我的MCP相关配置记录

1.VSCode的Cline中的MCP

{
  "mcpServers": {
    "github.com/modelcontextprotocol/servers/tree/main/src/github": {
      "autoApprove": [],
      "disabled": false,
      "timeout": 60,
      "command": "cmd",
      "args": [
        "/c",
        "npx",
        "-y",
        "@modelcontextprotocol/server-github"
      ],
      "env": {
        "GITHUB_PERSONAL_ACCESS_TOKEN": "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"
      },
      "transportType": "stdio"
    },
    
    "filesystem": {
      "timeout": 60,
      "command": "cmd",
      "args": [
        "/c",
        "npx",
        "-y",
        "@modelcontextprotocol/server-filesystem",
        "C:\\Users\\21031\\Desktop"
      ],
      "transportType": "stdio"
    },

    "time": {
      "timeout": 60,
      "command": "python",
      "args": [
        "-m",
        "mcp_server_time",
        "--local-timezone=Asia/Shanghai"
      ],
      "transportType": "stdio"
    },

    "browser-tools-mcp": {
      "autoApprove": [
        "getConsoleLogs",
        "getConsoleErrors",
        "getSelectedElement",
        "runAccessibilityAudit",
        "getNetworkLogs",
        "getNetworkErrors"
      ],
      "timeout": 60,
      "command": "cmd",
      "args": [
        "/c",
        "npx",
        "-y",
        "@agentdeskai/browser-tools-mcp@1.2.0"
      ],
      "transportType": "stdio"
    },

    "fetch": {
      "timeout": 60,
      "command": "uvx",
      "args": [
        "mcp-server-fetch"
      ],
      "transportType": "stdio"
    },

    "sequential-thinking": {
      "timeout": 60,
      "command": "cmd",
      "args": [
        "/c",
        "npx",
        "-y",
        "@modelcontextprotocol/server-sequential-thinking"
      ],
      "transportType": "stdio"
    },

    "firecrawl-mcp": {
      "timeout": 60,
      "command": "cmd",
      "args": [
        "/c",
        "npx",
        "-y",
        "firecrawl-mcp"
      ],
      "env": {
        "FIRECRAWL_API_KEY": "fc-bb842f816a1d42ad8867e808d579bcdc"
      },
      "transportType": "stdio"
    },

    "puppeteer": {
      "timeout": 60,
      "command": "cmd",
      "args": [
        "/c",
        "npx",
        "-y",
        "@modelcontextprotocol/server-puppeteer"
      ],
      "transportType": "stdio"
    },

    "mysqldb-mcp-server": {
      "timeout": 60,
      "command": "uvx",
      "args": [
        "mysqldb-mcp-server"
      ],
      "env": {
        "MYSQL_HOST": "localhost",
        "MYSQL_USER": "root",
        "MYSQL_PASSWORD": "1234",
        "MYSQL_DATABASE": "company",
        "MYSQL_READONLY": "false"
      },
      "transportType": "stdio"
    },

    "desktop-commander": {
      "disabled": false,
      "timeout": 60,
      "command": "cmd",
      "args": [
        "/c",
        "npx",
        "-y",
        "@smithery/cli@latest",
        "run",
        "@wonderwhy-er/desktop-commander",
        "--key",
        "c1616ea0-7665-4811-81d1-0f634d17dd9a"
      ],
      "transportType": "stdio"
    }
  }
}

2.CheeryStudio中的MC配置

{
  "mcpServers": {
    "filesystem": {
      "name": "@cherry/filesystem",
      "type": "inMemory",
      "description": "实现文件系统操作的模型上下文协议（MCP）的 Node.js 服务器",
      "isActive": true,
      "args": [
        "C:\\Users\\21031\\Desktop"
      ]
    },

    "Z5QlEkkaow3lL-Ccl6pps": {
      "isActive": false,
      "name": "MCP 服务器",
      "description": "",
      "baseUrl": "",
      "command": "",
      "args": [],
      "env": {}
    },

    "fetch": {
      "isActive": true,
      "name": "fetch-server",
      "type": "stdio",
      "description": "抓取网页",
      "registryUrl": "",
      "command": "uvx",
      "args": [
        "mcp-server-fetch"
      ]
    },

    "amap-maps": {
      "isActive": true,
      "name": "amap-maps",
      "type": "stdio",
      "description": "高德地图",
      "registryUrl": "",
      "command": "npx",
      "args": [
        "-y",
        "@amap/amap-maps-mcp-server"
      ],
      "env": {
        "AMAP_MAPS_API_KEY": "20cd06f09910513998ed05e863f19282"
      }
    },

