secureagentics/Adrian: Real-Time Runtime Security Monitoring That Acts Before the Agent Does

GitHub·low signal

Adrian is a Python runtime security layer that monitors AI agents to catch malicious tool use, prompt injection and policy drift in real time — before the agent acts — with native LangChain and LangGraph integration. Created 2026-05-11, it has reached ~264 stars. The 'pre-execution interception' framing, shared with permit0 and Sponsio, reflects a clear 2026 shift from after-the-fact agent auditing toward deterministic, blocking enforcement.