AI Sandbox MCP Server

A Model Context Protocol (MCP) server for AI Sandbox that provides custom tools and resources for AI assistants. This package can be installed globally to provide MCP capabilities to Claude and other AI assistants.

Features

• Greeting Resources: Personalized greeting messages
• Calculator Tools: Basic math operations (add, subtract, multiply, divide)
• Echo Tool: Simple tool for testing
• Global Installation: Can be installed as a global CLI tool

Prerequisites

• Node.js v18.x or higher
• npm or yarn
• GitHub account with access to the repository

Installation Options

Global Installation from GitHub Packages

1. Configure npm to use GitHub Packages for the scope:

Create or edit ~/.npmrc to add:

@dil-ddaradics:registry=https://npm.pkg.github.com/
//npm.pkg.github.com/:_authToken=YOUR_GITHUB_TOKEN

Replace YOUR_GITHUB_TOKEN with your GitHub Personal Access Token (PAT).

2. Install the package globally:

npm install -g @dil-ddaradics/ai-sandbox-mcp-server

3. Verify the installation:

ai-sandbox-mcp --version

Local Development Installation

1. Clone the AI Sandbox repository:

git clone https://github.com/dil-ddaradics/ai-sandbox.git
cd ai-sandbox/mcp-server

2. Install dependencies:

npm install

Usage

Running the Server

Using Global Installation

If installed globally, simply run:

ai-sandbox-mcp

This will start the MCP server using stdio transport.

Using Local Installation

To run the MCP server locally:

npm start

Connecting to Claude Code

With Global Installation

To use the globally installed MCP server with Claude Code:

claude mcp add --transport stdio ai-sandbox-mcp -- ai-sandbox-mcp

With Local Installation

To use the local MCP server with Claude Code:

claude mcp add --transport stdio ai-sandbox-mcp -- 'npm start --prefix /path/to/ai-sandbox/mcp-server'

Replace /path/to/ai-sandbox/mcp-server with the actual path to the mcp-server directory.

Testing the Server

Using Resources

Once connected to Claude Code, you can test the greeting resource:

Use the ai-sandbox-mcp server to get a greeting with my name.

Using Tools

To test the calculator tools:

Use the ai-sandbox-mcp server to add 25 and 17.

Use the ai-sandbox-mcp server to divide 100 by 5.

To test the echo tool:

Use the ai-sandbox-mcp server to echo "Hello, MCP!".

Development

Project Structure

mcp-server/
├── src/
│   ├── index.ts        # Entry point
│   ├── resources/      # Resource implementations
│   │   └── greeting.ts # Greeting resource
│   └── tools/          # Tool implementations
│       ├── calculator.ts # Calculator tools
│       └── echo.ts      # Echo tool
├── dist/              # Compiled JavaScript
├── tsconfig.json      # TypeScript configuration
├── package.json       # npm configuration
└── README.md          # This documentation

Building

To build the TypeScript code:

npm run build

This will compile the TypeScript code into JavaScript in the dist directory.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License.