AgentMail MCP server

What the model can do

The inboxes your AI agents send and receive from: list inboxes, browse and search threads and messages across them, read a thread or message in full with attachment text; can also send a message, reply in a thread, forward one on, save a draft, and label a thread or messages — mark them read, move them to trash, or tag them — when you ask.

Reads

• agentmail_inboxesThe agent inboxes the key can see: address, display name, status (active or paused) and pod.
• agentmail_threadsThreads newest first, across every inbox or in one (id or address), filtered by label, sender, recipient or subject substring and a time window.
• agentmail_searchFull-text search ranked by relevance: words in the subject, sender, recipients or body.
• agentmail_threadOne thread with its messages in order: who wrote what and when, each body as text (quoted replies stripped), attachment names, labels.
• agentmail_messageOne message in full (text body up to 6000 characters, headers, attachments).

Acts, when you ask

These change something in AgentMail. The model is told to use them only when you ask for exactly that, and each one is granted by name: per team on a shared connection, by the owner on a personal one.

• agentmail_composeSends an email from one of the agent inboxes — new, a reply, or a message forwarded on — or saves it there as a draft when draft is true.
• agentmail_labelAdds and removes labels on a thread or on particular messages, which is how AgentMail does triage — it has no read flag and no archive folder.

Connect it

First connect AgentMail on Dashboard → Connections, then add Memtro to your client once. The AgentMail tools appear alongside the tools of everything else you have connected; nothing is configured per connector.

claude mcp add --transport http memtro https://platform.memtro.com/api/mcp \
  --header "Authorization: Bearer sk-memtro-..."

{
  "mcpServers": {
    "memtro": {
      "url": "https://platform.memtro.com/api/mcp",
      "headers": {
        "Authorization": "Bearer sk-memtro-..."
      }
    }
  }
}

{
  "mcp": {
    "memtro": {
      "type": "remote",
      "url": "https://platform.memtro.com/api/mcp",
      "headers": {
        "Authorization": "Bearer sk-memtro-..."
      }
    }
  }
}

Who may use it

The key identifies the person, and the server offers that person the tools they may use everywhere else in Memtro. A shared AgentMail connection is used by the teams an administrator granted it to, tool by tool, on the Permissions page; a personal connection is used by its owner alone, and its tools that act stay off until the owner allows them. See Teams and permissions and Connectors.

Ask it like this

• › What did the support agent's inbox receive this morning?
• › Find the thread where the supplier confirmed the delivery date
• › Read the invoice attached to the latest email from Acme
• › Which threads in sales@ are still unanswered?

Questions

Is this an official AgentMail MCP server?

No. It is Memtro's MCP server with Memtro's AgentMail connector behind it: the connector reads AgentMail through its API with the credentials you connect on Dashboard → Connections, and AgentMail publishes nothing here. The same endpoint carries every other system you connect and Memtro's memory tools.

What does it cost?

Nothing on top of Memtro. One price per organisation by its number of users, from $50 a month for up to 10 users after a 7-day free trial, with every connector, the MCP server, jobs and memory included. You bring your own model keys.

Can it change anything in AgentMail?

Only when you ask, and only what it has been granted: send a message from an agent inbox, reply in a thread (to the sender or everyone), forward a message on, or save either as a draft; label a thread or particular messages: mark them read, move them to trash or back out of it, or add and remove tags of your own. Every other tool reads, searches or lists. An administrator grants each acting tool by name per team, and a personal connection acts only after its owner allows it.

Which clients work?

Any MCP client that speaks Streamable HTTP: Claude Code, Cursor, Claude Desktop, OpenCode, Codex, Zed, Windsurf, Cline and the Anthropic and OpenAI SDKs. There is no local process to run; the server is remote and authenticated with your key.

Other MCP servers · Email, calendar and chat

• Gmail MCP server
• Google Calendar MCP server
• Slack MCP server
• Outlook Mail MCP server
• Outlook Calendar MCP server
• Microsoft Teams MCP server