MCP Configurator Server
Manages, edits, backups, and validates Claude Desktop and other client JSON configurations.
README
MCP Configurator Server
A Model Context Protocol (MCP) server that provides tools to manage, edit, backup, and validate Claude Desktop and other client JSON configurations.
This application is built with the official @modelcontextprotocol/sdk and zod for argument schema validation, enabling client LLMs (like Claude or ChatGPT) to programmatically read and modify their own configuration and environment setups.
Folder Structure
mcp-configurator/
├── lib/
│ ├── backup.js # Handles backup creation, listing, and restores
│ ├── config.js # Handles JSON loading, syntax validation, and updates
│ ├── validator.js # Checks system path command executable existence
│ └── utils.js # Filepath resolvers and cross-platform defaults
├── backups/ # Stores configuration backup files (.bak)
├── package.json # Project metadata and dependencies (type: module)
├── server.js # Server entry point running StdioServerTransport
└── README.md # Documentation
Installation
- Install Dependencies:
Open a terminal in the project directory and run:
npm install
Running the Server
Since it uses Stdio transport, the server is meant to be spawned by an MCP host (such as Claude Desktop or Cursor). You can run it manually to check for syntax or launch issues:
npm start
Note: This command will run and listen on stdin/stdout. You will see MCP Configurator server running on stdio outputted to stderr to avoid polluting the JSON-RPC channel.
Integration Setup
1. Claude Desktop Configuration
To add this configurator server to Claude Desktop on Windows, open your %APPDATA%\Claude\claude_desktop_config.json and add the mcp-configurator to your mcpServers list:
{
"mcpServers": {
"mcp-configurator": {
"command": "node",
"args": [
"C:/Desktop/my workspace/work/server.js"
]
}
}
}
Replace C:/Desktop/my workspace/work/server.js with the actual absolute path to where you saved this project.
2. Cursor or other Local IDEs
For Cursor or VS Code, configure a new stdio-based MCP client with:
- Command:
node - Arguments:
C:/Desktop/my workspace/work/server.js
Tool Descriptions
The server registers the following tools:
get_config
Reads a configuration JSON file.
- Input:
path(string, optional): Custom absolute path to the configuration JSON file. If omitted, defaults to the current system's Claude Desktop config file path.
- Output:
- The full JSON config payload.
save_config
Saves configuration JSON after automatically creating a timestamped backup of the current state.
- Input:
path(string, required): Absolute path to the file to overwrite.config(object, required): The new JSON configuration object.
- Output:
- Success message with the name of the backup created.
list_backups
Lists all available configuration backups under the backups/ workspace directory.
- Input: None.
- Output:
- Formatted list of all files detailing filename, backup date, and file size.
restore_backup
Reverts the target configuration file using a selected backup file. Creates a new backup of the current state before overwriting as a safety measure.
- Input:
filename(string, required): Filename of the backup (e.g.claude_desktop_config.json.bak-2026-07-08T11-04-05-123Z).path(string, optional): Custom absolute path to restore into. Defaults to the active configuration path.
- Output:
- Confirmation message of successful restoration.
validate_command
Spawns standard shell diagnostic tools (where on Windows, which on macOS/Linux) or queries paths to confirm if an executable exists on the host machine. Helpful for verifying if custom MCP server commands (e.g. npx, python, git) are executable.
- Input:
command(string, required): The executable binary name or absolute path to check.
- Output:
- JSON object indicating success status, resolved PATH location, and descriptive message.
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。