Open Agent Email is open-source email infrastructure built specifically for AI agents. It gives every agent a real, secure inbox: agents can receive mail, reply inside the same thread, send new messages, and pause for a human to approve anything risky before it goes out. The project is self-hosted with Docker and PostgreSQL, exposes REST and MCP APIs plus SMTP/IMAP access, and ships Python and TypeScript SDKs. It has attracted around 2,100 GitHub stars and is under very active development, with code pushed in 2026.
Most agent frameworks treat email as an afterthought, bolting a shared Gmail account onto an autonomous workflow. Open Agent Email instead provides per-agent inboxes with signed webhooks, live event streams and human-in-the-loop approval workflows, so an agent can run an inbox-driven process (support triage, scheduling, lead follow-up) while a person stays in control of sensitive sends. Because it speaks MCP, coding agents such as Claude Code and Cursor can use it as a tool, and transactional sending is handled through Amazon SES.
Free. The project is open source under the Apache-2.0 license and self-hosted; you only pay for your own infrastructure and any Amazon SES sending costs.
Developers building autonomous agents that need a trustworthy email identity: agent-driven support desks, scheduling assistants, outreach workflows, and any multi-agent system where email actions need auditability and human oversight.
The project is part of a small family of open tools by the same author aimed at agent infrastructure, and its design reflects lessons from production agent deployments: every outbound message can be logged, threaded and audited, and the approval queue gives operators a single place to review what their agents are about to send. Setup guides and API documentation live in the repository’s docs folder, including a quickstart that takes you from Docker Compose to a working agent inbox.