pytorch-lightning
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides structural templates and documentation for deep learning workflows using PyTorch Lightning 2.6.6.
- [SAFE]: It explicitly recommends security best practices, such as using weights_only=True for checkpoint loading to prevent remote code execution via untrusted serialized data.
- [SAFE]: Documentation for external logging services (W&B, MLflow, Comet) correctly advises users to manage credentials via environment variables rather than hardcoding them in scripts.
- [SAFE]: All provided Python scripts (template_lightning_module.py, template_datamodule.py, quick_trainer_setup.py) contain standard, non-malicious code for model architecture definition, synthetic data generation, and trainer configuration.
- [SAFE]: External references are limited to official documentation domains, reputable GitHub organizations, and peer-reviewed academic archives.
Audit Metadata