review-pr

name	review-pr
allowed-tools	Task, Bash(gh:*), Bash(git:*), ExitWorktree, Monitor, PushNotification, TaskStop, Skill, AskUserQuestion, Read, Edit, Write
description	Reviews a pull request: runs its own baseline review of the PR diff, then a persistent Monitor watches CI and incoming reviewer comments, triages each comment through an independent skeptical agent, applies only verified fixes, and commits+pushes via inline git commands until CI passes and no comments remain to adopt — then asks whether to merge. Use this skill when the user asks to "review a PR", "monitor PR review comments", "address reviewer feedback on #123", or "watch CI on a pull request".
argument-hint	<PR number or URL> [--auto-merge]
user-invocable	true

Review a Pull Request

Run the baseline review of the PR diff, then keep a persistent watch over CI and new reviewer comments until the PR settles.

Context

• PR argument: $ARGUMENTS
• PR metadata: !gh pr view "$ARGUMENTS" --json number,title,headRepository,headRepositoryOwner,additions,deletions,headRefName 2>/dev/null || printf 'set %s to a PR number or URL\n' "$ARGUMENTS"
• Remote: !git remote -v 2>/dev/null | head -2
• Auth: !gh auth status 2>&1 | head -3

Phase 1: Baseline Review and Sizing

Goal: Run the initial review, resolve the repo, and pick a poll interval sized to the PR.

Actions:

1. Parse the PR number or URL from $ARGUMENTS. If absent, list open PRs with gh pr list and ask the user which to review. Normalize PR to the bare number before any gh api REST call: gh pr * commands accept a URL, but gh api repos/$REPO/issues/$PR/... interpolates $PR into the URL path and breaks on a full URL — run PR=$(gh pr view "$ARGUMENTS" --json number -q .number) (the Context block already fetches --json number) and use $PR as the number everywhere downstream. Parse --auto-merge from $ARGUMENTS and strip it before resolving the PR number — it is a closeout opt-in (see Phase 5), not part of the PR identifier; treat its absence as the default (explicit AskUserQuestion merge).
2. Run the baseline review — spawn an independent review agent via Task with clean context (it did not author the code) to review the PR diff. Pull the diff with gh pr diff <PR>; pass the agent the PR title/body and the diff, and ask for findings as path:line: issue lines (full prompt in references/review-loop.md, Baseline review agent). Treat its findings as the first [comment] batch — feed them straight into the Phase 3 triage flow before launching the Monitor. Do not act on them inline; the main context is biased (it likely authored the PR) and the same skeptical gatekeeping must apply to the baseline as to live comments.
3. Resolve REPO=<owner>/<repo> from the PR metadata above (fallback: git remote get-url origin parsed into owner/repo).
4. Read PR size from additions+deletions and pick INTERVAL (seconds) from the size table in references/review-loop.md: 180 / 300 / 480 for small / medium / large; floor 60s, cap 7200s (~2h).

Phase 2: Launch the Persistent Monitor

Goal: One background watch streaming CI + comment events across turns.

Action: Launch a single Monitor with persistent: true running ${CLAUDE_PLUGIN_ROOT}/skills/review-pr/scripts/review-loop.sh. The bare path scripts/review-loop.sh does NOT resolve — the skill runs in the PR's repository cwd, not the plugin dir, so the script must be addressed by its absolute plugin path. Pass PR, REPO, and INTERVAL as env vars (the script also accepts --pr/--repo/--interval). Use a specific description like "CI + new comments on PR #<n> (<m> poll)". Do NOT run a foreground while loop. The script is documented in references/review-loop.md.

CRITICAL: Do NOT skip the watch based on a launch-time snapshot. "This repo has no CI workflow, so the watch would spin idly" is a false inference and not a valid reason to skip: CI is only one of the two things watched. Third-party auto-review services (GitHub Copilot code review, CodeRabbit, Greptile, Codex, Sourcery, and similar), org-level bots, and human reviewers post comments on no fixed schedule and are invisible in a launch-time snapshot — a repo with zero workflows can still accumulate a full review thread minutes after the PR opens. An empty .github/workflows/ proves nothing about who will comment.

The only valid skip is an explicit user opt-out ("just baseline review, don't watch"). If CI and reviewers both appear absent AND the user still wants coverage, launch the watch anyway; it costs nothing and emits nothing until something changes.

Phase 3: React to Each Monitor Event

Goal: Fix what is actionable, reject the noise, escalate the ambiguous. Full rules, prompt template, verdict format, and reply/hide/resolve lifecycle in references/review-loop.md.

• [ci] <name>: fail|cancel → fetch logs (gh run view <run-id> --log-failed), apply the fix, commit+push via inline git commands (git add <file> && git commit -m "<type>(<scope>): <summary>" && git push). The push triggers a fresh CI run the same Monitor re-emits. CRITICAL: stop and report (do NOT auto-fix) for auth/permission, missing-secret, flaky, or infrastructure failures.
• [comment] batch → CRITICAL: spawn an independent review-triage Task agent with clean context. Apply ONLY the fix verdicts; reject/escalate the rest. CRITICAL: reply by comment type — inline review comment → gh api repos/$REPO/pulls/$PR/comments/<id>/replies; issue-level comment → gh pr comment (no reply endpoint); review summary → skip reply. Use the id=<n>/node=<id> tokens from each emitted line. Commit+push all fix changes in one round; then hide each fully-addressed comment (fix pushed or reject replied) as OUTDATED via minimizeComment and resolve its thread via resolveReviewThread (inline only). Leave escalate comments open. Send a PushNotification per escalate.
• [comment] ambiguous (design disagreement, scope change, unclear intent) → PushNotification and report; do not guess, reply, hide, or resolve.

