Mofei Life MCP Server
Enables browsing Mofei's blog posts, fetching article details, comments, and recommendations via MCP tools.
README
ChatGPT App MCP Backend for Mofei's Blog
This repository hosts the Cloudflare Workers MCP server that powers the blog app from the article “How to Build a ChatGPT App From Scratch: MCP Integration, Custom Widgets, and Real API Examples”. It exposes Mofei's blog APIs as MCP tools so ChatGPT (and any MCP client) can browse posts, fetch details, and surface recommendations.
What’s inside
- Cloudflare Worker + Durable Object MCP server (
src/index.ts) with SSE endpoints at/sseand/mcp. - Four blog-focused MCP tools:
list-blog-posts,get-blog-article,get-comment-list, andget-recommend-blog. - Wrangler configuration (
wrangler.jsonc) ready for local dev and production deploy.
Quick start (local)
- Install dependencies:
npm install - Start the dev server:
npm run dev - The MCP endpoints are available at
http://localhost:8787/sse(SSE) andhttp://localhost:8787/mcp.
If you need to expose local dev to ChatGPT, run ngrok http 8787 (or your tunneling tool) and append /sse to the public URL.
Deploy to Cloudflare Workers
- Configure your Cloudflare account with Wrangler (
npm create cloudflare@latestif you need an account setup). - Deploy:
npm run deploy - Your public MCP endpoint will be
https://<worker-name>.<your-account>.workers.dev/sse.
Connect from ChatGPT (Apps Developer Mode)
- In ChatGPT: Settings → Apps & Connectors → Advanced settings → Enable “Developer mode”.
- Create a connection and set “MCP Server URL” to your endpoint:
- Cloudflare:
https://<worker-name>.<your-account>.workers.dev/sse - Local via ngrok:
https://<random>.ngrok.io/sse - Keep the trailing
/sse.
- Cloudflare:
- Test the connection to confirm tools and resources are detected, then save.
- Try a prompt such as: “Show me the latest posts from Mofei’s blog in English.”
Available tools
list-blog-posts— Paginated blog list. Params:page(number),lang(en/zh).get-blog-article— Fetch a specific article byidandlang.get-comment-list— Retrieve comments withpageandpageSize.get-recommend-blog— Related articles for a givenidandlang.
Responses return the underlying JSON from https://api.mofei.life. For UI widgets, pair this server with the single-file widget bundle described in the blog post (structured content for the model + _meta for the widget data).
Useful scripts
npm run dev— Local dev server on port 8787.npm run deploy— Deploy to Cloudflare Workers.npm run format/npm run lint:fix— Biome formatting and linting.
More resources
- Full article: https://www.mofei.life/en/blog/article/chatgpt-app
- ChatGPT Apps SDK docs: https://developers.openai.com/apps-sdk
- MCP spec: https://modelcontextprotocol.io/
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。