MCP Server Colab Exec

🚀 mcp-server-colab-exec

An MCP server that allocates Google Colab GPU runtimes (T4/L4) and executes Python code on them, enabling any MCP-compatible AI assistant to run GPU-accelerated code without local GPU hardware.

🚀 Quick Start

This MCP server can allocate Google Colab GPU runtimes (T4/L4) and execute Python code on them. It allows various MCP-compatible AI assistants like Claude Code, Claude Desktop, Gemini CLI, Cline, etc., to run GPU-accelerated code (CUDA, PyTorch, TensorFlow) without the need for local GPU hardware.

✨ Features

• Allocate Google Colab GPU runtimes (T4/L4) and execute Python code.
• Enable MCP-compatible AI assistants to run GPU-accelerated code without local GPU hardware.

📦 Installation

You can install the package using pip:

pip install mcp-server-colab-exec

Or run it directly with uvx:

uvx mcp-server-colab-exec

📚 Documentation

Prerequisites

• Python 3.10+
• A Google account with access to Google Colab
• On first run, a browser window opens for OAuth2 consent. The token is cached at ~/.config/colab-exec/token.json for subsequent runs.

Configuration

Claude Code

Add to your project's .mcp.json or ~/.claude/.mcp.json:

{
  "mcpServers": {
    "colab-exec": {
      "command": "mcp-server-colab-exec"
    }
  }
}

Or via the CLI:

claude mcp add colab-exec mcp-server-colab-exec

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "colab-exec": {
      "command": "mcp-server-colab-exec"
    }
  }
}

Gemini CLI

gemini mcp add colab-exec -- mcp-server-colab-exec

Tools

colab_execute

Execute inline Python code on a Colab GPU runtime.

Returns JSON with per-cell output, errors, and stderr.

colab_execute_file

Execute a local .py file on a Colab GPU runtime.

Security policy: file_path must be a .py file inside the current workspace (cwd).

colab_execute_notebook

Execute code and collect all generated artifacts (images, CSVs, models, etc.).

Artifacts are downloaded as a zip and extracted into output_dir. Zip members are validated before extraction to prevent path traversal and special-file writes.

💻 Usage Examples

Basic Usage

Check GPU availability:

colab_execute(code="import torch; print(torch.cuda.is_available()); print(torch.cuda.get_device_name(0))")

Run nvidia-smi:

colab_execute(code="import subprocess; print(subprocess.run(['nvidia-smi'], capture_output=True, text=True).stdout)")

Advanced Usage

Train a model and download weights:

colab_execute_notebook(
    code="import torch; model = torch.nn.Linear(10, 1); torch.save(model.state_dict(), '/tmp/model.pt')",
    output_dir="./outputs"
)

Authentication

On first use, the server opens a browser window for Google OAuth2 consent. The access token and refresh token are cached at ~/.config/colab-exec/token.json. Subsequent runs use the cached token and refresh it automatically.

The OAuth2 client credentials are the same ones used by the official Google Colab VS Code extension (google.colab@0.3.0). They are intentionally public.

Troubleshooting

• "GPU quota exceeded" — Colab has usage limits. Wait and retry, or use a different Google account.
• "Timed out creating kernel session" — The runtime took too long to start. Retry — Colab sometimes has delays during peak usage.
• "Authentication failed" — Delete ~/.config/colab-exec/token.json and re-authenticate.
• OAuth browser window doesn't open — Ensure you're running in an environment with a browser. For headless servers, authenticate on a machine with a browser first and copy the token file.

📄 License

MIT