My First MCP
A simple MCP server that exposes a 'greet' tool which returns a friendly greeting for a given name.
README
My First MCP
NextFlow AI Internship — Submission 3 (Module 1.5 & 1.6)
This is my first working MCP (Model Context Protocol) server. The goal of this submission is to build the smallest possible end-to-end MCP pipeline — one server, one tool — and prove it actually works by connecting to it with MCP Inspector.
What this server does
The server exposes a single tool called greet:
greet(name)— takes a person's name as input and returns a friendly greeting, e.g.Hello, Ada!
It's intentionally simple. The point of this step isn't a useful tool — it's confirming that the server starts, registers a tool correctly, validates input with Zod, and responds over stdio the way a real MCP host would expect.
Tech stack
@modelcontextprotocol/sdk— official MCP server SDKzod— schema validation for tool input- TypeScript +
tsx— run TS directly without a separate build step
Project structure
my-first-mcp/
├── src/
│ └── index.ts # server setup + the "greet" tool
├── package.json
├── package-lock.json
├── tsconfig.json
├── README.md
└── .gitignore
Setup
npm install
Running the server
npm run dev
If it stays running with no crash and no visible output, that's expected —
the server is just waiting for input over stdio. Stop it with Ctrl+C.
Testing with MCP Inspector
Since this server talks over stdio, it needs a host or inspector to actually call it. To verify it manually:
npx @modelcontextprotocol/inspector npx tsx src/index.ts
Then, in the browser UI that opens:
- Go to the Tools tab and confirm
greetis listed with its description and input schema. - Call it with valid input (
name: "Ada") → returnsHello, Ada!. - Call it with invalid input (empty or wrong type) → rejected by Zod before the handler code runs, confirming input validation is working.
A screenshot of both the tool listing and a successful call is included with this submission.
Notes
- All logging in this server uses
console.error, neverconsole.logorprocess.stdout.write— the stdio transport uses standard output to talk to the host, so anything else written there would corrupt the connection. node_modules/anddist/are excluded via.gitignoreand are not part of this repo.
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。