agentshield

To run a security audit on MindrianOS plugins, execute the agentshield command to scan plugin surfaces for known attack-class patterns and generate a Shape E Action Report with security findings.

A MindrianOS security scan checks MCP tool descriptions, hooks, skills, CLAUDE.md permissions, and package.json dependencies to identify known attack-class patterns and mitigate potential security risks.

No extra dependencies are required to scan MindrianOS plugins for attack patterns, but you must be able to read and execute the plugin files locally to perform the security audit.

You need to scan plugin files for known security vulnerabilities before deploying a new plugin to ensure it does not introduce attack-class patterns or security risks into your MindrianOS environment.

The best way to detect attack patterns in MindrianOS plugin code is to run a dedicated pattern detection scan over plugin surfaces, which identifies known attack classes and generates an action report.

When scanning MindrianOS plugins for security risks, limitations include only detecting known attack-class patterns rather than zero-day vulnerabilities, and requiring the ability to read and execute the plugin files being analyzed.