Creator here. Built this after watching multi-agent systems fail silently in production.

The core problem: when AI agents talk to each other, they develop patterns humans can't audit. Jargon drifts, context gets lost, hallucinations chain.

Technical approach: - Local embeddings via sentence-transformers (BAAI/bge-small-en-v1.5) - Heuristic + statistical anomaly detection - No cloud dependency for core features - LangChain/CrewAI integrations included

Free tier works without API key. Happy to answer questions.