| name | sp-receive-review |
| description | Use when responding to code-review feedback — separates signal from noise, acknowledges what's right, defends with reasoning when appropriate, fixes what's wrong, and never accepts drive-by refactors in the same PR. |
| license | MIT |
| metadata | {"author":"Adapted from JZKK720/superpowers (obra/superpowers, MIT)","source":"https://github.com/JZKK720/superpowers","source_skill":"receiving-code-review","version":"1.0.0"} |
Receiving Code Review
The reviewer's job is to find issues. Your job is to triage them, fix the real ones, defend the defensible ones, and not let drive-by refactors leak in.
The triage flow
Receive review feedback
↓
For each finding:
↓
Read the finding carefully. Re-read the code. Re-read the spec.
↓
Classify:
- Right: I missed it. Fix it.
- Wrong: the reviewer is mistaken. Defend with reasoning.
- Out of scope: technically correct, but unrelated. Move to a follow-up.
- Negotiable: depends on team / project norms. Discuss.
↓
Apply the fix (or defend, or move to follow-up).
↓
Reply to the reviewer (acknowledgement + reasoning).
↓
Re-request review.
When the reviewer is right
- Acknowledge plainly: "Good catch. Fixing in the next commit."
- Don't over-explain or apologise. The fix is the apology.
- Don't add "while I was there" changes. Fix only the finding.
When the reviewer is wrong
- Defend with reasoning, not assertion. "I considered that approach. The downside is X; that's why I went with Y."
- Cite the spec, the test, the plan task, or the standard. "The coding-standards doc says N. The test verifies M. The plan task is L."
- If the reviewer still disagrees, that's a discussion, not a fight. The two of you may need a third opinion.
When the reviewer is out of scope
- Acknowledge the finding: "Yes, X is a real issue."
- Decline to fix in this PR: "I'll open a follow-up. This PR is scoped to Y; X is unrelated."
- Open the follow-up (issue, ticket, or a stub PR).
Don't accept drive-by refactors in code-review responses. They bloat the diff, delay the merge, and confuse the next reviewer.
When the review is unclear
- Ask a specific question. "I'm not sure what you mean by 'consider an alternative here'. Can you point at the file and line you're looking at?"
- Don't guess and implement. A wrong implementation is worse than no implementation.
Tone
- Thank the reviewer. They spent time on your code.
- Be specific in your reply. Quote the finding. Say what you did.
- Don't be defensive. The reviewer is on your side; they want the code to be good.
- Don't be sycophantic. "Great point, will fix" without the fix is noise.
Anti-patterns
- Fixing without acknowledging — push a commit without a reply. The reviewer doesn't know whether you saw their feedback.
- Defending without reasoning — "I disagree" without the "because". The reviewer will assume you didn't think it through.
- Accepting drive-by refactors — "Sure, I'll rename the variable while I'm in there" — no. Open a follow-up.
- Re-requesting review without re-running the tests — per
cubecloud-verify. The fix must be verified.
- Responding to many findings in one giant commit — split by finding. Easier to review, easier to revert.
Related skills
cubecloud-request-review — the author side of asking for review.
cubecloud-verify — re-run verification after the fix.
gstack-qa — the pre-ship gate that should run after the review cycle is done.
Source / license
Adapted from JZKK720/superpowers · receiving-code-review, MIT.