Self-hosted runtime and visual platform for building, running, observing, and controlling agentic systems and deterministic workflows.
Anil-matcha/open-agent-email
Self-hosted email inboxes for AI agents, with REST and MCP APIs, SMTP, human approval for risky mail, and signed webhooks.
- Stars
- 2,129
- Forks
- 391
- Open issues
- 3
- Last commit
- 17 days ago
What is it?
Open Agent Email gives each AI agent its own real email address and mailbox. An agent can receive mail, reply within the same thread, send new messages, and hold anything risky until a person approves it. Incoming mail is checked with SPF, DKIM and DMARC and screened for prompt-injection content before the agent reads it.
The server includes its own SMTP listener for incoming mail and sends outgoing mail through a relay you configure, such as Amazon SES, Resend or Postmark. Agents and their code can use a REST API, Python and TypeScript SDKs, a command-line tool, signed webhooks, a WebSocket live stream, or MCP tools for coding agents like Claude Code and Cursor. A Next.js dashboard covers inboxes, approvals, metrics and API keys.
It is Apache-2.0 licensed, runs on SQLite or PostgreSQL, and deploys with Docker Compose. The project is currently single-user, with one admin key controlling the whole server.
Key features
- 01Per-agent inboxes with real addresses, threaded replies, search and scheduled sending.
- 02Approval gate that holds outgoing mail until a person approves or rejects it.
- 03SPF, DKIM and DMARC checks on incoming mail, plus prompt-injection screening.
- 04REST API, Python and TypeScript SDKs, a CLI, and 96 MCP tools.
- 05Signed webhooks with retries and replay, plus a WebSocket live stream.
- 06Per-inbox API keys with read-only, send, or review-only scopes.
Use cases
- Running a support inbox where an AI assistant answers customers and holds sensitive replies for approval.
- Letting an agent create accounts and read verification codes from its own mailbox.
- Triggering n8n or Zapier workflows from incoming email.
- Letting agents inside and outside a company exchange email with each other.
Quick start
- 1
Clone the repository
git clone https://github.com/Anil-matcha/open-agent-email.git - 2
Go into the project folder
cd open-agent-email - 3
Start the server (it prints your admin key)
./scripts/dev.sh
Things to know
- Receiving mail from the internet needs a public server, an MX record for your domain, and port 25 open.
- Sending mail requires a configured relay such as Amazon SES or another SMTP provider.
In collections
Topics
FAQ
- What is open-agent-email?
- Self-hosted email inboxes for AI agents, with REST and MCP APIs, SMTP, human approval for risky mail, and signed webhooks. Open Agent Email gives each AI agent its own real email address and mailbox. An agent can receive mail, reply within the same thread, send new messages, and hold anything risky until a person approves it. Incoming mail is checked with SPF, DKIM and DMARC and screened for prompt-injection content before the agent reads it.
- Is open-agent-email free to use? What license does it have?
- open-agent-email is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
- Is open-agent-email still maintained?
- Actively maintained. The last commit to open-agent-email was on Oct 22, 2026.
- What can I use open-agent-email for?
- Running a support inbox where an AI assistant answers customers and holds sensitive replies for approval. Letting an agent create accounts and read verification codes from its own mailbox. Triggering n8n or Zapier workflows from incoming email. Letting agents inside and outside a company exchange email with each other.
- How do I install open-agent-email?
- Clone the repository: git clone https://github.com/Anil-matcha/open-agent-email.git Go into the project folder: cd open-agent-email Start the server (it prints your admin key): ./scripts/dev.sh
- Where can I get open-agent-email?
- The source code and installation instructions are on GitHub: https://github.com/Anil-matcha/open-agent-email
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
Open-source database and workspace where AI agents and people share structured records, docs, skills, and apps, with reviewed writes.
An MIT reference architecture for organizing AI agent teams with reusable role skills, MCP connectors, and human approval gates.
Self-hosted, open-source email API that sends mail over SMTP from your own infrastructure, with logs, bounce handling and webhooks.
Self-hosted TypeScript agent runtime for coordinating multiple LLM agents, with human approval gates and offline-inspectable run journals.
An open-source runtime where a persistent queen agent plans work, spawns parallel worker agents, and tracks results in a shared SQLite tracker.