AgentMail
Give an AI agent its own email inbox: create inboxes, read threads and send mail to anyone.
Search a Front inbox, draft replies and comment on conversations from an AI agent.
Opens Agent Studio, where connecting is one click. The connector URL below works in any MCP client.
Paste it into any MCP client. Setup docs
The Front MCP server connects a Front shared inbox to Claude and any MCP-compatible agent. Its 23 tools search and read conversations across every channel, draft and send replies, post internal comments, and tag, assign and re-status conversations — acting as you, within your Front permissions.
Connection verified
Connection checked by Agentman on .
Listed in Anthropic’s Connectors Directory at community tier — a listing, not a partner-level review.
Anthropic states this reflects the level of review a connector received, not a security audit.
Tool names from Anthropic's directory listing. This server requires sign-in, so we could not read tool descriptions or parameter schemas.
send_message reaches customers on the conversation's real channel. The draft tools are the safer path for agent-prepared replies.No, it is in open beta. Front states that everyone has access but that bug fixes, improvements and other details are still being worked on, and asks for feedback through its community group. Treat the tool set on this page as a snapshot: Front says more capabilities are in the works.
Yes. send_message sends on the conversation's native channel, which Front documents as covering email, SMS, chat and custom channels. Drafting tools exist separately, so an agent can prepare a reply without sending it. Front frames every action as taken on your behalf, with your permissions.
In two layers. A Front admin sets which read, write and send permissions the MCP server may use at all, which defines the superset. Each connected teammate is then restricted to what their own Front permissions allow, so enabling every inbox organization-wide still only gives each person the inboxes they can already open.
No. Front documents that the assistant acts as you, with your permissions, and describes the connection as a key that only works while you are authorized and only does what your role already allows. Because each teammate signs in to their own Front account, access tracks their existing inbox visibility.
Front names Claude, ChatGPT, Gemini and Microsoft Copilot. There are two connection routes: the official connector in your assistant's directory, where permissions are controlled by your company's assistant administrator, or a custom connection available now in any MCP-compatible client, where a Front admin configures the developer app.
Front lists six capabilities: search and read conversations across every channel, draft and send replies on the native channel, read and post internal comments, tag and assign and change conversation status, search and update contacts, and see teammates, inboxes and teams. Front suggests asking your assistant what tools are available once connected.
https://mcp.frontapp.com/mcp — HTTP 401 with a www-authenticate Bearer challenge, and RFC 9728 metadata at https://mcp.frontapp.com/.well-known/oauth-protected-resource naming https://app.frontapp.com as the authorization server, with an empty scopes_supported array (2026-09-13) · retrieved 2026-09-13Connect Front once and your agents call these tools on their own: on a schedule, in a workflow, with nobody at the keyboard.