My MCP Server
An extensible TypeScript-based MCP server designed for Claude Code with a modular architecture for easily adding custom tools. It includes built-in examples like a calculator and echo tool, utilizing Zod for robust input validation and error handling.
README
My MCP Server
A custom Model Context Protocol (MCP) server with an extensible tool architecture for Claude Code. This server makes it easy to add custom tools that Claude can use during conversations.
Features
- Extensible Architecture: Easy-to-use pattern for adding new tools
- TypeScript: Full type safety with TypeScript
- Example Tools Included: Calculator and Echo tools to demonstrate the pattern
- Production-Ready: Proper error handling and validation with Zod
- Well-Documented: Clear examples and documentation for adding new tools
Available Tools
1. Calculate
Perform basic arithmetic operations (add, subtract, multiply, divide).
Example usage:
- "Calculate 15 + 27"
- "What's 100 divided by 4?"
- "Multiply 8 by 9"
2. Echo
Echo a message back, with optional text transformation (uppercase/lowercase).
Example usage:
- "Echo 'Hello World'"
- "Echo 'test message' in uppercase"
- "Echo 'LOUD TEXT' in lowercase"
Installation
1. Install Dependencies
npm install
2. Build the Project
npm run build
This will compile the TypeScript code to JavaScript in the build/ directory.
3. Configure Claude Code
Add the following configuration to your ~/.claude.json file:
{
"mcpServers": {
"my-mcp": {
"type": "stdio",
"command": "node",
"args": ["/Users/martin/src/my-mcp/build/index.js"]
}
}
}
Important: Make sure to use the absolute path to your build/index.js file.
4. Restart Claude Code
After updating the configuration, restart Claude Code to load the MCP server.
Adding New Tools
Adding a new tool is simple and follows a consistent pattern:
Step 1: Create a New Tool File
Create a new file in src/tools/ with the naming convention your-tool.tool.ts:
import { z } from "zod";
import type { ToolDefinition } from "../types/index.js";
export const yourTool: ToolDefinition = {
name: "your_tool_name",
description: "Description of what your tool does",
inputSchema: z.object({
param1: z.string().describe("Description of parameter 1"),
param2: z.number().optional().describe("Optional parameter 2"),
}),
handler: async (args) => {
// Your tool logic here
const result = `Processed: ${args.param1}`;
return {
content: [{
type: "text",
text: result
}]
};
}
};
Step 2: Register the Tool
Add your tool to the registry in src/tools/index.ts:
import { yourTool } from "./your-tool.tool.js";
export const tools: ToolDefinition[] = [
echoTool,
calculatorTool,
yourTool, // Add your tool here
];
Step 3: Rebuild
npm run build
Step 4: Restart Claude Code
Restart Claude Code to load the new tool.
Step 5: Test
Try using your new tool in a conversation with Claude!
Tool Examples
Simple Text Processing Tool
import { z } from "zod";
import type { ToolDefinition } from "../types/index.js";
export const reverseTool: ToolDefinition = {
name: "reverse",
description: "Reverse a string",
inputSchema: z.object({
text: z.string().describe("The text to reverse"),
}),
handler: async (args) => {
const reversed = args.text.split("").reverse().join("");
return {
content: [{
type: "text",
text: reversed
}]
};
}
};
API Call Tool
import { z } from "zod";
import type { ToolDefinition } from "../types/index.js";
export const weatherTool: ToolDefinition = {
name: "get_weather",
description: "Get current weather for a location",
inputSchema: z.object({
location: z.string().describe("City name or zip code"),
}),
handler: async (args) => {
try {
// Make API call (example)
const response = await fetch(
`https://api.weather.example.com/current?location=${args.location}`
);
const data = await response.json();
return {
content: [{
type: "text",
text: `Weather in ${args.location}: ${data.temperature}°F, ${data.conditions}`
}]
};
} catch (error) {
return {
isError: true,
content: [{
type: "text",
text: `Failed to fetch weather: ${error instanceof Error ? error.message : 'Unknown error'}`
}]
};
}
}
};
Project Structure
my-mcp/
├── src/
│ ├── index.ts # Main entry point
│ ├── server.ts # Server initialization and request handlers
│ ├── tools/
│ │ ├── index.ts # Tool registry (add new tools here)
│ │ ├── calculator.tool.ts # Calculator tool implementation
│ │ └── echo.tool.ts # Echo tool implementation
│ └── types/
│ └── index.ts # TypeScript type definitions
├── build/ # Compiled JavaScript (generated)
├── package.json # Project dependencies and scripts
├── tsconfig.json # TypeScript configuration
├── .gitignore # Git ignore rules
├── .env.example # Environment variable template
└── README.md # This file
Development
Build the Project
npm run build
Run in Development Mode
npm run dev
This will build and run the server, showing any startup logs.
Error Handling
All tools should handle errors gracefully and return structured error responses:
// Success response
return {
content: [{
type: "text",
text: "Operation successful"
}]
};
// Error response
return {
isError: true,
content: [{
type: "text",
text: "Error: Something went wrong"
}]
};
Important Notes
- Logging: Always use
console.error()for logging, neverconsole.log(). The latter corrupts JSON-RPC messages over stdio. - Input Validation: All inputs are automatically validated using Zod schemas
- Error Messages: Provide clear, helpful error messages for debugging
- Async Handlers: All tool handlers are async functions
Troubleshooting
Server Not Appearing in Claude Code
- Check that the path in
~/.claude.jsonis absolute and correct - Verify the build succeeded:
npm run build - Restart Claude Code completely
- Check for errors:
node build/index.js(should start without errors)
Tool Not Working
- Verify the tool is added to the registry in
src/tools/index.ts - Rebuild the project:
npm run build - Restart Claude Code
- Check the Zod schema matches your expected input format
TypeScript Errors
- Make sure all imports use
.jsextensions (required for ES modules) - Run
npm run buildto see detailed error messages - Check that types are correctly imported from
../types/index.js
License
MIT
Contributing
Feel free to add more tools and share them! The modular architecture makes it easy to create and share tool implementations.
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。