How to connect Windsor MCP to any MCP-compatible AI client

Windsor MCP lets you connect your Windsor-integrated business data to MCP-compatible AI clients, so you can query your marketing, sales, analytics, CRM, and e-commerce data in natural language.

Windsor.ai offers native integrations for Claude, Copilot, and ChatGPT. However, you can also connect Windsor MCP to many other MCP-compatible agents and tools, including Cursor, Perplexity, Manus, OpenClaw, Claude Code, Cline, Continue, Roo Code, Goose, and more.

The setup pattern is nearly identical across all LLMs.

Before setup, make sure you have:

• An active Windsor.ai account
• At least one connected data source in Windsor.ai
• An MCP-compatible AI client
• Your Windsor API key (optional)

🚀 Start using Windsor MCP today with a free 30-day trial: https://onboard.windsor.ai/ and unlock smarter, AI-powered analytics.

Go to https://onboard.windsor.ai/ and connect the data sources you want to analyze in your AI tool.

Most MCP-compatible clients have a menu similar to:

• Settings
• Connectors/MCP Servers/Integrations
• Add MCP Server/Add Connector
• Import via JSON

The exact menu names vary by platform, but the setup process is usually the same.

Method 1: Enter the server name and URL

Add a new MCP server manually by entering:

• Server name: Windsor.ai
• URL: https://mcp.windsor.ai/

Method 2: Paste the configuration JSON

Paste the following configuration into your MCP client and save it:

{
  "mcpServers": {
   "windsor": {
    "url": "https://mcp.windsor.ai/",
    "headers": {
     "Authorization": "Bearer YOUR_WINDSOR_API_KEY"
   }
  }
 }
}

Replace YOUR_WINDSOR_API_KEY with your actual Windsor API key (you can find it in your Windsor.ai settings in the API Access section).

After saving the configuration, open a new chat in your AI client and try a simple prompt such as:

What data sources are connected to my Windsor.ai account?

You can also try:

• Show me the total spend for the last 30 days
• Which ad platforms are connected to my Windsor account?
• Compare ROAS across my connected channels

If the connection works, you can move on to deeper questions about spend, revenue, ROAS, conversions, pipeline quality, attribution, and cross-channel performance.

• Exact setup steps vary slightly by the AI client
• Windsor MCP uses read-only access for analysis
• Your original data is never modified in the connected platforms
• If your client supports native remote MCP servers, the direct URL setup (https://mcp.windsor.ai/) is usually enough
• Some clients may require restarting the app after adding a new MCP server

You can use Windsor MCP with clients such as:

• Cursor
• Perplexity
• Manus
• OpenClaw
• Claude Code
• Cline
• Continue
• Roo Code
• Goose
• Windsurf
• Open WebUI
• LibreChat
• Other MCP-compatible tools and agents

• About Windsor MCP – Explore general info about Windsor MCP: supported data sources and LLMs, benefits, use cases, etc.
• Windsor MCP use cases & examples (30+ prompts inside) – If you don’t know what to ask or are looking for advanced Windsor MCP use cases, read this guide for a batch of practical examples and useful prompts to make your ChatGPT data analysis even more effective and deeper.
• AI Prompt Library by Windsor.ai – Discover a list of advanced, ready-to-use prompts focused on different data sources and use cases.