home / mcp / remote mcp server authless

Remote MCP Server Authless

Provides a remote MCP server hosted on Cloudflare Workers, exposing tools via SSE for remote clients to use.

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "aninibread-remote-mcp-server-authless": {
      "url": "https://remote-mcp-server-authless.your-account.workers.dev/sse"
    }
  }
}

You can deploy a remote MCP server that runs on Cloudflare Workers and exposes a simple SSE endpoint for clients to connect. This lets you add and run your own MCP tools remotely, without hosting on your own infrastructure, and connect from playgrounds or desktop clients to perform tasks through the server.

How to use

Once your remote MCP server is up, you can connect to it from an MCP client like the Cloudflare AI Playground or a local client configured to reach the server’s SSE endpoint. In practice, you deploy the server, then point your client at the server URL, and start issuing tool-enabled tasks through the browser or your desktop client. You can customize the server to expose your own tools by editing the server code and redeploying.

How to install

Prerequisites you need before installing: Node.js and npm installed on your machine. You will also need access to Cloudflare for deploying Workers if you choose the Cloudflare route.

Step 1: Install the MCP server starter using the Cloudflare template.

npm create cloudflare@latest -- my-mcp-server --template=cloudflare/ai/demos/remote-mcp-authless

Step 2: Follow the prompts to set up your project configuration and deploy. You will receive a URL similar to: remote-mcp-server-authless.<your-account>.workers.dev/sse

Step 3: If you prefer running locally, you can create the server with the CLI shown above and run it in your local environment, then expose it to the internet or a custom domain as needed.

Additional notes

Customization is done inside the server’s initialization. To add your own tools, modify the init() method in src/index.ts and call this.server.tool(...) for each new tool. After adding tools, redeploy so clients can access the new capabilities.

Connecting to the Cloudflare AI Playground is straightforward: open the Playground, enter your deployed MCP server URL, and you can start using your MCP tools directly from the playground.

You can also connect the server to a local or remote MCP client, such as Claude Desktop, by configuring the client to point to the server’s SSE endpoint. See the example configuration snippet below for reference.

{
  "mcpServers": {
    "calculator": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "http://localhost:8787/sse"  // or remote-mcp-server-authless.your-account.workers.dev/sse
      ]
    }
  }
}