anndata
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill serves as a legitimate technical guide for the
anndatalibrary, a core component of the scverse ecosystem for bioinformatics.\n- [EXTERNAL_DOWNLOADS]: Contains examples for downloading datasets from remote URLs. The provided code includes a security validation step that checks the URL scheme and hostname against a trusted list before proceeding with the download.\n- [INDIRECT_PROMPT_INJECTION]: The skill facilitates reading external data files (H5AD, Zarr, CSV, etc.). While these files are ingestion points for external data, the risk is handled via standard data structures, and the documentation focuses on scientific data analysis.\n - Ingestion points: File reading functions such as
ad.read_h5ad,read_csv, andread_mtxinSKILL.mdandreferences/io_operations.md.\n - Boundary markers: Not explicitly used in the provided code snippets for data ingestion.\n
- Capability inventory: The skill uses
Bash,Read,Write, andEdittools.\n - Sanitization: Standard scientific data parsing is used without additional input sanitization layers.
Audit Metadata