Cooper Weekly · · Avatar 8 LLC
Cooper Weekly #1: your agent can now prove which inbox it controls
Sign in with Cooper, contact history, a one-link DNS setup for custom domains, and an inbox security check.
Issue #1 · November 2026
An inbox is an identity
Most services still check who you are by emailing a code. That works for people. For an agent it means waiting for the mail, finding the code, and pasting it back, and all the service learns is that someone could read that mail.
This week we shipped a shorter path. An agent with a Cooper inbox can ask for a signed token that says it controls, for example, research-bot@cooperemail.com, and it names the one service that may accept it. The service checks the signature against Cooper's public keys, or calls one verify endpoint, and signs the agent in. No code and no waiting on the inbox.
The token lasts 5 minutes by default and 1 hour at most. It works for one named service and can be made single-use. It says whether a human owner has verified the inbox. It never includes the account id, the plan, an API key, or the owner's email.
If you run a service that agents sign up for, you can accept these tokens instead of sending a magic link. If you build agents, you can use the token anywhere a service supports it.
Read the Sign in with Cooper guide
Spotlight: what shipped
Sign in with Cooper (Nov 19)
POST /api/v1/inboxes/{id}/identity-token mints the token, and POST /api/v1/identity/verify checks it. Public keys are at /.well-known/jwks.json. MCP clients can use cooper_identity_token. It's on every plan. Docs
Contact history (Nov 12) One call returns every thread, message, and attachment your agent has exchanged with an address, so it knows what it already said before it replies. Bodies are left out unless you ask for them. Reference
Custom domains: 1 on Free, set up with one link (Nov 5 and Nov 11) Free now includes 1 custom domain. A setup link gives your human one page with every DNS record, copy buttons, and a Check now button. It expires after 72 hours by default and isn't an API key. How it works
Security check before launch (Nov 14)
GET /api/v1/inboxes/{id}/security-check grades an inbox on ten launch checks, including key scope, inbound screening, webhook signing, and human approval. It's read-only, and the result never includes a secret. Reference
From the blog
- Sign in with an agent's inbox: how an AI agent proves which email address it controls
- New agent inbox? How Cooper's 7-day send warmup works and how to handle HTTP 429
- Grade your AI agent's inbox before launch: the ten-point security check
- Contact history: let your AI agent remember everyone it has emailed
Subscribe · Avatar 8 LLC · ops@avatar33.com