← ClaudeAtlas

code-reviewlisted

Runs a structured code review of a diff, pull/merge request, branch, or file set for correctness bugs, error-handling quality, readability/maintainability, API & contract design, test coverage, documentation accuracy, and version-control hygiene — then reports the results as one table (check, area, status, evidence, recommendation). Explicitly does not cover application security (injection, auth, access control, secrets, SSRF — that's the cybersecurity-check skill) or performance/resource efficiency (algorithmic complexity, N+1 queries, memory/CPU usage — that's the performance-audit skill). Use this whenever the user asks for a "code review", "review this PR", "review this diff", "review my code", "review this pull request", "review this merge request", "MR review", "code quality check", "quality audit", "maintainability review", "readability review", "review my changes", "sanity-check this code", "code smell check", "find code smells", "refactor candidates", "test coverage review", "is this well tested", "c
finnley07/AI-SKILLHUB · ★ 0 · Code & Development · score 72
Install: claude install-skill finnley07/AI-SKILLHUB
# Code Review A structured, evidence-based review of a diff, pull/merge request, branch, or file set against correctness, error-handling, readability/maintainability, API-design, test-coverage, documentation, and version-control-hygiene best practice. It is a peer to two other skills and deliberately does not re-cover their ground: - **cybersecurity-check** owns application/infrastructure security and GDPR — injection, auth, access control, secrets, SSRF, headers, rate limiting, and so on. If you spot something that's squarely a security issue (e.g. unparameterized SQL), note it exists in one line and point the user at `cybersecurity-check`, but don't build it out as a full finding here. - **performance-audit** owns algorithmic complexity, N+1 queries, memory/CPU/resource efficiency, and caching. Same treatment: a one-line pointer, not a full finding. This skill's lane is: is the code *correct*, is it *maintainable*, does it handle *failure* well, is its *contract* sane, and is it *tested and documented*. ## Ground rules - **Evidence or it didn't happen.** Every row needs a concrete pointer — a `file:line`, a quoted snippet, a grep match, or (for something code can't answer) an explicit note that it needs human judgment. Don't mark something ✅ because it "looks fine" or "follows best practice" — either you traced the actual logic/test and it holds up, or it doesn't get a ✅. - **This is static, read-only investigation, not execution.** You are reading the di