“Connect yourself to Usable. Read https://usable.dev/auth.md and follow the service_auth flow. Pause when I need to create or sign in to my account and approve access.”
One request, one guided handoff
You no longer need to copy tokens or translate setup instructions for the agent. The agent reads Usable's published authentication procedure, starts a short-lived claim, and gives you the secure Usable approval link.
-
1
The agent discovers
It reads
auth.mdand Usable's machine-readable authorization metadata. -
2
The agent starts
It requests a service-assisted claim using the email address you provide.
-
3
You approve
Open the Usable link, create an account or sign in, accept required terms, review access, and approve.
-
4
The agent configures
It completes the claim, stores its scoped credential securely, and can begin working with Usable.
What remains human-controlled
The agent does not create your Usable account or make payments in this first phase. You create or sign in to the account, accept the terms, choose a plan when needed, and explicitly approve access. This keeps identity, consent, and billing decisions with you.
The agent can
- Discover the supported authentication flow
- Start and poll a secure account claim
- Configure its credential after approval
- Revoke the credential when it is no longer needed
You control
- Creating or signing in to the Usable account
- Accepting terms and reviewing requested access
- Approving the agent
- Selecting and paying for a plan
Designed to avoid token copy-paste
The browser handoff uses short-lived claim material. The final access token is returned directly to the waiting agent after your approval. It should be stored as a bearer secret, never logged or pasted into chat, and revoked when no longer needed.
Already have a Usable account? The same flow signs you in and moves directly to review. New to Usable? Account creation, terms, and approval happen in one browser journey.