receiving-code-review

Receiving Code Review

When to invoke

WHEN receiving review feedback (from code-reviewer agent, human reviewer, or automated checks). BEFORE implementing any suggestion blindly.

This skill bans performative agreement ("I'll fix everything you said") and demands technical evaluation per finding.

Expert Operating Standard

Follow docs/references/skill-expert-operating-standard.md: start from source of truth, preserve retrieval evidence, apply scope safety, use real producers with runtime receipts for durable delegated outputs, verify before completion claims, and keep confidence below gate when evidence is partial.

Step 0 — Read source of truth (required)

1. Read the full review report (don't skim)
2. Read the original spec/plan to understand intent
3. Read the code each finding references
4. Note severity per finding (CRITICAL/MAJOR/MINOR/SUGGESTION)

Decision tree

Per finding, classify:
├─ AGREE — finding correct, fix straightforward → schedule fix
├─ AGREE-WITH-AMENDMENT — finding correct but suggested fix wrong → fix with note
├─ DISAGREE — finding based on misunderstanding → write counterargument with evidence
├─ CLARIFY — finding ambiguous → ask reviewer specific question
└─ DEFER — finding valid but out-of-scope → file follow-up issue

Procedure

1. Read review (Step 0)
2. Classify each finding using decision tree
3. For AGREE: implement fix, run verification
4. For AGREE-WITH-AMENDMENT: implement variant, document why
5. For DISAGREE: write 2-3 sentence counterargument with code/spec references
6. For CLARIFY: ask reviewer one question per ambiguity
7. For DEFER: create issue/note in .supervibe/artifacts/follow-ups.md
8. Mark each finding resolved with link to evidence (commit/PR/issue)
9. Score — supervibe:confidence-scoring artifact-type=agent-output; ≥9 required
10. Re-invoke reviewer if substantive changes made

Output contract

Returns per-finding resolution table:

| Finding | Classification | Action | Evidence |
| #1 ...  | AGREE          | fixed  | abc1234  |
| #2 ...  | DISAGREE       | <argument> | spec.md L42 |
| #3 ...  | DEFER          | follow-up issue #N | .supervibe/artifacts/follow-ups.md |

Guard rails

• DO NOT: blindly implement every suggestion (over-engineering risk)
• DO NOT: dismiss findings without writing counterargument
• DO NOT: ignore CRITICAL findings (must AGREE or escalate)
• DO NOT: claim "fixed" without verification command output
• ALWAYS: classify every finding (no "I'll think about it")
• ALWAYS: ask specific questions for CLARIFY (not "what do you mean?")

Verification

• Every finding has classification + action + evidence
• All AGREE/AGREE-WITH-AMENDMENT have verification output
• All DISAGREE have written counterargument
• No finding left unaddressed

Related

• supervibe:code-review — produces input
• supervibe:requesting-code-review — opposite flow
• supervibe:systematic-debugging — used when finding reveals real bug