MCP Demo

What is MCP Demo?

MCP Demo is a demonstration project that uses Cloudflare Workers MCP to create custom AI tools. These tools can be used with MCP‑capable AI assistants such as Claude Desktop, Cursor, and others. It is intended for developers who want to extend the capabilities of AI assistants with their own serverless functions.

How to use MCP Demo?

Clone the repository, install dependencies with pnpm install, log in to Cloudflare via wrangler login, run npx workers-mcp setup, then deploy with pnpm run deploy. For Claude Desktop, run pnpx workers-mcp install:claude. For Cursor, add a JSON configuration that includes the path to the worker and the worker’s URL. Free‑tier users must remove the browser section from wrangler.jsonc before deploying.

Key features of MCP Demo

• Custom AI tools running on Cloudflare Workers
• Seamless integration with Claude, Cursor, and other MCP clients
• Extend AI capabilities through your own serverless functions
• Includes tools like getRandomNumber, generateImage, screenshotURL, and extractImagesFromURL

Use cases of MCP Demo

• Prototype and test custom MCP tools quickly
• Build custom AI‑powered workflows with a personal serverless backend
• Experiment with the Model Context Protocol and Cloudflare Workers
• Integrate generative image creation into AI assistant interactions

FAQ from MCP Demo

What are the prerequisites for using MCP Demo?

You need Node.js v16 or later, the pnpm package manager, a Cloudflare account, the Wrangler CLI, and an MCP‑compatible client such as Claude Desktop or Cursor.

Does MCP Demo require a paid Cloudflare plan?

No, but free‑tier users must remove the browser binding from wrangler.jsonc. Doing so disables screenshotURL and extractImagesFromURL; all other tools remain functional.

How can I add my own tools?

Edit src/index.ts and add a new method to the Worker class. Then redeploy your worker with pnpm run deploy. Remember to restart your AI assistant after changing method names or parameters.

What should I do if my AI assistant cannot see the tools?

Restart the assistant. If the problem persists, check the Cloudflare Workers logs for deployment errors. Free‑tier users should verify they have removed the browser configuration.

Where does the project get its data or run?

The entire toolset runs on Cloudflare Workers. No data is stored locally; each tool execution is a stateless request to the deployed worker.