Workers MCP Demo

🚀 Cloudflare Workers MCP Demo

This is a simple demonstration project for creating custom AI tools using Cloudflare Workers MCP. These tools can be used with AI assistants such as Claude, Cursor, and other MCP - supported AI assistants.

🚀 Quick Start

What is MCP?

MCP (Model Context Protocol) is an open - source standard by Anthropic for connecting AI assistants to data sources, including content repositories, business tools, and development environments. Its goal is to help cutting - edge models generate better and more relevant responses.

This project leverages Cloudflare Workers MCP to create custom tools that can be accessed and used by AI assistants, thereby extending the capabilities of AI.

✨ Features

• Custom AI tools running on Cloudflare Workers.
• Seamless integration with Claude, Cursor, and other MCP clients.
• Ability to extend AI capabilities through your own serverless functions.

📦 Installation

Prerequisites

• Node.js (v16 or higher).
• pnpm package manager.
• [Cloudflare account](https://dash.cloudflare.com/sign - up).
• [Wrangler CLI](https://developers.cloudflare.com/workers/wrangler/install - and - update/).
• Claude Desktop, Cursor, or other MCP - supported clients.

Setup Steps

1. Clone this repository

git clone https://github.com/akazwz/workers-mcp-demo.git
cd workers-mcp-demo

2. Install dependencies

pnpm install

3. Configure your Cloudflare Worker

Ensure you are logged into Cloudflare via Wrangler:

pnpx wrangler login

Then run the MCP setup command:

npx workers-mcp setup

This will automatically configure your Worker to support MCP.

Note for free users

If you are a free user (non - paying) of Cloudflare Workers, you need to modify the wrangler.jsonc file by deleting or commenting out the following configuration:

"browser": {
  "binding": "BROWSER"
}

This is because the Browser Rendering API is only available to paying users. After removing this configuration, the screenshotURL and extractImagesFromURL functions will not work, but other functions will still function properly.

4. Deploy your Worker

pnpm run deploy

5. Configure your MCP client

For Claude Desktop:

Run the following command to install your MCP server in Claude:

pnpx workers-mcp install:claude

For Cursor:

Add the following content to Cursor's MCP configuration:

{
  "type": "command",
  "command": "/path/to/workers-mcp run workers-mcp-demo https://your-worker-url.workers.dev /path/to/workers-mcp-demo"
}

💻 Usage Examples

Basic Usage

After configuration, you can directly use your custom MCP tools from the AI assistant. For example, in Claude Desktop, you can use the following command:

@workers-mcp-demo getRandomNumber

Available Tools

This demo includes the following tools:

• getRandomNumber: Generate a random number.
• generateImage: Create an image based on a text prompt.
• screenshotURL: Take a screenshot of a specified URL (requires a paid account).
• extractImagesFromURL: Extract images from a web page (requires a paid account).

Customization

To add your own tools, edit the src/index.ts file and add new methods to the Worker class. After making changes, redeploy your Worker using the following command:

pnpm run deploy

📚 Documentation

Troubleshooting

• If your AI assistant cannot see your tools, try restarting the assistant.
• If you change method names or parameters, you need to restart the AI assistant for the changes to take effect.
• Check the Cloudflare Workers logs for any errors during deployment.
• If you encounter the Error: Browser Rendering API is only available for Workers Paid plans error as a free user, refer to the note for free users above.

📄 License

This project is licensed under the MIT License.

Acknowledgments

• Cloudflare Workers MCP for providing the underlying framework.
• Anthropic Model Context Protocol for the open - source MCP standard.
• Cloudflare Workers for providing the serverless platform.