Query Your Marketing Cloud Intelligence Data with the MCP Server

How the Marketing Cloud Intelligence MCP Server Works

The MCP server connects your MCI account to a compatible AI client, such as Claude Code or Cursor. When you ask a question, the client sends the request to the server. The server accesses only the data allowed by your configured permissions and returns the results in a readable format.

Available MCP Tools

Parameters for mci_query

Supported Date Ranges

Security

The MCI MCP server uses your configured MCI credentials and permissions to access data. Your account has access only to its authorized workspaces and data. The MCI MCP server sends all requests over HTTPS.

Example Conversation with the MCP Server

Show me last week's top campaigns by ROAS.

The AI client returns the top campaigns by ROAS for the requested period.

Compare these to last month.

The AI client compares the results with the previous month and summarizes the changes.

Set Up Marketing Cloud Intelligence MCP Server

Set up the MCI MCP server to query marketing data from Claude Code, Cursor, or another MCP-compatible AI client.

Before you begin, make sure that you have:

• A Marketing Cloud Intelligence account with API access.
• An MCP-compatible AI client.

1. Download your API credentials.

   1. Log in to Marketing Cloud Intelligence.
   2. On the navigation bar, open the dropdown list, and then select your name to access your profile settings.
   3. Next to API Access Token, select Generate/Refresh.
   4. Save the downloaded JSON credential file in a secure location, such as ~/.mci/credentials.json.

2. Configure your AI client.

   1. Add the MCI MCP server to your AI client's MCP server configuration.

      For Claude Code, the configuration file is typically located here:

      • MacOS and Linux ~/.config/claude/mcp_servers.json
      • Windows %APPDATA%\claude\mcp_servers.json
   2. Identify the host URL for your MCI instance.

      MCI Instance	Host URL
      EU	https://app-eu.datorama.com/api/mcp
      EU2	https://app-eu2.datorama.com/api/mcp
      US	https://app.datorama.com/api/mcp
      US2	https://app-us2.datorama.com/api/mcp

   3. Add this configuration. This example uses the US host URL.

      {
        "mci": {
          "command": "npx",
          "args": ["-y", "@datorama/mci-mcp-sdk"],
          "env": {
            "PRIVATE_KEY_PATH": "/absolute/path/to/credentials.json",
            "HOST": "https://app.datorama.com/api/mcp"
          }
        }
      }

   4. For PRIVATE_KEY_PATH, replace /absolute/path/to/credentials.json with the absolute path to your JSON credential file. For HOST, replace the example URL with the host URL for your MCI instance.
   5. Restart your AI client to apply the configuration.

3. Test your setup. In your AI client, ask:

   Show me my account and the workspaces I have access to.

   If the response includes your account and accessible workspaces, the setup is complete. Otherwise, verify the credential file path, regional endpoint, and account permissions, and then try again.