security-scan

To scan Claude Code configurations for security vulnerabilities, use AgentShield by running npx ecc-agentshield scan or installing it globally via npm install -g ecc-agentshield to audit your project files.

Security risks in CLAUDE.md and MCP configurations include prompt-injection vulnerabilities, misconfigurations, and excessive permissions. Auditing these files enforces least privilege and deny lists to prevent unauthorized access.

Yes, you need AgentShield installed to audit Claude Code settings.json files. You must run npx ecc-agentshield scan or npm install -g ecc-agentshield to detect security exposures and validate configurations.

You should run a security scan on Claude Code hooks and agent definitions when starting new projects, during ongoing maintenance, and before committing changes to surface potential security exposures and receive remediation guidance.

The best way to fix prompt-injection risks in Claude Code configurations is to scan CLAUDE.md with AgentShield, which surfaces vulnerabilities and provides clear remediation guidance to secure your setup.