scvi-tools
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADS
Full Analysis
- External Data and Model Retrieval: The skill provides instructions for downloading pre-trained genomic models from the Hugging Face Model Hub and installing necessary dependencies from official package registries like PyPI. These operations utilize established libraries such as
huggingface_hubandpip, which are standard tools for managing machine learning assets and software environments. - Scientific Data Processing: The provided Python scripts are designed to process specialized genomic data formats (.h5ad and .h5mu). The analysis confirms that these scripts perform expected data science operations—such as quality control filtering, dimensionality reduction, and differential expression analysis—without evidence of unauthorized data access or exfiltration patterns.
- Environment Configuration: The skill includes clear documentation for setting up computational environments with GPU support. It references official installation channels for PyTorch and other bioinformatics libraries (e.g., Scanpy, MuData), following best practices for reproducibility and security in scientific workflows.
Audit Metadata