MCP Builder — 构建 MCP 服务器
创建自定义 MCP (Model Context Protocol) 服务器。
支持的框架
| 框架 |
模板命令 |
| Python FastMCP |
pip install mcp → python -c "from mcp.server.fastmcp import FastMCP" |
| Node.js SDK |
npm install @modelcontextprotocol/sdk |
快速模板
Python FastMCP
from mcp.server.fastmcp import FastMCP
mcp = FastMCP("my-server")
@mcp.tool()
def my_tool(param: str) -> str:
"""工具描述"""
return f"处理: {param}"
@mcp.resource("config://app")
def get_config() -> str:
return "应用配置"
if __name__ == "__main__":
mcp.run(transport="stdio")
Node.js SDK
import { Server } from "@modelcontextprotocol/sdk/server/index.js";
import { StdioServerTransport } from "@modelcontextprotocol/sdk/server/stdio.js";
const server = new Server({ name: "my-server", version: "1.0.0" }, {
capabilities: { tools: {} }
});
server.setRequestHandler(ListToolsRequestSchema, async () => ({
tools: [{ name: "my_tool", description: "...", inputSchema: { type: "object", properties: {} } }]
}));
server.setRequestHandler(CallToolRequestSchema, async (request) => {
return { content: [{ type: "text", text: "结果" }] };
});
const transport = new StdioServerTransport();
await server.connect(transport);
传输方式
| 方式 |
适用场景 |
| stdio |
本地子进程，直接运行 |
| SSE |
远程 HTTP 连接 |
| Streamable HTTP |
MCP 规范新版，支持流式响应 |
发布
# npm 发布
npm publish
# PyPI 发布
python -m build && twine upload dist/*
测试
# 本地测试 stdio
node -e "const io = require('child_process').spawn('node', ['server.js']); io.stdout.on('data', d => console.log(d.toString()));"
# 使用 MCP Inspector
npx @modelcontextprotocol/inspector node server.js
tianque6916