xpepper/pr-review-agent-skill / pr-review-grill
pr-review-grill
VerifiedUse when addressing open pull-request review comments and the user wants to discuss, challenge, and decide each comment one at a time, applying the decisions in one batch at the end or as each one is made. Trigger for requests to grill review feedback, walk through comments individually, debate whether comments are valid, or work through PR feedback collaboratively. Do not use this skill for fast batch triage; use pr-review-loop instead.
| File | Kind | Size |
|---|---|---|
| README.md | markdown | 1.6 KB |
| SKILL.md | skill md | 16.4 KB |
| references/triage-guide.md | markdown | 2.8 KB |
| references/triage-state.md | markdown | 1.7 KB |
references/triage-guide.md
View on GitHub# Triage Guide
When classifying PR review comments, assign one of five categories:
## MUST_FIX
A blocking issue that must be resolved before the PR can merge.
Indicators:
- Reviewer explicitly says "blocking", "must fix", or "required"
- Correctness bug or security issue
- Breaks tests or build
- Data loss or data-isolation risk
- Violates project conventions that are explicitly documented
## SHOULD_FIX
Non-blocking but worthwhile to address in this PR.
Indicators:
- Reviewer says "nit", "suggestion", "consider", or "ideally"
- Clear readability or maintenance improvement
- Small, safe improvement that reduces future confusion
## PARK
A valid concern intentionally deferred outside this PR.
Use when:
- The fix belongs in a different PR or component
- The effort is disproportionate to the PR scope
- It requires a team/API/architecture decision
- The user explicitly wants to discuss it with the team first
Record the deferral reason. Create a follow-up issue only when the user
explicitly approves that action.
## OUT_OF_SCOPE
Not applicable to this PR and rejected with an explanation.
Use when:
- The comment relies on a disproven premise
- The suggestion conflicts with a stronger project invariant
- The concern refers to code outside this PR's scope
- A prior decision already settles the issue
Distinguish a substantive rejected suggestion from a no-op acknowledgment.
They may need different replies and resolution actions.
## NEEDS_CLARIFICATION
The reviewer's intent is genuinely ambiguous and acting without confirmation
risks the wrong change.
Use when:
- Multiple valid interpretations lead to materially different code
- The suggestion is unclear or self-contradictory
- The reviewer must choose between incompatible API or behavior contracts
Do not use this category to avoid difficult work. Ask one focused question and
leave the thread unresolved.
## Discussion prompts
For every item, examine:
- What premise does the comment depend on?
- Can the premise be verified in code, schema, tests, or documentation?
- Does the repository already enforce a stronger invariant?
- Is the concern about correctness, safety, API contract, observability, style,
or scope?
- Would a reasonable fix change behavior outside the PR's stated intent?
- Can the issue be fixed safely now, or does it require team agreement?
If the comment `body` contains a ` ```suggestion ` block, evaluate it as part of
the classification: does it compile, fit project conventions, and still apply
to the current code? A suggestion on lines the PR has since changed is likely
stale. Copilot autofix changesets are not exposed in the comment `body` by the
REST or GraphQL APIs, so classify those on the prose and do not assume a patch
exists.
When the user supplies a corrected domain fact, replace the old premise in the
decision record and reassess the classification.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |