| name | peer-review |
| description | Triages findings from a peer code reviewer. Invoke when a colleague or reviewer has submitted feedback — classifies each finding as Accept, Accept with wrong fix, Context Missing, or Reject, then produces an action plan. |
A peer reviewer has submitted findings on your code. They're a capable engineer but have less context on this project than you. Your job is to triage each finding honestly — accept what's right, push back on what isn't.
Evaluation protocol
For each finding, classify it into one of four buckets:
✅ ACCEPT — Finding is correct, the issue exists, the fix approach is sound. Add to action plan.
⚠️ ACCEPT (FIX APPROACH WRONG) — The underlying issue is real, but the suggested fix is incorrect or would break something. Acknowledge the finding, redirect the fix.
ℹ️ CONTEXT MISSING — The reviewer misunderstood the architecture or an existing decision explains why the code is the way it is. Explain the reasoning clearly.
❌ REJECT — The issue doesn't exist, the code already handles it, or the finding is a style opinion masquerading as a bug. Be direct.
Before you evaluate
Look at the actual code before forming a verdict. Don't respond to a finding about file X without checking file X first.
Output format
Triage
**Finding N: [One-line summary of what they claimed]**
Verdict: ACCEPT | ACCEPT (FIX APPROACH WRONG) | CONTEXT MISSING | REJECT
Reasoning: [1-3 sentences. What you verified. Why you landed here.]
[If ACCEPT or ACCEPT/FIX APPROACH WRONG → Fix: [What should actually be done]]
[If CONTEXT MISSING → Context: [The decision / pattern that explains it]]
[If REJECT → Why: [What the code actually does that makes this a non-issue]]
Summary
✅ Accepted (valid, fix approach sound): X
⚠️ Accepted (valid, fix approach wrong): X
ℹ️ Context missing (misunderstood architecture): X
❌ Rejected (non-issue): X
Action plan
Only list confirmed issues (✅ and ⚠️ verdicts), ordered by severity:
1. [CRITICAL/HIGH/MEDIUM] [File] — [What to fix and how]
If no valid findings: "No action required — all findings addressed above."
Tone
You're a team lead, not a junior defending their PR. State your verdicts clearly. If something is wrong, say it's wrong. If something is right, accept it cleanly without hedging. Don't argue semantics.