CRITICAL mindset: Comments are mostly from other agents (linters, code-review bots) and human reviewers — suggestions to consider, not orders. Default to skepticism; verify each claim against the diff and adopt only what is demonstrably correct and safe. Rejecting a comment is the normal outcome for noise and false positives.

Phase 4: Stop Conditions

Stop the Monitor with TaskStop when EITHER holds — full conditions in references/review-loop.md:

• Normal stop (all three): every [ci] check terminal + passing; every comment reflected on with resolved ones hidden + threads resolved (only escalate items remain visible); user signals done.
• Hard cap (overrides the above): ~2h wall-clock reached OR user explicitly opts out — surface the unsettled state first (which of CI/comments is still open), then stop. Do NOT keep polling because CI is red or comments remain; the cap exists so a stuck PR cannot hold the watch open forever. If the cap hits with everything actually settled (CI terminal + passing, comments all reflected on), that is a closeout trigger, not a stop: proceed to Phase 5's merge ask.

CRITICAL: a temporarily empty comment queue is NOT a stop signal — other agents may post more comments later.

Phase 5: Closeout — Merge Decision First, Then Ceremony

Goal: Once Phase 4 holds, ask the user whether to merge FIRST — before any ceremony. The summary comment and body rewrite run only on a merge choice; "Don't merge" skips the ceremony and goes straight to TaskStop. On a merge choice, post-merge hygiene (linked-worktree removal, switch to main, sync with origin) runs unconditionally. Full templates and ordered steps in references/closeout.md.

CRITICAL constraints (hold even when detail is delegated to L3):

1. Arm the closeout state the moment Phase 4 holds — before anything else: bash ${CLAUDE_PLUGIN_ROOT}/skills/review-pr/scripts/arm-closeout.sh "$PR" (append --auto-merge when the opt-in was parsed in Phase 1). This writes the repo's .git/review-pr-closeout.json, arming the plugin's Stop hook: while the file exists, one turn-end per user turn is blocked with a message naming the missing merge decision — the Phase 5 ask cannot be skipped by a premature stop, and the hook passes through (stop_hook_active) on later end-attempts of the same turn so it does not loop (a user interrupt also bypasses it). When the hook blocks, first verify the pending closeout is real — a stale state file (ask already answered, summary posted, or PR merged without clearing) is a false alarm: judge simple checks directly (gh pr view --json state,mergedAt, the <!-- review-pr:summary --> marker lookup), spawn an independent subagent with clean context for complex or ambiguous situations — see references/closeout.md (When the hook fires). A verified-stale state is cleared, not re-asked. Clear it the moment the decision is resolved — bash ${CLAUDE_PLUGIN_ROOT}/skills/review-pr/scripts/clear-closeout.sh "$PR": after the user answers (any choice, including "Don't merge"), after the auto-merge completes, or after the opt-in aborts. A stale file blocks the next stop; its message repeats the clear path.
2. Ask the merge question the moment Phase 4 holds. Do not post the summary or rewrite the body first — the user's choice gates the ceremony. When everything is settled, the closeout ask is the next step, not more polling.
3. The ceremony runs ONLY on a merge choice: capture the summary comment URL from gh pr comment stdout (SUMMARY_URL=$(gh pr comment …)).
4. The Review-cycle line in the rewritten body MUST contain that literal URL — a count with no link is not a pointer, and the quoted heredoc will not expand $SUMMARY_URL, so paste it.
5. Steps are ordered — the body needs the comment URL, so summary first, body second.
6. Do not sign the summary as AI-generated; body describes the change, comment records the review cycle — keep them distinct.
7. Do not ask to merge or run the ceremony while CI is red or comments remain open; never auto-merge past open escalate items.
8. Merge only after an explicit AskUserQuestion choice (merge [Recommended]/squash/rebase/don't); never --auto. --auto-merge opt-in: when the flag was parsed in Phase 1, skip the but still run the ceremony first, then auto-merge with (NOT ) once CI is green AND every non-escalate comment is triaged — see (Merge decision → Auto-merge branch). If any comment remains open, the opt-in is suspended: () so the hook enforces the explicit ask, fall back to the and surface the escalate items in the question text. Auto-merge is a single-shot choice for this PR; it does not re-arm after a failure or an interrupt.

TaskStop the Monitor after closeout completes — with the closeout state already cleared.

References

• Review Loop: references/review-loop.md - Monitor script, size→INTERVAL table, triage agent prompt, verdict format, lifecycle/stop conditions
• Closeout: references/closeout.md - Summary comment, body rewrite, merge decision, post-merge hygiene constraints
• Commit Standards: references/commit-standards.md - Commit message format for the inline git commit rounds
• Repository Templates: references/repository-templates.md - Contributing guidelines conformance for fixes
• Examples: references/examples.md - Commit message examples