The project is a platform to safely operate AI agents in production. It auto-discovers agents and endpoints, enforces policy-as-code before and after generation, and provides deep observability with searchable audit trails. The system detects prompt-injection, jailbreaks, PII leaks, and hallucination risk; maps controls to frameworks (EU AI Act, NIST AI RMF); and supports canarying, rollback, and human-in-the-loop approvals. Secure RAG features check document provenance and citation confidence, while latency/cost SLOs and drift monitors keep performance and spend on track.