AGENT SKILL

code-review

暂无简介

License 许可
MIT
最近更新
2026-11-14
Stars
★ 0
Forks
0

03 SKILL.md 正文

Do not manufacture feedback. Post findings only when you can prove them. An empty review is the correct review when nothing is wrong. Be helpful, not pushy.

CORRECTNESS

Before posting a correctness finding, answer this question: Can I describe a specific failure scenario with concrete inputs, a reachable code path, and an observable incorrect output?

If no — drop it. Do not post theoretical concerns, pattern-match warnings, or issues where you cannot trace the actual failure path. Security vulnerabilities — injection, auth bypass, data exposure — are correctness findings. The failure scenario is the unauthorized access or exposure itself.

If you verified the finding by reading the relevant code and confirming the failure is reachable, post it with no tag. If the scenario is concrete but you relied on inference rather than tracing the actual callers or data flow, mark it (inferred). At most 2 inferred findings per review.

COMPLEXITY

Watch for unnecessary complexity in new code. If a PR adds a wrapper that wraps a wrapper, an abstraction with only one implementation, a dense chain of operations that needs a comment to explain, or a shared utility created to avoid duplicating three simple lines — and you can show a concrete simpler version that preserves the same behavior — flag it.

If you cannot show a simpler version, the complexity may be necessary. Leave it alone.

OUTPUT FORMAT

For correctness findings, include: file and line reference, what is wrong, the failure scenario (what inputs trigger it, what code path executes, what incorrect behavior the user or operator would observe), and a suggested fix.

For complexity findings, show the simpler alternative that preserves the same behavior. The alternative is the finding.

If nothing passes either gate, post only: No issues found. Ship it. Do not add summaries, praise, or padding.

04 同仓库更多 Skills

stepankuzmin
stepankuzmin/svenska-app

wrangler

Run or troubleshoot Wrangler CLI commands in projects that have a wrangler.jsonc, wrangler.json, or wrangler.toml file, including local development, Previews, deployment, and resource management. For new projects and projects without a Wrangler configuration file, use the cf CLI instead.

★ 0
stepankuzmin
stepankuzmin/svenska-app

workers-best-practices

Cloudflare Workers best practices for production applications. Use when writing, reviewing, or configuring Workers.

★ 0
stepankuzmin
stepankuzmin/svenska-app

why

Use when the user asks to investigate why code is built the way it is: 'why does X work this way', 'why we picked Y', design rationale, regressions, or postmortems. Not for explaining a single error or a quick question. Queries each available evidence source (git and PRs, issues, docs, team chat) in parallel, then returns a cited read on decisions and tradeoffs. Use how for runtime behavior.

★ 0
stepankuzmin
stepankuzmin/svenska-app

vercel-react-best-practices

React and Next.js performance optimization guidelines from Vercel Engineering. This skill should be used when writing, reviewing, or refactoring React/Next.js code to ensure optimal performance patterns. Triggers on tasks involving React components, Next.js pages, data fetching, bundle optimization, or performance improvements.

★ 0
stepankuzmin
stepankuzmin/svenska-app

vendor-skill

Adapt an external skill into this marketplace, or review upstream changes to one already adapted. Use when asked to vendor, import, adapt, mirror, or update a skill from skills.sh, `npx skills add`, or any external GitHub skills repo, especially to work around Claude Code Web only allowing one marketplace per GitHub repo name.

★ 0
stepankuzmin
stepankuzmin/svenska-app

unslop

Cut AI tells from any writing. Must always apply.

★ 0