Authenticate AI agents with AuthPI

AI agents shouldn't borrow user accounts or share API keys. In AuthPI, agents are first-class identities (agt_) with their own credentials, scopes, organization memberships, and audit trail. Secret verifiers mint short-lived OAuth tokens through the standard client_credentials grant; suspending an agent cuts its access within one five-minute token TTL.

Start

• Give an AI agent an identity in 10 minutes — create an agent, mint a token, validate it in your API
• Agents & services — the full path — from the identity model and credential choice to the go-live checklist

Understand

• Agents — the concept model: lifecycle, verifiers, scopes, and how agents differ from M2M clients
• Multi-org tokens — how organization membership shapes what a token can do

Build

• Agent identities — model your fleet: one agent per workload, scoped and auditable

Reference

• Agents — Core API — create, update, and delete agents and their verifiers
• Token claims — every claim in an agent access token, including dat