    "sequential-thinking": {
      "isActive": true,
      "name": "sequential-thinking",
      "type": "stdio",
      "description": "赋予大模型推理能力，适合解决复杂问题",
      "registryUrl": "",
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-sequential-thinking"
      ]
    },

    "firecrawl": {
      "isActive": true,
      "name": "firecrawl",
      "type": "stdio",
      "description": "高级网页抓取与爬取（用于上下文）处理 JS 渲染、批量作业、提取结构化数据，并能在多个页面进行深度研究。",
      "registryUrl": "",
      "command": "npx",
      "args": [
        "-y",
        "firecrawl-mcp"
      ],
      "env": {
        "FIRECRAWL_API_KEY": "fc-bb842f816a1d42ad8867e808d579bcdc"
      }
    },

    "puppeteer": {
      "isActive": true,
      "name": "puppeteer",
      "type": "stdio",
      "description": "自动化浏览器操作。与页面交互、截图以及在实际浏览器环境中执行 JavaScript，用于测试或自动化。",
      "registryUrl": "",
      "command": "npx",
      "args": [
        "-y",
        "@modelcontextprotocol/server-puppeteer"
      ]
    },

    "desktop-commander": {
      "isActive": true,
      "name": "desktop-commander",
      "type": "stdio",
      "description": "控制台，终端命令交互",
      "registryUrl": "",
      "command": "npx",
      "args": [
        "-y",
        "@wonderwhy-er/desktop-commander"
      ]
    }
  }
}

补充MCP文档：

昨天，Cline发布了自己 MCP Marketplace 中最受欢迎的10款 MCP，以及四款荣誉提名的MCP（在最后）：

如果你平时更经常用Cursor、Windsurf、Cherry Studio等MCP客户端，可以去之前提到的MCP资源站搜索找到。当然，这篇文章也整理了这14款MCP在GitHub上的仓库地址，大家根据需求自取即可：

1、GitHub：可以在 Cline 直接管理仓库、文件、问题和 PR。将版本控制工作流程无缝集成到你的开发过程中的必备工具。

🔗：https://github.com/github/github-mcp-server

2、Browser Tools：通过 Chrome 扩展监控浏览器活动、捕获屏幕截图、分析日志以及通过 DOM 进行交互。非常适合可视化调试 Web 应用。

🔗：https://github.com/AgentDeskAI/browser-tools-mcp

3、File System：允许 Cline 以编程方式访问详细的文件元数据。使用 get_file_info 函数可以检索大小、修改时间、权限和类型（文件/目录），用于脚本编写或分析。

🔗：https://github.com/modelcontextprotocol/servers/tree/main/src/filesystem

4、Git Tools：使用状态、diff、提交、分支管理等命令与 Git 仓库交互。可以以编程方式自动化常见的 Git 任务。

🔗：https://github.com/modelcontextprotocol/servers/tree/main/src/git

5、Sequential Thinking ：有条不紊地解决复杂问题。这个 MCP 支持结构化、逐步推理，将复杂任务分解为可管理的部分，以实现更稳健的解决方案。

🔗：https://github.com/modelcontextprotocol/servers/tree/main/src/sequentialthinking

6、Fetch：直接将网页内容（HTML、JSON、文本、Markdown）作为上下文拉取到 Cline。适用于快速使用特定 URL 的信息让 Cline 定位。

🔗：https://github.com/zcaceres/fetch-mcp

🔗：https://github.com/modelcontextprotocol/servers/tree/main/src/fetch

7、FireCrawl：高级网页抓取与爬取*（用于上下文）*。处理 JS 渲染、批量作业、提取结构化数据，并能在多个页面进行深度研究。

🔗：https://github.com/mendableai/firecrawl-mcp-server

8、Browser Use：使用自然语言控制网络浏览器。浏览页面、填写表单、理解视觉内容——适用于自动化复杂的网络交互。

🔗：https://github.com/Saik0s/mcp-browser-use

9、Puppeteer：使用 Puppeteer 自动化浏览器操作。与页面交互、截图以及在实际浏览器环境中执行 JavaScript，用于测试或自动化。

🔗：https://github.com/modelcontextprotocol/servers/tree/main/src/puppeteer

10、Playwright：另一个使用强大的浏览器自动化服务器。支持交互、截图和 JS 执行，用于跨浏览器测试和自动化。

🔗：https://github.com/microsoft/playwright-mcp

还有四个 MCP 获得荣誉提名（是Cline用户大量应用的MCP）：

Context7（AI 友好的文档）：https://github.com/upstash/context7

Perplexity（AI 研究）：https://github.com/ppl-ai/modelcontextprotocol

21st Dev 的 Magic UI（UI 生成）：https://github.com/21st-dev/magic-mcp

Supabase（后端集成）：https://github.com/supabase-community/supabase-mcp