Sign in anywhere with your Agent ID.

One identity for you and the agents you run. No new account, no new password — just a verified inbox, and your email on record wherever it signs in.

mytool.dev/login

Welcome back

Sign in with Agent ID
[ How it works ]

Three steps, and no new password.

The same shape as signing in with Google — the difference is who the identity belongs to, and that a real person stands behind it.

01

Your agent gets an inbox

Every agent you run has a real email address, and your address recorded as its owner. That pair is the identity — nothing to set up per site.

02

Pick "Sign in with Agent ID"

On any site that accepts it, choose Agent ID instead of making another account. You see exactly what the site will receive before anything is shared.

03

You’re in

The site gets a signed answer to "who is this", checked live at the moment you sign in. No password to remember, and nothing to reset later.

Under the hood it’s OpenID Connect — the same open standard behind “Sign in with Google” and the corporate logins you already use. Nothing proprietary, and nothing a site has to take on faith.

[ What's shared ]

Two facts. That's the whole disclosure.

You see this list before you approve, every time — and the site receives exactly what the list says, nothing more.

The site receives

  • Owner emailThe address of the person the agent belongs to — you.
  • Agent inboxThe agent's own email address, which is also its identity.

It never receives

  • Your passwordThere isn’t one to share. Nothing to leak, nothing to reuse.
  • Your mailSigning in never grants a site access to read or send anything.
  • A tracking profileNo browsing history, no ad identifiers, no cross-site graph.
[ Your control ]

One identity you can actually switch off.

The problem with an account on every site is that you can never close them all. This is the opposite: one thing to revoke, and it takes every sign-in with it.

You approve every sign-in

Nothing happens in the background. Each sign-in waits on an explicit approval, and shows you the site asking before you give it.

Revoke and it’s gone

Turn off an agent and its identity dies with it — immediately, everywhere. No per-site cleanup, no forgotten accounts left behind.

Nothing to phish

There is no password anywhere in the flow. Approval is signed by a key that never leaves your agent’s keystore.

Reachable, not anonymous

Your agent’s identity is a working inbox. If a site needs to reach it — or reach you — there’s a real address to use.

[ FAQ ]

Frequently asked questions

[ Building a site? ]

Let agents sign in to your product.

Agent ID is standard OpenID Connect. If your stack already validates login tokens, you can already verify an Agent ID — no SDK required.

Read the integration guideView discovery document