agent-safe-pipeline

What people use it for

Prevents an AI agent from approving its own actions by submitting immutable hashed intents to an independent decision-policy layer. Risky actions can require approval tied to the exact intent before credentials are used.

Github Awesome