codex-code-review

SKILL.md

To perform thorough, automated code reviews using the OpenAI Codex CLI agent, use this skill. Codex runs locally and can analyze code changes, identify issues, suggest improvements, and provide security/performance insights through non-interactive automation.

⚠️ CRITICAL: When reviewing code that involves dependency versions, latest releases, or current best practices, you MUST use the WebSearch tool to verify information before making any claims. Never assume version numbers or release status - always search first to avoid false positives. See the "Web Search Verification" section for details.

When running codex for automated code reviews, you MUST use the --full-auto flag to grant all necessary permissions for headless operation. Without this flag, codex may hang waiting for user approval.

使用 OpenAI Codex CLI 執行全面的代碼審查。當用戶請求代碼審查、想要分析差異/PR、需要安全審計、性能分析或想要自動代碼質量反饋時，應該使用此技能。支持查看分階段更改、特定文件、整個目錄或 git diff。 來源：tyrchen/claude-skills。

查看原文