MCP

      How to integrate mcpfy.ai with Claude, OpenAI, Cursor & Windsurf

      Rick

      Are you tired of setting up local environments and managing Node.js just to connect AI tools? With mcpfy.ai, you can integrate mcpfy.ai hosted Model Context Protocol (MCP) servers that you can plug directly into your favorite AI platforms—no local setup needed.

      Below, we’ll walk you through how to use these remote MCP servers across four platforms: Claude, OpenAI (ChatGPT), Cursor, and Windsurf.

      MCP

      🔑 How to Retrieve Your Remote MCP URL from mcpfy.ai

      1. Login to your account at https://mcpfy.ai
      2. Create a new MCP server – name it appropriately (e.g., “filesystem”, “github-fetcher”)
      3. Add your tool (like file system access, web search, API client, etc.)
      4. Authenticate if the tool requires it (OAuth or API token)
      5. Copy the generated URL – typically in the format:
        https://mcp.mcpfy.ai/<your-token>/sse

      🧠 Claude (Anthropic): Integrate Remote MCP with Claude Desktop & API

      1. Ensure Claude Desktop is Installed & Updated
      2. Follow the URL retrieval steps above to get your endpoint and token
      3. Edit Claude’s Config File:
        Go to Settings → Developer → Edit Config
        Paste the following: 
         {
    "mcpServers": {
        "MCPfy-server": {
            "command": "npx",
            "args": ["-y", "mcp-remote", "https://mcp.mcpfy.ai//sse"]
        }
    }
}
      4. Restart Claude
      5. Use Tools – Claude will prompt you to approve the use of tools provided by the remote server

      🤖 OpenAI (ChatGPT): Connect to Remote MCP Servers

      For ChatGPT (Enterprise or Team)

      1. Go to Settings → Connectors → Add Connector
      2. Paste in the remote MCP URL from mcpfy.ai
      3. Once saved, ChatGPT will automatically call these tools when relevant

      💻 Cursor IDE: Plug & Play Remote MCP

      1. Open Cursor
      2. On the top right, click on the Settings button
      3. Navigate to the MCP section
      4. Click on Add a New Global MCP Server
      5. Paste the JSON configuration that you copied from mcpfy.ai:
      {
  "mcpServers": {
    "MCPfy-server": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://mcp.mcpfy.ai/<your-token>/sse"
      ]
    }
  }
}
      1. Save the configuration
      2. You’re ready to go!

      🌊 Windsurf (Cascade AI Agent): Add Remote MCP Plugins

      1. Login to Windsurf – authentication is required to manage plugins
      2. Open the Windsurf application
      3. Navigate to Windsurf Settings from the main menu
      4. Go to the Cascade section
      5. Click on Manage Plugins
      6. Click VIEW RAW CONFIG
      7. Paste the following JSON that you copied from mcpfy.ai:
      {
  "mcpServers": {
    "MCPfy-server": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://mcp.mcpfy.ai/<your-token>/sse"
      ]
    }
  }
}
      1. Save the configuration
      2. Refresh the tools list
      3. Use the tools directly within Windsurf as needed

      ✅ Conclusion: Hosted MCP is the Future

      By using mcpfy.ai, you can skip local Node.js installs and jump straight into productivity. These hosted MCP servers work seamlessly across:

      • Claude Desktop & API
      • ChatGPT (via Connectors)
      • Cursor IDE
      • Windsurf IDE (Cascade)

      No more fiddling with environments. Just plug the URL and go.

      🔗 Check out mcpfy.ai to get started.

      Related Posts