Example MCP Server
A Node.js/TypeScript MCP server template with sample tools (ping and system_info) that demonstrates how to build custom tools for Claude Desktop using stdio transport.
README
Example MCP Server
A simple Model Context Protocol (MCP) server implementation in Node.js with TypeScript for use with Claude Desktop.
What is MCP?
The Model Context Protocol (MCP) is an open protocol that enables AI models like Claude to interact with external tools and data sources. It allows Claude Desktop to call local or remote tools, retrieve information, and perform actions through standardized JSON-RPC messaging.
This server exposes tools that Claude can discover and invoke over a stdio (standard input/output) transport, making it perfect for local integrations.
Project Structure
.
├── src/
│ ├── mcp/
│ │ └── server.ts # Main MCP server implementation
│ ├── tools/
│ │ ├── ping.ts # Ping tool example
│ │ ├── system-info.ts # System info tool example
│ │ └── registry.ts # Tool registry & dispatcher
│ └── types/
│ └── index.ts # TypeScript type definitions
├── dist/ # Compiled JavaScript (generated)
├── package.json
├── tsconfig.json
├── mcp.json # Claude Desktop manifest
└── README.md
Features
- ✅ MCP-compliant JSON-RPC 2.0 server
- ✅ Stdio transport (works with Claude Desktop local tools)
- ✅ Two sample tools:
pingandsystem_info - ✅ Extensible tool registry system
- ✅ Proper error handling and validation
- ✅ TypeScript for type safety
Prerequisites
- Node.js 16+ (or compatible version)
- npm or yarn
Getting Started
1. Install Dependencies
npm install
2. Build TypeScript to JavaScript
npm run build
This compiles TypeScript from src/ to JavaScript in dist/.
3. Run the MCP Server
npm start
The server will start and listen for JSON-RPC requests on stdin.
Available Tools
ping
A simple test tool that returns "pong".
Request:
{"jsonrpc": "2.0", "id": 1, "method": "tools/call", "params": {"name": "ping"}}
Response:
{"jsonrpc": "2.0", "id": 1, "result": {"content": [{"type": "text", "text": "pong"}]}}
system_info
Returns detailed system information including OS, architecture, CPU count, memory usage, and uptime.
Request:
{"jsonrpc": "2.0", "id": 2, "method": "tools/call", "params": {"name": "system_info"}}
Response:
{
"jsonrpc": "2.0",
"id": 2,
"result": {
"content": [{
"type": "text",
"text": "{\n \"platform\": \"win32\",\n \"arch\": \"x64\",\n \"osType\": \"Windows_NT\",\n ...\n}"
}]
}
}
MCP Protocol Methods
tools/list
Lists all available tools.
Request:
{"jsonrpc": "2.0", "id": 1, "method": "tools/list"}
Response:
{
"jsonrpc": "2.0",
"id": 1,
"result": {
"tools": [
{
"name": "ping",
"description": "A simple ping tool...",
"inputSchema": { "type": "object", "properties": {} }
},
...
]
}
}
tools/call
Calls a specific tool with optional arguments.
Request:
{
"jsonrpc": "2.0",
"id": 1,
"method": "tools/call",
"params": {
"name": "tool_name",
"arguments": { "arg1": "value1" }
}
}
Response:
{
"jsonrpc": "2.0",
"id": 1,
"result": {
"content": [
{ "type": "text", "text": "Result here" }
]
}
}
Registering with Claude Desktop
To use this MCP server with Claude Desktop:
-
Build the project:
npm run build -
Locate your Claude Desktop configuration file:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json - Linux:
~/.config/Claude/claude_desktop_config.json
- macOS:
-
Add the server to your
claude_desktop_config.json:{ "mcpServers": { "example-mcp-server": { "command": "node", "args": ["/absolute/path/to/dist/mcp/server.js"], "env": {} } } }Replace
/absolute/path/towith the full path to your project folder. -
Restart Claude Desktop to load the new MCP server.
-
Claude will now be able to discover and call the tools provided by this server.
Adding New Tools
To add a new tool:
-
Create a new tool file in
src/tools/(e.g.,src/tools/my-tool.ts):import { Tool, ToolResult, ToolHandler } from "../types/index.js"; export const myTool: Tool = { name: "my_tool", description: "Description of what the tool does", inputSchema: { type: "object", properties: { param1: { type: "string" }, }, required: ["param1"], }, }; export const myToolHandler: ToolHandler = async (args) => { // Implementation return { content: [{ type: "text", text: "Result" }], }; }; -
Register it in
src/tools/registry.ts:import { myTool, myToolHandler } from "./my-tool.js"; const toolRegistry: Map<string, ToolEntry> = new Map([ // ... existing tools ["my_tool", { definition: myTool, handler: myToolHandler }], ]); -
Rebuild and restart the server:
npm run build npm start
Development
For development with auto-reload, you can use ts-node:
npm run dev
This will run the TypeScript directly without compilation (requires ts-node to be installed).
Error Handling
The server implements proper JSON-RPC 2.0 error handling:
- -32700: Parse error
- -32600: Invalid request
- -32601: Method not found
- -32603: Internal error
All errors are returned with descriptive messages to help with debugging.
Architecture
MCP Server (src/mcp/server.ts)
- Reads JSON-RPC 2.0 messages from stdin
- Routes requests to appropriate handlers
- Validates incoming requests
- Writes responses to stdout
- Handles errors gracefully
Tool Registry (src/tools/registry.ts)
- Central registry of available tools
- Provides tool list for discovery
- Dispatches tool calls to appropriate handlers
Tool Handlers
Each tool has:
- A
Tooldefinition (name, description, input schema) - A
ToolHandlerfunction (async execution logic)
Troubleshooting
Server not starting
- Ensure Node.js is installed:
node --version - Check that dependencies are installed:
npm install - Build the project:
npm run build
Tools not visible in Claude
- Verify the full path in
claude_desktop_config.jsonis correct - Restart Claude Desktop after configuration changes
- Check server logs for errors
Connection issues
- Ensure the server process stays running
- Check file permissions on the compiled files
- Verify stdin/stdout are not being intercepted
License
MIT
References
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。