Subact ID documentation

Subact ID is an agent identity and delegation control plane. It issues short-lived, scoped tokens that let an AI agent act on behalf of a specific human. It records every decision in an audit ledger sealed by signed checkpoints, so each action traces back to that human.

Subact ID uses standard OAuth: RFC 8693 for token exchange, RFC 7662 for introspection, RFC 7009 for revocation and RFC 8707 for resource indicators. Agents authenticate with private_key_jwt: they sign a short-lived assertion with their own key. There is no shared client secret.

Start here

What Subact ID is, and what it is notWhat it does, what it does not do, and what it needs you to run.

QuickstartThe whole system on one machine, then the same steps by hand.

Your first exchangeThe request, the checks, the response and every claim in the token.

The five invariantsThe rules the server keeps, and the tests that check them.

Understand it

Delegation, not impersonationWhy the subject is the human and the agent is the actor.

Tasks and task grantsA long task, short tokens, and the task grant that renews them.

ScopeThe intersection of three sets, which only ever narrows.

Delegation depthWhat a chain of agents is, and what v0.1 issues.

RevocationLocal validation or per-call introspection, chosen per audience.

The audit ledgerWhat the signed checkpoints prove, and what they do not.

Build with it

Register an agent

Build an agent

Protect a tool server

Protect an MCP server

Operate it

SDK reference

Trust it

Threat modelWhat Subact ID defends against, and what is out of scope for v0.1.

The conformance suiteThirteen adversarial tests run against a live instance through its public endpoints.

Reporting a vulnerabilityWhere to send it and what to expect back.