v0.10.0

What's New

Features

• Visual workflow guidance system - Flashing highlights guide users through the recording → training → online workflow
  • Highlights Training tab after 2+ recordings
  • Auto-selects newly created datasets
  • Auto-loads trained models in Online protocol
• CLI entry point - Run MyoGestic directly with myogestic command after installation
• Package installable - Proper pip/uv installation support with package = true

Fixes

• Fixed asset path resolution to work from any working directory
• Hide default recording UI when no visual interface is selected

Security

• CVE-2025-3730 (Medium): Fixed PyTorch Improper Resource Shutdown vulnerability
• CVE-2025-2953 (Low): Fixed PyTorch local Denial of Service vulnerability
• Upgraded PyTorch from 2.6.0 to 2.10.0
• Switched CUDA index from cu124 to cu126

Dependencies

• Added vispy>=0.16.0
• Updated torch>=2.8.0, torchvision>=0.23.0, torchaudio>=2.8.0