1.202 sync (docs-only):
/review <pr>reverted upstream to fast single-pass — MANUAL.md note updated; manifest bumped. The new dynamic-workflow-size/configsetting is config-state, not asettings.jsonkey, so the strict schema needs nothing.VERSION 12.1.0 → 12.2.0 (setup.ts + plugin.json), CHANGELOG entry.
Test Plan
bun run lint:skills(37, 0 errors) +bun run lint:agents(10, 0 errors)bun run typecheck+bun test(767 pass)bun run upstream:scan— no drift at 2.1.202