Fire in da houseTop Tip:Paying $100+ per month for Perplexity, MidJourney, Runway, ChatGPT and other tools is crazy - get all your AI tools in one site starting at $15 per month with Galaxy AI Fire in da houseCheck it out free

workers-mcp-demo

MCP.Pizza Chef: akazwz

workers-mcp-demo is a Cloudflare Workers-based MCP server that enables developers to create custom AI tools accessible by MCP clients like Claude and Cursor. It runs serverless functions on Cloudflare Workers, extending AI capabilities by connecting AI assistants to various data sources and tools. This server facilitates seamless integration with multiple MCP clients, providing a scalable, cloud-native environment for deploying AI-enhanced workflows and tools.

Use This MCP server To

Deploy custom AI tools on Cloudflare Workers for scalable serverless execution Integrate AI assistants with business tools via MCP protocol Extend AI capabilities using serverless functions in Cloudflare Workers Enable seamless interaction between AI models and custom data sources Create and host MCP servers accessible by Claude, Cursor, and other clients

README

MCP Demo

这是一个使用 Cloudflare Workers MCP 创建自定义 AI 工具的简单演示项目,这些工具可以与 Claude、Cursor 和其他支持 MCP 的 AI 助手一起使用。

什么是 MCP?

MCP (Model Context Protocol) 是由 Anthropic 开源的一个标准,用于连接 AI 助手与数据源,包括内容仓库、业务工具和开发环境。它的目标是帮助前沿模型生成更好、更相关的回应。

本项目利用 Cloudflare Workers MCP 创建自定义工具,这些工具可以被 AI 助手访问和使用,从而扩展 AI 的能力。

特点

  • 在 Cloudflare Workers 上运行的自定义 AI 工具
  • 与 Claude、Cursor 和其他 MCP 客户端的无缝集成
  • 能够通过自己的无服务器函数扩展 AI 能力

前提条件

设置步骤

1. 克隆此仓库

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

2. 安装依赖

pnpm install

3. 配置你的 Cloudflare Worker

确保你已经通过 Wrangler 登录到 Cloudflare:

pnpx wrangler login

然后运行 MCP 设置命令:

npx workers-mcp setup

这将自动配置你的 Worker 以支持 MCP。

免费用户配置注意事项

如果你是 Cloudflare Workers 的免费用户(非付费用户),需要修改 wrangler.jsonc 文件,删除或注释掉以下配置:

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

这是因为 Browser Rendering API 仅适用于付费用户。删除此配置后,screenshotURLextractImagesFromURL 功能将无法使用,但其他功能仍然可以正常工作。

4. 部署你的 Worker

pnpm run deploy

5. 配置你的 MCP 客户端

对于 Claude Desktop:

运行以下命令在 Claude 中安装你的 MCP 服务器:

pnpx workers-mcp install:claude
对于 Cursor:

在 Cursor 的 MCP 配置中添加以下内容:

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

使用方法

配置完成后,你可以直接从 AI 助手使用你的自定义 MCP 工具。例如,在 Claude Desktop 中,你可以使用如下命令:

@workers-mcp-demo getRandomNumber

可用工具

此演示包括以下工具:

  • getRandomNumber:生成一个随机数
  • generateImage:根据文本提示创建图像
  • screenshotURL:对指定 URL 进行截图(需要付费账户)
  • extractImagesFromURL:从网页中提取图像(需要付费账户)

自定义

要添加自己的工具,编辑 src/index.ts 文件并向 Worker 类添加新方法。更改后,使用以下命令重新部署你的 Worker:

pnpm run deploy

故障排除

  • 如果你的 AI 助手看不到你的工具,尝试重启助手。
  • 如果你更改了方法名称或参数,你需要重启 AI 助手才能使更改生效。
  • 检查 Cloudflare Workers 日志以查看部署中的任何错误。
  • 如果使用免费账户时出现 Error: Browser Rendering API is only available for Workers Paid plans 错误,请参考上面的免费用户配置注意事项。

许可证

本项目采用 MIT 许可证。

致谢

workers-mcp-demo FAQ

How do I deploy the workers-mcp-demo server?
Clone the repository, install dependencies with pnpm, configure your Cloudflare Worker using Wrangler CLI, and deploy via Wrangler commands.
What prerequisites are needed to run workers-mcp-demo?
You need Node.js v16+, pnpm package manager, a Cloudflare account, Wrangler CLI, and an MCP-compatible client like Claude or Cursor.
Can I extend the AI capabilities with custom logic?
Yes, you can add your own serverless functions within Cloudflare Workers to create custom AI tools accessible via MCP.
Which MCP clients are compatible with workers-mcp-demo?
It works seamlessly with Claude, Cursor, and other MCP-supporting AI assistants.
Is workers-mcp-demo suitable for production use?
While primarily a demo, it can be adapted for production with proper configuration and scaling on Cloudflare Workers.
How does workers-mcp-demo enhance AI assistant interactions?
By exposing custom tools and data sources through MCP, it allows AI assistants to generate more relevant and context-aware responses.
What is the role of Cloudflare Workers in this MCP server?
Cloudflare Workers provide a serverless environment to host and run MCP server logic, enabling scalable and low-latency AI tool execution.
How do I update or customize the MCP server?
Modify the serverless functions in the repository, then redeploy using Wrangler CLI to update the MCP server.