security-scan

To scan Claude project configuration files for security vulnerabilities, apply this skill to your codebase to detect hard-coded keys, unsafe permissions, and injection vectors using AgentShield during setup, changes, or CI pipeline execution.

Security issues in Claude settings and agent definitions include hard-coded secrets, unsafe permissions, and injection points, which risk breaches and execution attacks if left misconfigured.

You can run a Claude config security scan in a CI/CD pipeline via GitHub Action to enforce security gates before merges, ensuring no new vulnerabilities are introduced to the repository.

To fix hard-coded keys found in Claude configuration files, use the skill's automatic remediation options to easily correct problems by replacing hard-coded keys with environment variables.

You should scan a Claude repository for misconfigured settings and hooks before committing changes to ensure no new vulnerabilities are introduced, or during initial setup and CI pipeline execution.