> For the complete documentation index, see [llms.txt](https://missiveapp.com/docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://missiveapp.com/docs/ai/mcp/server/connect.md). # Connecting an AI app Add Missive as a custom connector in Claude, ChatGPT, or any MCP client, then approve the permissions it asks for. Connecting takes two steps: add the Missive server in your AI app, then approve the request in Missive. **Server URL:** `https://mcp.missiveapp.com` That's the only value you need. Missive issues no API key and no client secret for MCP. Authorization happens in your browser. {% hint style="info" %} Before you start, make sure an admin has [enabled MCP access](/docs/ai/mcp/server/admin.md) in every organization you belong to. Otherwise the approval screen will refuse the request and name the organizations blocking it. {% endhint %} ## Step 1: Add Missive in your AI app Every MCP client words this differently, and their menus move around. Look for **Connectors**, **Custom connectors**, **MCP servers**, or **Integrations**, then add a remote server and paste the URL above. {% tabs %} {% tab title="Claude" %} 1. Open **Settings** > **Connectors**. 2. Click **Browse,** then search for **Missive.** 3. Click on `Missive` 4. Click **Connect to Claude**. Claude opens a browser window on Missive's authorization screen. Continue to [step 2](#step-2-approve-the-request-in-missive). {% endtab %} {% tab title="ChatGPT" %} 1. Open **Settings** > **Apps & Connectors**. 2. Add a custom connector, then paste `https://mcp.missiveapp.com` as the server URL. 3. Confirm, and ChatGPT sends you to Missive's authorization screen. Custom MCP connectors are not available on every ChatGPT plan, and OpenAI changes where this lives fairly often. If you can't find it, check OpenAI's current documentation. {% endtab %} {% tab title="Other clients" %} Point the client at `https://mcp.missiveapp.com` as a **remote** MCP server over HTTP. Missive registers your client automatically the first time it connects, so there is nothing to create in Missive beforehand. The client must support OAuth with PKCE and must send the `resource` parameter defined in the MCP specification. Most current clients do. Older ones fail to connect, and there is no workaround. See the [FAQ](/docs/ai/mcp/server/faq.md). Missive's MCP server is stateless and doesn't support SSE connections. {% endtab %} {% endtabs %} ## Step 2: Approve the request in Missive Your browser lands on Missive's **Authorize access** screen. If you're logged out, you'll be asked to log in first. You cannot create a new Missive account during this flow, so sign up separately if you don't have one yet. The screen shows the app requesting access and the permissions it asked for. {% stepper %} {% step %} ### Choose the permissions to grant Every requested permission is checked by default. Uncheck anything you'd rather not give away. Unchecking is a real restriction, not a preference. A permission you don't grant removes the matching capabilities from the app entirely, and the app can't ask for them later without going through this screen again. See [Permissions and scopes](/docs/ai/mcp/server/permissions.md) for what each one covers. Grant at least one permission, or **Approve** stays disabled. {% endstep %} {% step %} ### Verify the redirect URL Under **Verify redirect URL**, Missive shows the exact address it will send you back to after approval, and asks you to check **I recognize and trust this URL**. Read it. This is what stops a malicious site from talking you into handing over access to your email. If the address doesn't belong to the app you're connecting, click **Deny access**. {% endstep %} {% step %} ### Approve Click **Approve**. Missive sends you back to the app, and a notice tells you that you can close the window. If nothing seems to happen, check that your browser isn't blocking pop-ups. {% endstep %} {% endstepper %} The app is now connected and appears in **Settings** > **Login & Security** > **Connected apps**. {% hint style="warning" %} Only connect apps you trust. Granting **Send messages** lets an app deliver email as you, and granting **Manage conversations** lets it reorganize your inbox. Grant the least you need. {% endhint %} ## Changing permissions later There's no way to edit an app's permissions after the fact. To change them, [revoke the app](/docs/ai/mcp/server/connected-apps.md#revoking-an-app) and connect it again with a different selection. Reconnecting an app you've already authorized replaces the old access. You won't end up with two live grants for the same app. ## Related * [Permissions and scopes](/docs/ai/mcp/server/permissions.md) * [Managing connected apps](/docs/ai/mcp/server/connected-apps.md) * [MCP server FAQ](/docs/ai/mcp/server/faq.md) --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://missiveapp.com/docs/ai/mcp/server/connect.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.