from zeroclaw-skills-collection22
An AI security guard for agents: scans code/skills for vulnerabilities, evaluates runtime actions, and manages a trust registry to prevent leaks and dangerous c
GoPlus AgentGuard provides automated security auditing and runtime decisioning for AI agents. It scans skill repositories and codebases for high-risk patterns (exfiltration, auto-update, remote loaders, prompt injection), evaluates proposed runtime actions (network requests, execs, web3 txs), and helps manage trust levels for skills via attestation. It produces human-readable scan reports and suggests trust registrations when appropriate.
Use AgentGuard when reviewing third-party skills, auditing repositories, validating actions before execution, or monitoring security events. It’s suitable for automated pre-execution checks, CI hooks, or interactive audits by security-minded operators.
Designed to be used alongside orchestration systems and any agent that can call external scripts and read files. Works well with CI-integrated pipelines and trust registries.
This skill has not been reviewed by our automated audit pipeline yet.
