问题说明

mcp就相当于给AI安装了工具包，它可以调用获取接口文档，网页，数据库等，基本上所有的mcp都是node程序安装到npx缓存目录，少数需要python环境

使用说明

使用mcp-mysql举例，下面是配置json

 "MySQL": {
	 "command": "npx",
	   "args": [
	     "mcprunner",
	     "MYSQL_HOST=",
	     "MYSQL_PORT=",
	     "MYSQL_USER=",
	     "MYSQL_PASS=",
	     "MYSQL_DB=",
	     "ALLOW_INSERT_OPERATION=true",
	     "ALLOW_UPDATE_OPERATION=true",
	     "ALLOW_DELETE_OPERATION=false",
	     "--",
	     "npx",
	     "-y",
	     "@benborla29/mcp-server-mysql"
	   ]
	 }

可以看到启动命令就是 npx mcprunner MYSQL_HOST= ....

npx 命令会去查找这个模块：

1. 先查当前项目的 node_modules/.bin
2. 找不到再临时下载到缓存目录
3. 执行脚本

npm config get prefix
可以查看当前node的缓存目录
一般都在 缓存目录\node_cache\\_npx 可以看最新的日期来判断

查看启动日志，output选择mcp logs

我遇到的报错问题，就是缺少相关依赖，需要管理员cmd
通过报错信息找到依赖存在的npx文件夹将它删除掉

然后通过命令全局安装依赖，或者缺少的依赖

npm install -g @upstash/context7-mcp@latest

然后在运行它原本npx命令，正常运行即可

常用mcp

browser-tools-mcp：获取浏览器控制台还需要安装浏览器插件
还需要启动服务端（npx @agentdeskai/browser-tools-server@1.2.0）
API 文档：APIFOX文档获取
MySQL：读取MYSQL数据库
interactive-feedback：翻倍token使用
mcp-server-time：配置时间
playwright：操作浏览器进行测试
sequential-thinking：获取上下文
context7：获取最新工具包的文档进行开发

{
  "mcpServers": {
    "browser-tools-mcp": {
      "command": "cmd",
      "args": [
        "/c",
        "npx",
        "-y",
        "@agentdeskai/browser-tools-mcp@1.2.0"
      ]
    },
    "API 文档": {
      "command": "cmd",
      "args": [
        "/c",
        "npx",
        "-y",
        "apifox-mcp-server@latest",
        "--project=6425249"
      ],
      "env": {
        "APIFOX_ACCESS_TOKEN": "APS-shml7cvY71yXcAZcqRdKIgVofXrwCbWm"
      }
    },
    "MySQL": {
      "command": "npx",
      "args": [
        "mcprunner",
        "MYSQL_HOST=39.107.252.127",
        "MYSQL_PORT=3306",
        "MYSQL_USER=root",
        "MYSQL_PASS=xxmj666.",
        "MYSQL_DB=xxmj_app",
        "ALLOW_INSERT_OPERATION=true",
        "ALLOW_UPDATE_OPERATION=true",
        "ALLOW_DELETE_OPERATION=false",
        "--",
        "npx",
        "-y",
        "@benborla29/mcp-server-mysql"
      ]
    },
    "interactive-feedback": {
      "command": "uv",
      "args": ["--directory", "D:/workSpace/pyWorkSpace/interactive-feedback-mcp", "run", "server.py"],
      "timeout": 600,
      "autoApprove": ["interactive_feedback"]
    },
    "mcp-server-time": {
      "command": "uvx",
      "args": [
        "mcp-server-time",
        "--local-timezone=Asia/Shanghai"
      ]
    },
    "playwright": {
      "command": "npx",
      "args": ["@playwright/mcp@latest"]
    },
    "sequential-thinking": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-sequential-thinking"]
    },
    "context7": {
      "command": "npx",
      "args": ["-y", "@upstash/context7-mcp@latest"]
    },
    "shrimp-task-manager": {
      "command": "npx",
      "args": ["-y", "mcp-shrimp-task-manager"],
      "env": {
        "DATA_DIR": "D:/workspace/Aother/mcp-shrimp-task-manager/data",
        "TEMPLATES_USE": "en",
        "ENABLE_GUI": "false"
      }
    },
    "mcp-deepwiki": {
      "command": "npx",
      "args": ["-y", "mcp-deepwiki@latest"]
    }
  }
}