See what your AI agents can cause, route them safely, and prove what happened. Open-source execution authority for AI agents — consequence preview, trust routing, signed receipts.