| name | receiving-feedback |
| description | Acts on code-review feedback (from a human, a PR comment, or an AI reviewer) by verifying each item against the codebase before implementing, and asking when items are unclear. Use when receiving code review comments, PR feedback, or suggestions from another reviewer — before implementing any of them. |
Receiving Feedback
Adapted from the MIT-licensed superpowers plugin.
Code review requires technical evaluation, not a reflex to agree. Verify
before implementing, ask before assuming, and prefer technical correctness
over social comfort.
Workflow
- Read all feedback first. Take in the complete set of comments before
reacting to any single one. Do not reply or start fixing item by item —
items are often related and a partial read leads to a wrong fix.
- Restate each item as a concrete technical requirement. For every
comment, write down in your own words exactly what change it implies. If
an item is genuinely ambiguous, mark it unclear instead of guessing.
- If any item is unclear, ask about all unclear items before implementing
anything. Do not implement the clear items first and ask about the rest
later — resolving the unclear ones can change how the clear ones should
be done.
- Verify each clear item against the actual codebase. For every item,
check:
- Does it match current behavior, or is the reviewer assuming something
that isn't true?
- Would the change break existing tests or other callers?
- Is there a reason the code is the way it is (compatibility, a prior
decision, a constraint the reviewer may not see)?
- Is the suggestion actually used, or does it add something nothing
calls (YAGNI)?
- Evaluate technical soundness for this stack. A suggestion can be
generically reasonable and still wrong for Nico's stack (Go stdlib
patterns, React/TS conventions, sqlc-generated code, etc.). Judge it in
that context, not in the abstract.
- Push back with reasoning when a suggestion is wrong. State the
specific technical reason — a failing assumption, a broken test, a
compatibility constraint — and propose an alternative or ask which
tradeoff to take. Do not rubber-stamp a suggestion just because it came
from a reviewer.
- Implement one item at a time, in this order: blocking issues (bugs,
security) first, then simple fixes, then complex/structural fixes. Verify
each one (tests, build, manual check) before moving to the next.
- Report back factually. State what changed and where, or state why an
item was not implemented. If you pushed back and turned out to be wrong,
say so plainly and fix it — no lengthy apology, no over-explaining.
Constraints
- Acknowledge with the fix itself or a one-line factual note — skip the social
filler ("you're right", "great point", "thanks for catching that").
- Architectural pushback goes to Nico, not into a unilateral decision — if a
suggestion conflicts with an existing architectural choice, flag it and
ask rather than silently overriding either the reviewer or the prior
decision.
- When replying to inline PR comments on GitHub, reply in the comment thread
(
gh api repos/{owner}/{repo}/pulls/{pr}/comments/{id}/replies), not as a
new top-level PR comment.
Quality
- Before presenting results, run the shared self-review checklist. Surface issues in the chat only if found.