security-scan

To scan Claude Code configurations for security vulnerabilities, run a security audit on your .claude directory to parse configuration files and detect hardcoded secrets, prompt injection patterns, and overly permissive permissions using AgentShield.

Common security risks in Claude Code settings.json and CLAUDE.md files include hardcoded API keys, overly permissive tool access, command injection vulnerabilities in hooks, and malicious prompt injection patterns.

Yes, you can run a pre-commit security check on your .claude directory configurations to catch hardcoded API keys, overly permissive tool access, or command injection vulnerabilities before pushing changes to your project.

Yes, the security scan supports MCP server configs and agent definitions, comprehensively checking all core Claude Code configuration files including CLAUDE.md, settings.json, hooks, and agent definitions for misconfigurations.

The best way to audit Claude Code hooks for command injection vulnerabilities is to run a comprehensive config scan that parses hook definitions and categorizes any detected threats by severity from Critical to Info.

Detected misconfigurations and injection risks are prioritized by categorizing findings into severity levels ranging from Critical to Info, helping you address the most urgent security vulnerabilities first.