anndata
Pass
Audited by Gen Agent Trust Hub on Apr 9, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill serves as a documentation resource for the official anndata Python package. All code examples demonstrate standard scientific data processing tasks, such as creating, subsetting, and transforming annotated matrices.
- [EXTERNAL_DOWNLOADS]: The skill includes instructions for downloading datasets from remote URLs and cloud storage (e.g., S3) using well-known libraries like urllib and fsspec. These patterns are expected for accessing remote research data and do not target suspicious sources.
- [DATA_EXFILTRATION]: The skill demonstrates reading and writing local data files as part of its primary data management functionality. These operations are limited to research data formats and do not involve unauthorized access to sensitive system information or credentials.
- [PROMPT_INJECTION]: The skill ingests data from various external file formats (CSV, Loom, 10X), creating a potential attack surface for indirect prompt injection. However, this is an inherent property of data processing tools, and no specific exploits or instruction-following triggers were identified.
Audit Metadata