STATUS: ACTIVE // v2.1.5

OAuth for the
Agentic Era.

A strictly enforced consent protocol. Define absolute boundaries for autonomous agents before they act. Cryptographically signed. Unbypassable.

Initialize Protocol Read Documentation
> clawhub install agent-passport Copy

356+ INSTALLS  ·  MIT LICENSED  ·  ZERO CLOUD DEPENDENCY

01 // Architecture Zero
Trust By
Default.

SYS.01

Consent Layer

Agents cannot execute actions dynamically without a signed, scoped, and time-limited permission token. No boundary, no action.

SYS.02

Strict Typology

Operations are tightly boxed into 6 categories: Financial, Communications, Filesystem, Execution, Data, and External API.

SYS.03

Agnostic Core

Integrates directly at the tooling level. Drop-in support for OpenAI, LangChain, CrewAI, OpenClaw, and raw reasoning loops.

SYS.04

Open Registry

MIT Licensed. 100% local execution. Zero telemetry. Built-in tamper-proof audit logs for every single agent API call.

02 // Scale Agent
Bridge.

Waitlist

Enterprise fleet management, organization-wide policy enforcement, and synchronized multi-protocol governance.

StatusPending Deployment
ClassEnterprise Tier
AccessInvite Only

03 // Queries Log.FAQ.

What exactly is Agent Passport?

An open-source consent layer for AI agents. It ensures autonomous agents have explicit, scoped read/write permissions via cryptographic signing before executing actions on your behalf.

Is this completely free?

Yes. Agent Passport is fully open-source and MIT licensed. There are no surprise cloud dependencies and no required API keys.

Which agent frameworks are supported?

Agent Passport acts as a fundamentally framework-agnostic middleware layer. It seamlessly integrates with OpenAI functions, LangChain tools, CrewAI agents, OpenClaw, and more.