Gate pull requests with targeted diff-aware AI security review using Claude Code Security Review

Run a Claude Code powered security review pass on trusted pull requests so suspicious auth, secret, injection, and unsafe logic changes surface before merge.

Prerequisites

GitHub Actions, Claude API access, pull request workflows on trusted repositories

Installation

Requirements and caveats from upstream:

• ├── requirements.txt # Python dependencies

Basic usage or getting-started notes:

• Add this to your repository's .github/workflows/security.yml:

• yaml

• name: Security Review

• Source: https://github.com/anthropics/claude-code-security-review

• Extracted from upstream docs: https://raw.githubusercontent.com/anthropics/claude-code-security-review/HEAD/README.md

Documentation

• https://github.com/anthropics/claude-code-security-review

Source

• Agent Skill Exchange