scvi-tools
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION] (SAFE): No attempts to bypass safety filters or override agent behavior were found in the skill instructions or metadata.
- [EXTERNAL_DOWNLOADS] (SAFE): The skill installs reputable scientific packages (scvi-tools, scanpy) from standard package registries; downloads of example datasets are from trusted library maintainers.
- [DATA_EXFILTRATION] (SAFE): No suspicious data access or exfiltration patterns were identified; operations focus on local scientific data processing.
- [DYNAMIC_EXECUTION] (SAFE): While the skill includes model loading functionality, this is a standard requirement for its machine learning workflow and is used as intended for scientific analysis.
- [INDIRECT_PROMPT_INJECTION] (SAFE): The skill processes genomic data files (h5ad, peak matrices), which is its primary intended use case, and does not exhibit high-risk capabilities associated with data poisoning.
Audit Metadata