agentsclimarketplace
Sign in

Code review

Skill mtvrkan/senior-dev-kit/skills/code-review

Use to review diffs without editing code. Focus on bugs, regressions, security risks, data loss, missing validation, missing tests, and performance risks.From its SKILL.md

Install
npx -y skills add mtvrkan/senior-dev-kit --skill code-review

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 4 stars4 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

2.2 KB, as published. Nobody here has run it

code-review

Review for actionable risks only. Skip subjective style. Read changed files + directly imported deps only. Auto-fires after meaningful changes; also invocable directly via /code-review for a deliberate pre-merge pass. If the diff touches a guarded domain (auth/payment/DB/CI), raise effort and escalate per the last line.

Do not use for: a dedicated, exhaustive security audit (security-review) — items (5) and (7) below are an inline sanity check on the diff, not a substitute for one; dependency CVE checks (security-scan); pre-release gating (release-gate); or a deep performance investigation (performance-check) — item (10) below is a quick flag, not a profiling pass.

Review order: (1) correctness bugs — logic errors, off-by-one, wrong condition, unhandled null/undefined. (2) regressions — breaks behavior callers depend on? (3) breaking changes — endpoint removed/renamed, required field added → grep for callers. (4) missing validation — unvalidated input reaching DB or executing code. (5) auth gaps — unauthenticated access, missing ownership check, role bypass. (6) data loss — overwriting without backup, missing transaction, destructive op. (7) injection — SQL/NoSQL/XSS/command via unsanitized input. (8) race conditions — non-atomic check-then-act, missing lock. (9) missing tests — behavior changed with no test update. (10) performance — N+1, expensive loop, unnecessary full-table scan.

Severity: critical (ship blocker) | high (fix before merge) | medium | low — format: SEVERITY | file:line | issue | fix

Critical:[N] High:[N] Medium:[N] Low:[N]
BREAKING CHANGES: [N — list | none]
VERDICT: ship ✓ | fix critical first | needs work
ESCALATE: [security-guard | db-guard | none]

Keep looking

Skills are one crate of 627,801. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.