Agent Guardrail

Action-level governance for AI agents with policy enforcement, spend caps, and flight recording.

Provides action-level governance for AI agents, controlling what agents do rather than what they say. Includes a policy engine with allowlists and denylists, daily and total spend caps, emergency kill switches, a flight recorder for full session replay, and approval gates for risky actions. Framework-agnostic with LangChain and CrewAI integrations.