AGENT SKILL

pr-comment-fact-check

STOP and fact-check PR review comments before applying or dismissing reviewer or bot feedback. Use on CodeRabbit, Copilot, Cursor bot, dependabot, or human reviewer comments.

License 许可
MIT
最近更新
2026-11-01
Stars
★ 5
Forks
0

01 前言 Front Matter

name
pr-comment-fact-check
description
STOP and fact-check PR review comments before applying or dismissing reviewer or bot feedback. Use on CodeRabbit, Copilot, Cursor bot, dependabot, or human reviewer comments.

02 资源文件 1

03 SKILL.md 正文

PR comment fact-check

STOP before applying or dismissing any PR comment. Each comment is a claim — verify against the actual code and this repo's authoritative sources (.agents/, docs/architecture.md, JSDoc, tests) before acting. Workflow: WORKFLOW.md.

Verdict Default action
✅ Correct Apply + resolve
⚠️ Partial Apply salvageable part + explain nuance
❌ Hallucinated Push back with evidence; resolve-on-merge-gate exception in WORKFLOW
🕒 Outdated Point at fix commit + resolve
💭 Style Apply if cheap, else defer

Hallucination catalog (scrutinize harder)

Common LLM-reviewer patterns on this repo:

  1. "This isn't tested" without checking siblings — persist-core contracts are pinned across src/core/persist-core.test.ts, src/adapters/codecs/seroval.test.ts, src/adapters/backends/idb.test.ts, src/adapters/sources/tanstack-store.test.ts; useHydrated spans src/adapters/frameworks/react.test.ts (bun, SSR/snapshot) and tests-dom/**/*.test.tsx (vitest, rerender/detach). Verify coverage before accepting.
  2. Type-safety alarms — if bun run typecheck (tsgo) passes, the claim is almost always wrong, or about runtime behavior the type system can't see (then the reviewer must justify with the runtime case).
  3. Generic "best practice" claims unsupported by our rules — "always destructure", "prefer interfaces over types", "add useMemo/useCallback" — stylistic; we either have a rule or we don't. Grep .agents/ for the convention.
  4. Convention citations that don't exist — "this breaks the library's API conventions" — grep .agents/ + docs/architecture.md. If not codified, it's preference, not rule.
  5. Memory-leak / race-condition claims with no concrete trigger — "this could leak" without a scenario is speculation; ask for the path. Real candidates here: hydration-gate ordering, throttle trailing edge, cross-tab listener teardown (destroy()), useSyncExternalStore unmount-detach — demand the specific path.
  6. Wrong API for our seams — bot suggests coupling to a specific store when the seam is PersistableSource; suggests a factory-per-combination when factory policy is "backend earns a factory only when it needs real adaptation (IndexedDB)".
  7. Public-API suggestions that leak internals — PersistStorage.raw is public and typed unknown on purpose (identity compare only); internal listener aliases are deliberately kept out of public signatures. Push back on "tighten this type" / "export this helper".

Anti-patterns

  • ❌ Applying every suggestion to clear the queue.
  • ❌ Replying "fixed!" without verifying.
  • ❌ Dismissing without evidence.
  • ❌ Resolving a thread you rejected as hallucinated (the reviewer needs to see the receipts).

Reference

  • verify-after-each-step — run after applying a fix.
  • harden-pr — optional full pass on the branch once comments are triaged.
  • docs/architecture.md — seam model, public-surface policy, test matrix (authoritative for "is this the right seam / is this tested" claims).

04 同仓库更多 Skills