Guardrails for AI coding agents — block catastrophic tool calls before they run. Semantic, agent-neutral, near-zero false positives.
By chatting or signing in you agree to the Terms and chat-message logging (revocable in History).