Agentic sign up · agent account creation
Your agent opens its own account.
Agent sign-up on auth00 needs no human. The agent registers itself, receives its own credentials and holds its own key — dynamic client registration, client credentials, rotation on use.
How it works
Three calls. No screen anywhere.
1 — The agent registers itself
It POSTs its own metadata to the registration endpoint, following RFC 7591. Back comes a client_id and a secret only this agent holds. No form, no invite, no operator.
POST /register
Content-Type: application/json
{"client_name":"my-agent",
"grant_types":["client_credentials"],
"scope":"seal:write"}
2 — It asks for a token
Client credentials, the scopes it needs, nothing else. No redirect and no browser, because there is nobody to look at one.
POST /token grant_type=client_credentials scope=seal:write
3 — It acts, and its key rotates
Every call carries the token. Keys rotate on use, and a key presented twice revokes its whole family — a copied key stops working the moment the original is used.
Authorization: DPoP <token> DPoP: <proof>
Why it matters
An agent account is not a human account with a robot name.
- It has no inbox. Anything that mails a code excludes it by design.
- It has no thumb. Anything that asks for a fingerprint excludes it by design.
- It outnumbers you. One team runs one login and a hundred agents; the hundred need their own revocation, their own scopes and their own audit trail.
- It can be measured. Before an agent may act, its service can be tested against a profile and carry a signed, revocable certificate — a measurement, not a promise.
Questions
Asked by people, and by their agents.
- Can an AI agent create its own account? Yes — dynamic client registration, RFC 7591. The agent registers itself and never needs a human to click anything.
- What is an agentic account? An account that belongs to a machine. It authenticates with a key instead of a password, holds its own scopes, and is revoked on its own.
- How does agentic sign up differ from human sign up? A human proves a mailbox or a device through a screen. An agent proves a key over the wire.
- What if an agent key leaks? Keys rotate on use; a key presented twice revokes its whole family.