bio-proteomics-data-import
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill instructions focus on valid bioinformatics data processing tasks using established scientific libraries like pandas and pyOpenMS for handling proteomics files.
- [SAFE]: The provided code examples demonstrate safe data handling practices, including input sanitization by filtering out decoy and contaminant protein entries.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data files, creating a theoretical attack surface for indirect prompt injection. Ingestion points: Uses pd.read_csv, pd.read_parquet, and pyopenms.MzMLFile().load to ingest external proteomics data. Boundary markers: No explicit prompt boundary markers or ignore instructions directives are used. Capability inventory: The skill performs local data transformations using pandas and numpy; no subprocess calls, arbitrary command execution, or network operations are present. Sanitization: The skill includes logic to sanitize data by filtering out search engine bookkeeping rows (decoys, contaminants).
Audit Metadata