Code review
Handle incoming code review feedback with verification before changes. Use when review comments land (human or bot) and before implementing suggestions — especially if feedback is unclear or technically questionable.From its SKILL.md
npx -y skills add AllHailSeizure/synapse --skill code-reviewAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
2.3 KB, as published. Nobody here has run it
Code Review (receiving)
Technical evaluation, not social performance. Verify before implementing. Ask before assuming. Correctness over comfort.
Response pattern
- Read the full feedback without reacting
- Understand — restate the requirement, or ask
- Verify against this codebase
- Evaluate — sound for this stack and design?
- Respond — technical ack or reasoned pushback
- Implement — one item at a time; test each
Forbidden
- Performative agreement ("You're absolutely right!", "Great point!")
- Implementing before verification
- Partial batches when some items are unclear
Clarify unclear items before any implementation. Related items break if you only understood half.
Source handling
From the user
Trusted after understanding. Still ask if scope is fuzzy. Skip fluff — act or give a technical one-liner.
From external / bot reviewers
Before implementing:
- Correct for this codebase?
- Breaks existing behavior?
- Why does current code look that way?
- Platform/compat constraints?
- Does the reviewer have full context?
Wrong or incomplete → push back with evidence. Can't verify → say what you need. Conflicts with prior user decisions → ask the user first.
YAGNI
If "implement it properly" appears: check actual usage. Unused → propose removal. Used → implement properly.
Order
- Clarify unknowns
- Blocking (breaks, security)
- Simple fixes
- Complex fixes
- Test each; watch for regressions (
verification)
Push back when
- Breaks working behavior
- Reviewer lacks context
- Unused feature creep
- Wrong for this stack
- Conflicts with user architecture
How: technical reasoning, specific questions, reference tests/code. No defensiveness.
Acknowledge correctly
✅ Fixed — [what changed]
✅ Good catch: [issue]. Fixed in [place].
✅ [Just fix and show the diff]
❌ Performative praise / gratitude rituals
If you pushed back and were wrong: state the correction factually and fix. No long apology.
GitHub threads
Reply in the review comment thread, not as a new top-level PR comment.