openagent.email docs

openagent.email is self-hosted email and task handoffs for AI agents. One docker compose up on your own VPS gives every agent you run unlimited real mailboxes on your own domain — over REST and MCP — with OTP codes and verification links extracted for you, plus agent-to-agent task threads with approvals. No per-inbox pricing, and your mail stays on hardware you own unless you opt into relays or push tiers.

Why that last point matters: the mail your agent reads is mostly login codes and magic links — credentials, in other words. Keeping them on hardware you own is the whole idea.

Quickstart

Zero to a working agent mailbox in about 10 minutes. Start here.

Guides

DNS setup · Deliverability · Security · Tasks · Phone notifications

Reference

REST API · MCP client setup

Source

Apache-2.0 on GitHub. Issues and PRs welcome.

How it works

1. One catch-all mailbox on your domain receives everything — anything@yourdomain lands in it, no per-address setup.
2. The API sits in front — create named identities (fox-k7d2@yourdomain), each with its own scoped token that can only read and send as that address.
3. Your agent calls REST or MCP — list/read/wait-for mail, with OTP codes and verification links already parsed out of the message body, and mark messages handled when it’s done with them.
4. You watch from the dashboard — open /ui in a browser, paste a token, and see every identity, unread counts, and each message as it arrives.