Imported from ucdavis/ai-skills-registry (skills/code-review-general/SKILL.md). Install upstream with npx skills add ucdavis/ai-skills-registry --skill code-review-general. Copyright stays with the author.

Code Review — General

Review Priority Order

1. Correctness — does it do what it should? Are edge cases handled?
2. Security — does it introduce vulnerabilities? (See security skill for checklist.)
3. Performance — N+1 queries, unnecessary allocations in hot paths, missing indexes.
4. Readability — will the next engineer understand this without context?
5. Style — does it match project conventions? (Never block on style alone.)

Red Flags (always block)

• Hardcoded secrets, API keys, or credentials.
• Missing input validation at API/service boundaries.
• Unhandled error paths that leave state corrupted.
• Race conditions in concurrent code.
• Silent exception swallowing (catch (e) {}).

Structural Issues (should fix)

• Functions doing more than one thing.
• Deep nesting — suggest early returns / guard clauses.
• Duplicated logic that should be extracted.
• Magic numbers or strings without named constants.

Suggestions (non-blocking)

• Better variable/function names.
• Simpler implementations.
• Additional test cases for uncovered paths.

Tone

• Quote the specific line. Explain the problem. Offer a concrete alternative.
• Distinguish: "must fix", "should fix", "consider".
• Approve with comments for minor issues — block only for correctness, security, or missing tests.
• Separate your opinion from objective issues.