ever-works/orgs/code-review-standards
Claude Code and agent skill by ever-works/orgs
- community
Checking that the SKILL.md file is still in the repository.
Use when reviewing a code change, or preparing your own change for review, so every diff meets the same bar for correctness, safety, and maintainability regardless of who reviews it.
Install
npx vskill@latest install ever-works/orgs/code-review-standardsSKILL.md
Code Review Standards
A review is a verdict on a claim: "this change does X safely." Work through the passes below in order; stop and request changes as soon as a pass fails hard.
Pass 1 — Understand the claim
- Read the task or slice description and its acceptance criteria first, not the diff.
- Read the author's handoff note: what was built, what was excluded, where to look.
- If you cannot state in one sentence what the change claims to do, stop and ask —
reviewing an unclear claim produces an unclear verdict.
Pass 2 — Correctness
- Trace the main path end to end; confirm the code does what the claim says.
- Hunt the edges: empty inputs, nulls, zero and negative numbers, duplicates,
concurrent calls, and the largest realistic payload.
- Check every error path: what does the caller see on failure? No swallowed
exceptions, no bare 500s, no error branch that "cannot happen".
- Verify state changes are atomic or recoverable — no half-written records if the
second step fails.
Pass 3 — Safety
- Every new input is validated at the boundary: type, length, format, range.
- Every object ID in a request is checked for ownership or permission — the caller
being authenticated is not the caller being authorized.
- Queries are parameterized; anything templated into a command or query from
user input is a finding.
- No secrets, tokens, or personal data in code, logs, or test fixtures.
Pass 4 — Maintainability
- Tests exist for the behavior claimed, and at least one failure path each.
A test that cannot fail is a decoration.
- Names say what things are; no dead code or commented-out alternatives left in.
- Migrations, new config, and new environment variables are documented in the
change itself.
Verdict
- Approve, approve-with-nits (author fixes, no re-review), or request-changes.
- Every request-changes item states the problem, why it matters, and an acceptable
fix. "I would have done it differently" is not a finding.
- Quality bar: you are willing to be on call for this change after approving it.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingcompanies/atlas-engineering/skills/code-review-standards/SKILL.md
Popularity
- 7-day trend score
- 5
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/ever-works/orgs/code-review-standards)