docs: keep PR titles/descriptions uncompressed under caveman mode

.claude/skills/handoff/SKILL.md

@@ -35,8 +35,14 @@ force-push, or discard work.
 5. **Push** the feature branch: `git push -u origin <branch>`. The pre-push guards run
  (auto-merge sentinel, format, drift) — heed a block rather than overriding blindly.
 6. **Open a PR** with `gh pr create --base main`, body ending with the Claude Code
- attribution line. Enabling squash-auto-merge (`gh pr merge --squash --auto`) is the
- repo norm but requires explicit user confirmation before enabling; the PR lands on green.
+ attribution line. Write the body from `.github/pull_request_template.md` in full normal
+ prose — exact `## Summary` / `## Verification` / `## Risk and rollout` / (when clinical-risk
+ or RAG-ranking files are touched) `## Clinical Governance Preflight` headings, every governance
+ box checked, and a satisfying `RAG impact:` line — never caveman-compressed; `pr-policy.yml`
+ parses this text verbatim and hard-blocks the merge on a paraphrased or dropped item (see
+ AGENTS.md "External skill precedence"). Enabling squash-auto-merge is the repo norm but
+ requires explicit user confirmation before enabling (`gh pr merge --squash --auto`); the
+ PR lands on green.
 7. **Record** the review with `npm run ledger:append`, passing `--ref <branch>`, `--head`
  (the full 40-character SHA), `--scope`, `--outcome`, and `--checks`. Do not hand-write
  the row into `docs/branch-review-ledger.md`.

AGENTS.md

@@ -262,6 +262,18 @@ Output-style plugins such as caveman mode may compress prose. They must never co
  inspecting already-fetched local refs (`git log`, `git show`) first; `git fetch` or other
  network/provider access requires explicit user confirmation per the "API and provider confirmation
  boundary" section.
+- **PR titles and descriptions are parsed input, not prose.** `.github/workflows/pr-policy.yml`
+ runs `scripts/pr-policy.mjs` against the exact PR title/body text and hard-blocks the merge
+ when a clinical-risk diff lacks a complete `## Clinical Governance Preflight` (every item from
+ `requiredClinicalGovernanceItems` checked) or a RAG-ranking-surface diff lacks a satisfying
+ `RAG impact:` line (see "RAG ranking protection" below). Caveman-style fragment-dropping breaks
+ this exact-format contract — a paraphrased checklist item or a shortened `RAG impact:` reason can
+ silently fail `governanceItemSatisfied`/`ragImpactDeclared` even though the PR is otherwise fine.
+ `gh pr create`/`gh pr edit` bodies and any `PR_POLICY_BODY.md` content therefore always get
+ written in full normal prose from `.github/pull_request_template.md`, regardless of the active
+ output style — this is "commits" territory under the caveman carve-out, not chat. Before
+ push, sanity-check clinical-risk/RAG-ranking bodies against `scripts/pr-policy.mjs`'s
+ `evaluatePullRequestPolicy` shape (run `npm run check:pr-policy` if the script itself changed).
 
 <!-- END:external-skill-precedence -->