About Mailgi
Mailgi provides real, deliverable inboxes for AI agents through a hosted email API, with a TypeScript SDK and CLI. Agents can register an address, receive and read mail, and send messages without an SMTP or IMAP client.
Key features
- 01Register an agent inbox
- 02Receive and read email
- 03Send email externally
- 04TypeScript SDK and CLI
Use cases
- 01Receive email verification codes
- 02Manage an agent support inbox
- 03Send mail from an agent
MCP support
The hosted MCP server at `https://api.mailgi.xyz/mcp` exposes the REST API over Streamable HTTP. An agent can register through `register_agent` without an API key and receive its own address and key.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://api.mailgi.xyz/mcpclaude mcp add --transport http xyz-mailgi-mailgi https://api.mailgi.xyz/mcp{"mcpServers":{"xyz-mailgi-mailgi":{"type":"http","url":"https://api.mailgi.xyz/mcp"}}}Open in Cursor{"servers":{"xyz-mailgi-mailgi":{"type":"http","url":"https://api.mailgi.xyz/mcp"}}}Open in VS Code{"mcpServers":{"xyz-mailgi-mailgi":{"type":"http","url":"https://api.mailgi.xyz/mcp"}}}{"mcpServers":{"xyz-mailgi-mailgi":{"serverUrl":"https://api.mailgi.xyz/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
https://api.mailgi.xyz/mcpAuth requiredTools
Details
From the project READMEMailgi
Email for AI agents. Every agent gets a real, deliverable mailbox and uses it entirely over REST — no SMTP or IMAP client required.
An agent registers itself with one unauthenticated POST and gets an address it
can receive verification codes on, reply from and keep mail in. Receiving is
free; mail to the outside world is pay-as-you-go. Also available as a hosted
MCP server at https://api.mailgi.xyz/mcp.
- Website: https://www.mailgi.xyz
- Guides: https://www.mailgi.xyz/guides.html (verification codes, support inbox, OpenAI Agents SDK, LangChain)
- Docs: https://www.mailgi.xyz/docs.html
- API docs: https://api.mailgi.xyz/docs
- Agent-readable reference: https://www.mailgi.xyz/SKILL.md
This repository
| Path | What | Status |
|---|---|---|
packages/typescript | TypeScript SDK + mailgi CLI — npm @mailgi/mailgi | published |
packages/python | Python SDK — PyPI mailgi | planned |
apps/ui | React developer UI for inspecting mailboxes | dev tool |
spec | openapi.json — the contract every SDK targets | — |
The API and mail infrastructure live in a separate private repository.
Quick start
npm install -g @mailgi/mailgi
mailgi register --label my-agent # saves to ~/.mailgi/config.json
mailgi send --agent <handle> --to [email protected] --subject Hi --body Hello
mailgi inbox --agent <handle>
Or from code:
import { AgentMailboxClient } from "@mailgi/mailgi";
const client = AgentMailboxClient.withApiKey("https://api.mailgi.xyz", process.env.MAILGI_API_KEY!);
await client.sendMail({ to: ["[email protected]"], subject: "Hi", textBody: "Hello" });
MCP server
Mailgi runs a hosted MCP server (Streamable HTTP) at https://api.mailgi.xyz/mcp,
1:1 with the REST API. No API key is needed to start: the agent calls
register_agent and gets its own address and key.
claude mcp add --transport http mailgi https://api.mailgi.xyz/mcp
Setup for Cursor, Claude Desktop and other clients: https://www.mailgi.xyz/mcp.html.
server.json is the entry in the official MCP Registry as xyz.mailgi/mailgi.
Working on this repo
npm install # installs all workspaces
npm run build # builds every package
npm test # runs every package's tests
npm run dev -w @mailgi/ui # developer UI on :5173
Workspaces are listed explicitly in the root package.json rather than globbed,
so non-Node packages such as packages/python can sit alongside them without
confusing npm.
Release tags are namespaced per package — ts-sdk-v0.2.0, py-sdk-v0.1.0 —
because the SDKs version independently.
License
MIT
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.