"use_case": "For AI agents"
An inbox your agent can create and read
Agents that sign up for tools, verify accounts or wait on a confirmation need email. Add the MCP server and your agent mints its own address, waits for the message and gets the code back as JSON.
Create your account
Email
We’ll email you a 6-digit code
Verification code
6-digit code
Sign up for Acme with a new inbox and verify the email.
create_inboxreceiver
wait_for_emailreceiver
847291
Who it's for
- Teams building agents that complete sign-ups and onboarding on other services
- Browser and computer-use agents that hit an email verification step
- QA agents that explore your product and need to receive its email
- Anyone wiring an agent framework that speaks MCP
The job
An agent mid-task hits “check your email for a code”. It needs an address nobody else is using, a way to wait without burning turns, and the code without reading HTML.
- 01Create an inbox. create_inbox returns a fresh address and a since timestamp.
- 02Use it. The agent types the address into the sign-up form like any user would.
- 03Wait for the email. wait_for_email holds until the message lands, up to five minutes.
- 04Read the code. With extract: "otp" the code comes back on its own; links come extracted too.
Set it up
One command for Claude Code, a few lines of JSON for any other MCP client. A project API key is the only setting. Prefer a shell? The CLI does the same from a script.
claude mcp add receiver \
-e RECEIVER_API_KEY=recv_live_... \
-- npx -y @receiverdotemail/mcpOne project per agent
A project is the isolation boundary. Give each agent or deployment its own, with its own key, so one agent never reads another’s mail.
Keys that can’t do much
An API key reads its own project’s mail and mints addresses. It cannot create keys, manage webhooks or delete the project.
Receive only
Addresses receive mail; they never send it. Your agent can’t be used to spam anyone from a Receiver.email address.