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.
InsAIts Monitor AI agent communications for anomalies local, privacy-first · HackerTrans
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.