scanpy
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No security issues detected. The skill provides standard bioinformatics workflows using well-known, established libraries.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface through genomic data files and has the capability to write to the file system.
- Ingestion points: Data loading functions sc.read_h5ad, sc.read_10x_mtx, and sc.read_csv in scripts/qc_analysis.py, assets/analysis_template.py, and SKILL.md.
- Boundary markers: None present; input data is treated as valid scientific observations.
- Capability inventory: File reading, writing (adata.write, to_csv), and directory creation (os.makedirs) as seen in scripts/qc_analysis.py and assets/analysis_template.py.
- Sanitization: The skill performs standard bioinformatics quality control but does not sanitize potential natural language metadata within input files.
Audit Metadata