create
CLI to set up and deploy MCP Servers to Cloudflare Workers in seconds. Just write TypeScript functions to make Cursor MCP tools.
claude mcp add --transport stdio fefergrgrgrg-create-mcp bun create mcp
How to use
This MCP server is generated by the create-mcp CLI and deploys a Cloudflare Workers-based MCP server. It enables you to define MCP tools by adding TypeScript methods to the MyWorker class in src/index.ts, with JSDoc comments that describe parameters and return values. Each method becomes an MCP tool that can be called by your Cursor Agent. After scaffolding, the CLI deploys a Hello World MCP server to your Cloudflare account and wires it into Claude Desktop for easy access. To use it, install the prerequisites, scaffold a new server with bun create mcp, customize the tools in MyWorker, redeploy with bun run deploy, and reload Cursor to start invoking your new tools.
How to install
Prerequisites
- Install Bun: https://bun.sh/
- Cloudflare Wrangler CLI installed and logged in: https://developers.cloudflare.com/workers/wrangler/install-and-update/
- Claude Desktop App installed (required by the workflow, though this may be removed in the future)
Installation steps
- Install Bun (if you haven't already)
- macOS/Linux: curl -fsSL https://bun.sh/install | bash
- Windows (PowerShell): iwr https://bun.sh/install | iex
- Log in to Cloudflare via Wrangler
- wrangler login
- Scaffold a new MCP server
- bun create mcp
- You can pass a name directly: bun create mcp --name <server-name>
- Navigate into the generated directory and install dependencies (if not done by the CLI)
- cd <server-name>
- npm install (or yarn/pnpm) if your project uses them
- Deploy the server to Cloudflare
- bun run deploy
- Open Claude Desktop and connect your new MCP server (the CLI will copy the MCP server command to your clipboard for easy pasting into Cursor)
Additional notes
Tips and notes:
- The CLI scaffolds a Cloudflare Workers MCP server; your tools live as TypeScript methods inside MyWorker in src/index.ts.
- Each method's JSDoc @param and @return annotate tool inputs/outputs, which are used to generate MCP tool metadata.
- After deploying, you can redeploy anytime with bun run deploy to update the worker.
- Ensure you have Wrangler logged in to the correct Cloudflare account before deploying.
- If you encounter issues, check the generated README and the generated src/index.ts for tool descriptions and typings.
- If Cloudflare access or Claude Desktop integration changes, refer to the CLI output and docs for updated steps.
Related MCP Servers
gemini -tool
MCP server that enables AI assistants to interact with Google Gemini CLI, leveraging Gemini's massive token window for large file analysis and codebase understanding
Alph
Universal MCP Server Configuration Manager
obsidian
MCP server for Obsidian vault management - enables Claude and other AI assistants to read, write, search, and organize your notes
driflyte
The Driflyte MCP Server exposes tools that allow AI assistants to query and retrieve topic-specific knowledge from recursively crawled and indexed web pages.
attio
Model Context Protocol server for Attio CRM - enables Claude Desktop and Cursor to interact with your Attio workspace
GameMaker
GameMaker MCP server for Cursor - Build GM projects with AI