How to integrate data into Gemini with Windsor MCP

This guide covers the Gemini CLI, Google’s command-line agent. With Windsor MCP, the Gemini CLI can query data from 350+ sources, so you can ask questions about your business data in plain language without uploading files.

❗ Important: Since 18 June 2026, the Gemini CLI works only with a Gemini Code Assist Standard or Enterprise licence, through Google Cloud, or with a paid Gemini API key. Free, Google AI Pro and Google AI Ultra accounts moved to Google’s Antigravity CLI. If you use one of those accounts, follow the Windsor MCP setup guide for Antigravity instead. To add Windsor.ai to Gemini Enterprise as a data store, follow the Gemini Enterprise guide.

To use Windsor MCP with the Gemini CLI, make sure you have the following:

• An active Windsor.ai account (free or paid plan) with at least one connected data source
• Access to the Gemini CLI through a Gemini Code Assist Standard or Enterprise licence, Google Cloud, or a paid Gemini API key
• Node.js version 20 or higher (check with node -v)

🚀 Get started with Windsor MCP on the free forever plan, no credit card needed: https://onboard.windsor.ai/

1. Install Gemini CLI (command line interface).

npm install -g @google/gemini-cli

gemini

cd ~/.gemini

nano settings.json

{
 "mcpServers": {
  "windsor": {
   "httpUrl": "https://mcp.windsor.ai/"
  }
 }
}

gemini

/mcp auth windsor

6. Check the connection with the command:

/mcp list

The windsor server appears with its tools listed. Then ask a first question, for example “Which data sources are connected to my Windsor.ai account?”. Gemini answers with your connectors and the accounts connected under each one.

❗ Important: Windsor MCP for Gemini reads your data for analysis. On platforms that support write actions, Gemini can also make changes, such as pausing a campaign or changing a budget, when you ask it to. See the list of write actions by platform.

• Requests fail after you sign in with a personal Google account. Since 18 June 2026, the Gemini CLI does not serve free, Google AI Pro or Google AI Ultra accounts. Sign in with a Gemini Code Assist licence, a Google Cloud project or a paid Gemini API key, or set up Windsor MCP in Antigravity.
• The install fails or the gemini command does not start. Run node -v. The Gemini CLI needs Node.js 20 or higher.
• Gemini CLI shows “MCP issues detected”. Run /mcp list. If windsor shows as DISCONNECTED, open settings.json and check that the file is valid JSON and that httpUrl is exactly https://mcp.windsor.ai/. Then run /mcp reload.
• The browser does not open when you run /mcp auth windsor. The sign-in needs a browser on the same machine as the Gemini CLI. It does not work over SSH or inside a container without a browser. Run the Gemini CLI on your own computer and run /mcp auth windsor again.

• About Windsor MCP: general information about Windsor MCP, including supported data sources and LLMs, benefits and use cases.
• Windsor MCP use cases & examples (30+ prompts inside): practical examples and prompts to use with Gemini when you are not sure what to ask.
• How to integrate data into Antigravity with Windsor MCP: the setup for free, Google AI Pro and Google AI Ultra accounts.
• How to integrate data into Gemini Enterprise with Windsor.ai: add Windsor.ai as a data store in Gemini Enterprise.