A simple API that turns risky agent actions into an auditable workflow: propose → approve/deny → verify. Built for production teams that need control, compliance, and a tamper-evident trail.
LIVE Propose actions, approve with a signed receipt, verify authenticity in one call.
Gate high-severity actions before anything executes. Your agents propose, humans (or policies) decide.
Every decision produces a signed receipt you can verify later. Cryptographic proof of what was approved.
Clear trail of what was proposed, who approved, and when. Compliance teams love this.
Integrate in minutes with simple HTTP endpoints. Works with any agent framework.
Your agent submits an intended action with context and severity level.
A human (or policy engine) approves or denies based on severity and risk.
Receive a signed receipt and cryptographically verify authenticity anytime.
# 1. Propose an action
POST /v1/actions/propose
→ {action_id: "abc123", status: "pending", severity: "high"}
# 2. Approve (or deny)
POST /v1/actions/abc123/approve
→ {decision: "APPROVED", receipt: {signature: "04dcfa..."}}
# 3. Verify the receipt
POST /v1/verify
→ {valid: true, message: "Signature valid. Receipt is authentic."}
No—Policy Gate is a control layer that works with any framework. LangChain, CrewAI, custom agents—all compatible.
Yes—human approvals, policy rules, or both. Automation features on roadmap.
Hit /docs or run the 3-call demo above. You'll have a verified receipt in 30 seconds.
Start with the live API, then add gating to the actions you can't afford to get wrong.