discovered 03 Aug 2026
crucible
→ View on GitHubCrucible is a security testing tool designed specifically for AI agents, enabling comprehensive behavioral integrity testing and automated red-teaming against a wide range of attacks, including those aligned with OWASP guidelines. It offers rapid deployment via CI/CD integration, producing detailed compliance reports, and incorporates a unique Model Context Protocol security module. The tool encompasses over 90 tested attack vectors, ensuring agents are hardened against potential threats before production deployment.