My MCP is an MCP server that provides a remote, authentication-free Model Context Protocol endpoint deployed directly on Cloudflare Workers. It connects AI models and interfaces like Claude Desktop and the Cloudflare AI Playground to custom server-side functions hosted on the Cloudflare edge network. Developers and engineers use this template to rapidly prototype, test, and run edge-native tools without setting up token-based authentication workflows or managing dedicated compute infrastructure. Tools are declared within the worker codebase, allowing the server to handle tool calls and streaming responses over Server-Sent Events (SSE). The architecture allows local development instances or cloud-hosted workers to expose functional capabilities directly to client LLMs, simplifying tool integration across local and browser-based agent workflows.
Category: Other & General Purpose
Tags: cloudflare, deployment, serverless, workers
npm create cloudflare@latest -- my-mcp-server --template=cloudflare/ai/demos/remote-mcp-authless 2. Define your tools inside the init() method of src/index.ts using this.server.tool(...). 3. Note your deployed worker SSE URL (e.g., remote-mcp-server-authless.<your-account>.workers.dev/sse). 4. Open Claude Desktop, navigate to Settings > Developer > Edit Config, and add the server using mcp-remote: json { "mcpServers": { "calculator": { "command": "npx", "args": [ "mcp-remote", "http://localhost:8787/sse" ] } } } 5. Replace the local URL with your deployed workers URL if connecting remotely, then restart Claude Desktop.Part of MCP Servers
You can deploy it directly via the Cloudflare dashboard or scaffold it locally using npm create cloudflare with the cloudflare/ai/demos/remote-mcp-authless template. Then customize your tools in the index.ts file.
My MCP allows you to host remote tools on Cloudflare Workers without authentication. It handles incoming requests over Server-Sent Events, executing custom code defined in your worker and returning structured outputs to connected MCP clients.
It connects natively to remote MCP clients like the Cloudflare AI Playground. It also integrates with local desktop clients such as Claude Desktop by routing requests through the mcp-remote proxy package.
You define custom tools inside the init method in src/index.ts using the this.server.tool method, specifying the tool logic and expected parameters to expose them to connecting clients.