Python Remote Workspace ML Tools

Automatic compatibility validation for Python ML environments in remote workspaces. Detects and resolves version mismatches between TensorFlow, CUDA, cuDNN, and GPU drivers across SSH, container, and WSL sessions to prevent silent runtime failures and ensure stable remote development.

Features

• Validates TensorFlow, CUDA, and cuDNN version compatibility across remote development sessions (SSH, Dev Containers, WSL)
• Detects GPU driver versions and cross-references against TF requirements
• Validates toolkit completeness (nvcc, headers, runtime libraries, cuDNN)
• Python environment introspection (virtualenv, conda, system Python)
• Compatibility index with automatic updates and local caching
• Background scanning — runs silently without interrupting your workflow
• Persists results across sessions to minimize overhead on reconnect

Performance

The compatibility analysis engine and license verification are implemented as optimized WebAssembly modules compiled from Rust. This keeps the extension lightweight (under 100 KB total) and ensures that version matching across 200+ TF/CUDA/cuDNN combinations runs in milliseconds without blocking the editor.

The extension scans your environment once on activation and caches results locally. Subsequent sessions reuse the cache unless your installed packages change.

Requirements

• VS Code 1.75.0 or later
• Python workspace with TensorFlow or ML framework dependencies
• Remote development session recommended (SSH, Dev Containers, or WSL)

Configuration

You can optionally create a tf_cuda_config.json in your workspace root to override detected paths:

{
  "cudaHome": "/usr/local/cuda-12.2",
  "cudnnPath": "/usr/local/lib",
  "pythonPath": "/home/user/.venv/bin/python"
}

License & Privacy

This extension includes license verification for subscription features. On activation, it validates your license by sending a hashed machine identifier to our license server. No personal data, source code, or workspace contents are transmitted.

The verification runs once on activation and periodically when you switch workspaces or open Python files. If the license server is unreachable, the extension continues to function with basic compatibility detection.