paidf-anomalygen
Pass
Audited by Gen Agent Trust Hub on Jun 30, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a professional machine learning workflow for manufacturing defect simulation. It adheres to security best practices by utilizing environment variables for authentication (HF_TOKEN) and recommending non-root Docker execution with proper user ID mapping.
- [EXTERNAL_DOWNLOADS]: Fetches model weights and datasets from official and well-known sources including Hugging Face (NVIDIA, Google, Meta/Facebook organizations), NGC, and public research repositories. These downloads are verified as legitimate dependencies for the model's functionality.
- [COMMAND_EXECUTION]: Orchestrates the pipeline through documented utility scripts and standard ML tools (torchrun, python modules). It includes an internal 'guard' preflight check to verify environment integrity before execution.
- [CREDENTIALS_UNSAFE]: Secret management is handled correctly by instructing users to use environment variables for Hugging Face tokens rather than hardcoding credentials.
Audit Metadata