一种可重复使用的AI代码安全审查技能，源自anthropics/claude代码安全审查。

原始项目是一个GitHub Action，使用Claude代码对拉取请求进行安全审计。该技能提取并重组其核心提示、过滤规则和方法论，形成一个独立且工具无关的格式，适用于GitHub Copilot、Claude Code或任何AI辅助代码审查工作流程。

结构

许可证

麻省理工学院——详情请参见原始仓库。

Code Security Review Skill

A reusable AI code security review skill derived from anthropics/claude-code-security-review.

The original project is a GitHub Action that uses Claude Code to perform security audits on Pull Requests. This skill extracts and reorganizes its core prompts, filtering rules, and methodology into a standalone, tool-agnostic format suitable for use in GitHub Copilot, Claude Code, or any AI-assisted code review workflow.

Structure

claude-code-security-skills/
├── README.md                                # This file
├── SKILL.md                                 # Skill entry point and usage guide
└── resources/
    ├── audit-prompt.md                      # Security audit prompt template
    ├── filtering-rules.md                   # False positive filtering rules (19 exclusions + 17 precedents)
    ├── hard-exclusion-patterns.md           # Regex-based auto-exclusion patterns
    └── customization-guide.md              # Guide to extend scan/filter rules

License

MIT — see the original repository for details.