AI Apps Configuration MCP Server

A Model Context Protocol (MCP) server that provides easy access to configuration files for popular AI applications. Pre-configured with common AI tools like Gemini, Claude Desktop, Claude Code, VS Code, and Cursor.

Features

• Pre-configured mappings for popular AI applications
• Supports multiple configuration file formats (.plist, JSON, YAML, plain text)
• Provides tools to search, locate, and read config files
• Secure file operations with proper permission handling

Installation

npm install
npm run build

Usage

Start the server:

npm start

For development:

npm run dev

Restarting the Server After Code Changes

If you modify the server's source code (e.g., adding new tools or features), you'll need to restart the server to apply the changes. Follow these steps:

1. Stop the current server process: First, find the process ID (PID) of the running Node.js server. You can usually do this with grep:

   ps aux | grep 'node dist/index.js'

   Identify the PID from the output (it's typically the second column) and then stop the process:

   kill <PID>

   If the server was run with npm run dev, you can often stop it by pressing Ctrl+C in the terminal where it's running.

2. Recompile the TypeScript code:

   npm run build

3. Restart the server:

   npm run start

   Or for development with automatic reloading (if configured in package.json):

   npm run dev

Making the code executable

To make the script directly executable from the command line, follow these steps:

1. Add a bin entry to package.json: This tells npm how to create a symbolic link to your executable file.

   "bin": {
     "ai-apps-config-mcp": "dist/index.js"
   }

2. Add a shebang to src/index.ts: This line at the very top of your main script tells the system which interpreter to use.

   #!/usr/bin/env node

3. Make the file executable: After building, you need to give the output file execution permissions.

   chmod +x dist/index.js

4. Link the package: npm link creates a global symbolic link from ai-apps-config-mcp to your project, allowing you to run it from anywhere.

   npm link

MCP Tools

• list_apps - List all available applications with their configuration status
• find_config - Locate configuration files for a specific application
• read_config - Read and format configuration file contents
• search_config - Search within configuration files for specific settings
• add_config_location - Add new configuration locations for applications

Supported Applications & Configuration Locations

The server currently supports these AI applications and their configuration files:

Google Gemini

• ~/.gemini/settings.json - Gemini settings configuration

Claude Desktop

• ~/Library/Application Support/Claude/claude_desktop_config.json - Claude Desktop configuration

Claude Code

• ~/.claude/ - Claude Code configuration directory

Visual Studio Code

• ~/.vscode/mcp.json - VS Code MCP configuration

Cursor

• ~/.cursor/mcp.json - Cursor MCP configuration

Adding New Applications

You can add support for new applications using the add_config_location tool:

# Example: Add a new config location
{
  "app": "new-app",
  "path": "~/.config/new-app/config.json",
  "type": "file",
  "description": "New App configuration",
  "format": "json"
}

Example Configuration

Claude Desktop

To use this MCP server with Claude Desktop, add the following configuration to your claude-desktop-config.json file. This assumes you have made the script executable and linked it using npm link.

{
  "mcpServers": {
    "ai-apps-config": {
      "command": "npx",
      "args": [
        "ai-apps-config-mcp"
      ],
      "env": {}
    }
  }
}

Ask Claude Desktop to find my VScode mcp config