e2a: Open-Source Email API for Applications and AI Agents

Introduction

e2a is an open-source email API for applications and AI agents, enabling transactional email sending and real two-way inboxes. Developed by Token Canopy, it allows agents to interact with email, incorporating human oversight when needed. Whether you're building a simple application or a complex AI agent, e2a provides the infrastructure to send and receive emails with structured data and enhanced security.

This project, written in Go, boasts a strong community with 191 stars and 25 forks, licensed under Apache-2.0. It's designed to be flexible, offering both a managed hosting service at e2a.dev and the option to self-host using Docker.

Why Use e2a and Its Advantages

e2a offers several compelling advantages over traditional email services, especially for AI agents and applications requiring advanced email interaction:

No Public URL Required for Inbound Mail: Agents can authenticate with their API key and consume inbound mail over a WebSocket, by polling the REST API, or through MCP tools. This is ideal for agents on developer laptops, edge devices, or behind corporate firewalls, eliminating the need for public webhook URLs.
Email Threading on Every Reply: e2a resolves the RFC Message-ID, In-Reply-To, and References graph within each agent mailbox, emitting correct reply headers and maintaining a server-owned topology identity for message reads. This ensures conversations remain coherent and easily trackable.
Slug Provisioning on a Shared Domain: Operators can set shared_domain and users can immediately register an agent on a shared domain with no DNS configuration, simplifying onboarding.
Built-in Human-in-the-Loop (HITL) Review: A per-agent protection policy can hold outbound mail in a pending_review state. Reviewers can approve via a dashboard, magic-link email, MCP tools, or the API, with configurable auto-expiration policies.
Inbound Content Screening: Opt-in content scanning flags prompt-injection payloads, jailbreaks, obfuscation, data-exfiltration patterns, and even phishing attempts. Messages are routed to allow, review, or block based on scan sensitivity, providing a critical layer of security for AI agents.
Authenticated Inbound Identity: Messages expose normalized SPF, DKIM, and DMARC evidence, with an explicit aligned DMARC verdict, ensuring the authenticity of incoming emails.

Installation

Self-host with Docker

To self-host e2a, you'll need Docker installed. Follow these steps:

git clone https://github.com/tokencanopy/e2a.git
cd e2a

docker compose up -d

This will bring up Postgres, the API server, and the dashboard.

curl http://localhost:8080/api/health

docker compose exec e2a e2a -config /etc/e2a/config.yaml -bootstrap-email you@example.com

Save the API key, as it's shown only once.

Connect your agent to hosted e2a

e2a provides quickstart guides for connecting various AI agent frameworks to its hosted service:

Claude Code: claude plugin marketplace add tokencanopy/e2a then claude plugin install e2a@e2a
Codex: codex plugin marketplace add tokencanopy/e2a then install e2a via /plugins

Cursor: Add the MCP server directly to .cursor/mcp.json:

{
  "mcpServers": {
    "e2a": { "url": "https://api.e2a.dev/mcp" }
  }
}

Examples

e2a is highly versatile and can be used to build a wide range of AI-powered email applications. Here are some use cases:

AI Support Agent: Triage customer requests, retrieve context, reply in-thread, and hold sensitive replies for approval.
AI Receptionist: Answer common inquiries, route messages, and forward conversations to a human team.
AI Scheduling Agent: Coordinate participants, propose times, and preserve state across multi-turn replies.
E-commerce Agent: Handle order questions, returns, and coordinate with vendors through persistent email threads.

For complete, runnable examples, explore the e2a-runbooks repository, which demonstrates various use cases with different agent frameworks like Mastra, OpenAI Agents SDK, Claude Agent SDK, LangChain, CrewAI, and Pydantic AI.