pydeseq2
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data files (CSV, TSV, H5AD) which could theoretically contain malicious instructions designed to influence the agent's behavior during analysis or reporting.
- Ingestion points: The skill uses pandas.read_csv() and anndata.read_h5ad() in SKILL.md and scripts/run_deseq2_analysis.py to ingest count matrices and metadata from local storage.
- Boundary markers: The provided code snippets and instructions do not include explicit prompt delimiters or 'ignore instructions' warnings when processing these data sources.
- Capability inventory: The skill uses file read/write operations for data and results, and utilizes the multiprocessing library for parallel computation via the DefaultInference class. No network exfiltration or arbitrary shell command execution capabilities were identified.
- Sanitization: The scripts/run_deseq2_analysis.py script performs basic data validation, including checking for negative count values and ensuring sample index consistency between datasets.
Audit Metadata