MCP Multi-Language Sandbox MCP Server

MCP Multi-Language Sandbox

Execute code securely in 6 programming languages with Docker isolation, designed for Claude Code via Model Context Protocol (MCP).

What is this?

A local MCP server that lets Claude execute code in isolated Docker containers. Think of it as your own private code sandbox - 100% free, 100% local, no cloud dependencies.

Why use this instead of cloud sandboxes?

• Free: No per-execution costs (vs ~$0.10/run on cloud services)
• Fast: 0ms container acquisition with pooling (vs 2-5s cold starts)
• Private: Code never leaves your machine
• Customizable: Add your own languages, packages, security rules

Features

• 6 Languages: Python, TypeScript, JavaScript, Go, Rust, Bash
• Container Pooling: Pre-warmed containers for instant execution
• Package Caching: Install once, reuse forever (SHA256-based)
• ML Runtime: numpy, pandas, sklearn, torch, mlx pre-installed
• Security: Seccomp profiles, capability dropping, audit logging
• Sessions: Persistent state with TTL and auto-cleanup

Quick Start

Prerequisites

• Node.js >= 18.0.0
• Docker Desktop
• Claude Code CLI (optional, for MCP integration)

Installation

# Clone the repository
git clone https://github.com/Pit-CL/mcp-multilang-sandbox.git
cd mcp-multilang-sandbox

# Install dependencies
npm install

# Build
npm run build

# Run tests (optional)
npm run test:mcp

Add to Claude Code

# Add as MCP server
claude mcp add multilang-sandbox node /path/to/mcp-multilang-sandbox/dist/mcp/server.js

# Verify it's connected
claude mcp list
# Should show: multilang-sandbox ✓ Connected

Manual Configuration

Add to your Claude settings (~/.claude.json or VS Code settings):

{
  "mcpServers": {
    "multilang-sandbox": {
      "command": "node",
      "args": ["/path/to/mcp-multilang-sandbox/dist/mcp/server.js"],
      "env": {
        "LOG_LEVEL": "info"
      }
    }
  }
}

Usage

Once configured, Claude can use these tools:

Execute Code

// Python
sandbox_execute({ language: 'python', code: 'print("Hello!")' })

// TypeScript
sandbox_execute({ language: 'typescript', code: 'console.log("Hello!")' })

// With ML libraries (numpy, pandas, sklearn, torch)
sandbox_execute({
  language: 'python',
  code: 'import numpy as np; print(np.array([1,2,3]))',
  ml: true
})

Persistent Sessions

// Create a session
sandbox_session({ action: 'create', name: 'my-project', language: 'python' })

// Execute in session (state persists)
sandbox_execute({ language: 'python', code: 'x = 42', session: 'my-project' })
sandbox_execute({ language: 'python', code: 'print(x)', session: 'my-project' })  // prints 42

// Install packages
sandbox_install({ session: 'my-project', packages: ['pandas', 'requests'] })

// Cleanup
sandbox_session({ action: 'destroy', name: 'my-project' })

File Operations

// Write a file
sandbox_file_ops({ session: 'my-project', operation: 'write', path: 'data.csv', content: 'a,b\n1,2' })

// Read it back
sandbox_file_ops({ session: 'my-project', operation: 'read', path: 'data.csv' })

System Stats

// View pool, cache, and session stats
sandbox_inspect({ target: 'all' })

// Security audit
sandbox_security({ action: 'stats' })

MCP Tools Reference

Architecture

┌─────────────────────────────────────────────────────────┐
│                    Claude / MCP Client                   │
└───────────────────────────┬─────────────────────────────┘
                            │ JSON-RPC (stdio)
┌───────────────────────────▼─────────────────────────────┐
│                    MCP Sandbox Server                    │
│  ┌────────────────────────────────────────────────────┐ │
│  │  Tools: execute | session | install | file_ops     │ │
│  │         inspect | security                          │ │
│  ├────────────────────────────────────────────────────┤ │
│  │  Core: ContainerPo