Intercom MCP

How to Install and Use Intercom MCP

If you want your AI tools to connect securely with your Intercom data, you can use the Intercom MCP (Model Context Protocol). This guide will walk you through how to set it up and start using it step by step. You'll find simple instructions and real examples from the official documentation.

What is Intercom MCP?

Before setting it up, know that Intercom MCP is a secure protocol. It lets AI applications like customer service bots access your Intercom conversations, contacts, and tools safely and easily. It works by connecting to a remote MCP server that Intercom hosts, so you don’t need to handle complex data access yourself.

Step 1: Choose Your Authentication Method

To connect to the Intercom MCP server, you need to authenticate, or prove who you are. There are two ways:

1. OAuth Flow (Recommended): This is a browser-based method that handles signing in automatically.
2. Bearer Token: This uses an API token that you get from Intercom and put directly into your connection.

Choose whichever is easiest for you. The OAuth method is simpler for most people.

Step 2: Set Up the Connection

Now, you can configure the MCP server connection. Here are examples for both authentication methods you can use in your setup.

Using OAuth Authentication

This example uses npx to run the MCP remote tool connecting to Intercom’s MCP server:

{
  "mcpServers": {
    "intercom": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://mcp.intercom.com/mcp"
      ]
    }
  }
}

This will ask you to log in when you connect, making it easy and secure.

Using Bearer Token Authentication

If you prefer using a token, replace YOUR_INTERCOM_API_TOKEN with your actual token from Intercom, and use this configuration:

{
  "mcpServers": {
    "intercom": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://mcp.intercom.com/mcp",
        "--header",
        "Authorization:Bearer YOUR_INTERCOM_API_TOKEN"
      ],
      "env": {
        "AUTH_HEADER": "Bearer YOUR_INTERCOM_API_TOKEN"
      }
    }
  }
}

Make sure your token has the right scopes for reading users/companies and conversations.

Step 3: Start Using MCP Tools

Once connected, you can use one of the 6 available tools Intercom MCP provides to interact with your data. For example, to search conversations with open state and limit the results to 50:

object_type:conversations state:open limit:50

If there are more results than 50, you’ll receive a response with a “starting_after” cursor. Use this cursor in your next query to get the next page of results:

object_type:conversations state:open limit:50 starting_after:cursor_abc123

Keep doing this until there are no more pages.

Step 4: Test Your Connection (Optional but Helpful)

Before full use, test if your setup works using the MCP Inspector tool with this command:

npx @modelcontextprotocol/inspector

Connect to the Intercom MCP server using:

• Transport Type: Streamable HTTP
• URL: https://mcp.intercom.com/mcp

This will help ensure everything is working correctly.

Final Tips

• Always restart your AI tools after changing the setup.
• For token method, double-check your token permissions.
• Use the official documentation for your AI provider for detailed integration steps.

With these simple steps, you’ll have Intercom MCP installed and ready to securely link your AI tools with your Intercom workspace easily.