docs: stop PRs from conflicting in HANDOFF.md

[tuckcode]

Why

Every PR was appending a State Origin line and a Recent sessions line near the bottom of docs/HANDOFF.md. Whichever PR merged second always conflicted. That happened on #142, #143, #145, and #148.

What changed

Per-PR notes now live only in docs/plans/handoffs/ as one new file (YYYY-MM-DD-HHMM-<agent>-<topic>.md).

• docs/HANDOFF.md keeps its existing body. The top now says new PRs must add a file under docs/plans/handoffs/ instead of editing this file.
• Agent instructions that said “update HANDOFF in place” now say to add a handoffs file.
• pnpm handoff:check still caps the index at 900 lines, forbids inline ## Session handoff sections, and requires frontmatter. It also requires the new top pointer.

No product code. No State or Recent line added to HANDOFF.md by this PR.

Files

• .claude/commands/handy.md
• .cursor/rules/continual-learning.mdc
• .cursor/skills/rhizome-ship/SKILL.md
• AGENTS.md
• docs/CROSS-MODEL-HANDOFF.md
• docs/GETTING-STARTED.md
• docs/HANDOFF.md
• docs/YOU-SHOULD-KNOW.md
• docs/plans/handoffs/2026-10-11-0755-cursor-grok-4-6-handoff-per-pr-files.md
• docs/plans/rhizome-ship-skill.md
• scripts/check-handoff-shape.mjs

QA

• pnpm handoff:check — 898 lines, pointer + frontmatter OK
• bash .husky/pre-push by hand — passed in 8m 39s (frontend, Rust, Playwright)
• Second hook run on git push — passed in 7m 55s
• No Playwright feature spec or native UI — docs/scripts only

Completion

• Tests/coverage: pre-push frontend + Rust coverage lanes passed; no new product tests
• Codacy: docs/scripts instruction change; no new Critical/High in product code
• Localization: none — English only (C18)
• PostHog: no event needed because no user-facing feature
• Refactoring: none needed
• ADRs: none
• Docs: listed above
• Demo vault dirt checked: empty

Ready for review. Do not merge from this agent. knispo merges.

  

[repowise-bot]

✅ Health of changed files: 9.3 (unchanged)

Check	Result
Coverage	Not set up · add one CI step
Tests	0 of 1 changed code file reached by a test · 1 reached by no test
Security	No new findings on changed lines
Doc drift	5 refuted claims in a doc this PR edits

Doc drift

• docs/CROSS-MODEL-HANDOFF.md line 106 refers to src/hooks/useMcpBridge.ts, which the tree no longer has (high confidence)
• docs/CROSS-MODEL-HANDOFF.md line 227 refers to src/types/laputaTestBridge.ts, which the tree no longer has (high confidence)
• docs/GETTING-STARTED.md line 364 refers to src/mock-tauri.ts, which the tree no longer has (medium confidence)
• +2 more on the PR page

---

📊 See the full report for this PR
_{Blast radius, every caller of the contracts it changes, and health before and after. No sign-in.
Plain markdown for agents}

_{Settings and updates}

_{Updated 2026-10-11 08:07 UTC
Silence one PR with [skip repowise] in the title · Bot settings
⭐ Star Repowise · 📥 Install on another repo}

[codacy-production]

Up to standards ✅

🟢 Issues 0 issues

Results:
0 new issues

View in Codacy

AI Reviewer: first review requested successfully. AI can make mistakes. Always validate suggestions.

_{TIP This summary will be updated as you push new changes.}

[codacy-production]

Pull Request Overview

The handoff-shape checker does not enforce that the required per-PR pointer is located in the expected top section, allowing a pointer placed elsewhere to pass validation. This should be corrected before relying on the checker to enforce the documented workflow. No Codacy issues were reported; Codacy results are up to standards.

About this PR

• Add automated tests for the new handoff checker behavior to prevent regressions in pointer placement, inline-session rejection, line limits, and frontmatter validation.

Test suggestions

• The checker accepts a valid HANDOFF.md index with the required per-PR pointer and valid handoff files.
• The checker rejects HANDOFF.md when the required per-PR pointer is absent.
• The checker rejects inline ## Session handoff sections.
• The checker rejects an index exceeding 900 lines.
• The checker rejects handoff files with missing or invalid frontmatter.

Prompt proposal for missing tests

Consider implementing these tests if applicable:
1. The checker accepts a valid HANDOFF.md index with the required per-PR pointer and valid handoff files.
2. The checker rejects HANDOFF.md when the required per-PR pointer is absent.
3. The checker rejects inline ## Session handoff sections.
4. The checker rejects an index exceeding 900 lines.
5. The checker rejects handoff files with missing or invalid frontmatter.

_{TIP Improve review quality by adding custom instructions
TIP How was this review? Give us feedback}

[codacy-production]

_{🟡 MEDIUM RISK}

Validate that the per-PR pointer appears in the expected top section, rather than using an unrestricted substring search, so the checker actually enforces the documented top-pointer requirement.