bun-mcp

A Model Context Protocol (MCP) server that enables execution of TypeScript and JavaScript code using the Bun runtime. This server allows AI assistants to run code snippets with full TypeScript support and automatic package installation.

Features

• TypeScript Support: Execute TypeScript code directly without compilation setup
• Auto Package Installation: Bun automatically installs npm packages on first import
• Environment Variables: Automatically loads .env files in the working directory
• Fast Execution: Leverages Bun's speed for quick script execution
• Secure Timeout: 10-second execution timeout to prevent runaway scripts

Prerequisites

• Node.js 18+
• Bun installed on your system

Installation

npm install bun-mcp

Usage

As an MCP Server

Add to your MCP client configuration:

{
  "servers": {
    "bun-executor": {
      "command": "npx",
      "args": ["bun-mcp"]
    }
  }
}

Or run directly:

node server.js

Available Tools

run_script

Executes TypeScript/JavaScript code using Bun runtime.

Parameters:

• script (string, required): The TypeScript/JavaScript code to execute

Example:

// Simple calculation
console.log(2 + 2);

// Using TypeScript
interface User {
  name: string;
  age: number;
}

const user: User = { name: "Alice", age: 30 };
console.log(`Hello ${user.name}`);

// Import npm packages (auto-installed)
import chalk from 'chalk';
console.log(chalk.blue('Hello World'));

Security Considerations

⚠️ WARNING: This server executes arbitrary code with full system access. Only use in trusted environments.

Risks

• Full filesystem access
• Network access
• Environment variable access
• Can install and run any npm package

Recommended Safeguards

• Run in isolated environments (Docker, VMs)
• Implement resource limits
• Add package allowlisting
• Enable audit logging
• Use only with trusted inputs

Development

# Clone the repository
git clone https://github.com/ebeloded/bun-mcp.git
cd bun-mcp

# Install dependencies
npm install

# Run the server
npm start

License

MIT

Contributing

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