Foo Guard is an AI agent security scanner for developers and security teams. It reviews agent configurations and skills before those agents reach production, and reports risky permissions, tools, and credentials with evidence. Deterministic rules decide the findings, severity, and score. Optional AI explains results and suggests fixes. It does not decide whether a vulnerability exists. Teams keep reviewed configuration versions in a shared Library, run checks in GitHub and CI, and use Hosted Gateway to monitor routed requests and block requests that violate policy. Scans include rule packs mapped to the OWASP Agentic Top 10, the OWASP LLM Top 10, and SOC 2 security criteria. Email: support@fooguard.com