Windsor MCP (Model Context Protocol) enables your LLM to query, explore, and analyze your full-stack business data integrated into Windsor.ai with zero SQL writing or custom scripting.
It connects seamlessly to 325+ platforms, giving AI-native tools such as Claude, Perplexity, Cursor, or others, real-time access to your performance marketing, sales, and customer data to help you unlock valuable insights.

Windsor MCP is a natural language interface that connects your integrated Windsor.ai datasets with the LLM platform, enabling you to better understand your data by asking questions like:

• “What campaigns had the best ROAS last month?”
• “Give me a breakdown of spend by channel over the past 90 days.”
• “What campaigns are wasting our advertising budget?”

All in real-time, directly inside your LLM chat interface.

Sync data from Facebook Ads, GA4, HubSpot, Salesforce, Shopify, TikTok Ads, and more via native Windsor.ai connectors.

Windsor MCP works via the Claude Desktop or with a lightweight dev proxy. No custom integrations required.

Built on Anthropic’s open MCP spec, it’s compatible with Claude, Perplexity, Cursor, and more.

Get instant breakdowns, summaries, and performance insights from your integrated data.

You connect Windsor MCP to your preferred LLM as an external connector using the MCP protocol. The LLM can then issue real-time data queries and receive structured results, all within the chat interface.

• What was total ad spend by channel last month?
• Break down ROAS for Meta vs Google Ads for Q2
• Are there any campaigns overspending vs target ROAS?

https://windsor.ai/introducing-windsor-mcp/

• Claude Pro or higher-tier Claude Desktop plan
• Your Windsor API key

1. Go to Claude settings → Connectors → Add custom connector
2. Use one of the following URLs for Windsor MCP:
   • https://mcp.windsor.ai
   • https://mcp.windsor.ai/sse
3. Open a new chat and start with:

My Windsor.ai API key is {your-key}. {Your question here}

4. Accept connector permissions and start querying your data!

For users on lower-tier Claude plans or requiring custom setups for advanced flexibility.

• Claude Desktop with dev mode enabled

1. Inatall mcp-proxy and copy its path.

uv tool install mcp-proxy
which mcp-proxy  # Copy full path

2. Configure Claude Desktop: Open Settings → Developer → Edit Config and add:

{
  "mcpServers": {
    "windsor": {
      "command": "/Users/{your-username}/.local/bin/mcp-proxy",
      "args": ["https://mcp.windsor.ai/sse"]
    }
  }
}

💡 Replace with your system username.

3. Fully quit and reopen Claude. You should now see “windsor” listed in your MCP options.

• Cursor Desktop installed
• Your Windsor API key

1. Install mcp-proxy

uv tool install mcp-proxy
which mcp-proxy  # Copy full path

2. Open settings in Cursor Desktop. Select Tool & Integrations > New MCP Server.

3. The mcp.json file will open. Paste the following script into it:

{
  "mcpServers": {
    "windsor": {
      "command": "/Users/{your-username}/.local/bin/mcp-proxy",
      "args": ["https://mcp.windsor.ai/sse"]
    }
  }
}

4. Windsor MCP will now become active in Cursor. It will ask for Windsor’s API Key in a prompt; just paste it, and you are good to go with any questions related to your data.

1. Install mcp-proxy

uv tool install mcp-proxy
which mcp-proxy  # Copy full path

2. Install Gemini CLI Use Node.js to globally install the Gemini CLI (make sure you have Node.js 18 or later installed).

npm install -g @google/gemini-cli

3. Configure Gemini to use Windsor MCP Navigate to the Gemini config directory:

cd ~/.gemini

If the .gemini directory doesn’t exist yet, run gemini once to generate it. Open the settings.json file:

nano settings.json

Add the following configuration inside the JSON object:

{
  "mcpServers": {
    "windsor": {
      "command": "/Users/{your-username}/.local/bin/mcp-proxy",
      "args": ["https://mcp.windsor.ai/sse"]
    }
  }
}

Note: Make sure the overall file remains valid JSON (no trailing commas or syntax errors).

4. Start Gemini with Windsor MCP Now, simply run Gemini:

gemini

You’ll be asked for your Windsor API key — paste it in to authenticate. You’re all set now!

Yes, it's available during our beta phase. You’ll need a Windsor.ai account with integrated data and API key access. But keep in mind that Claude Desktop allows you to add external connectors only on the paid plans.

Any AI agent compatible with MCP, including Claude Desktop, Perplexity, Cursor, and custom tools.

Marketing performance, sales pipelines, spend summaries, ROAS trends, campaign anomalies, and more. If it’s in your Windsor.ai data, you can ask it.

No. Just ask your questions in plain English and get structured responses in real-time.

Windsor MCP is currently in beta. All features are fully functional, but you may encounter occasional quirks. We're actively improving performance, authentication, compatibility, and feature coverage.

Start querying your business data via Windsor MCP.
👉 Get your API Key
👉 Watch the demo

For support or feedback, contact us at support@windsor.ai.