tooluniverse-metabolomics-analysis
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [SAFE]: The skill facilitates standard bioinformatics workflows for metabolomics data processing. No malicious patterns or unauthorized operations were detected.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted metabolomics data from peak tables and raw spectra, representing a potential surface for indirect prompt injection if malicious instructions were embedded in metabolite names or metadata.
- Ingestion points:
SKILL.md(Phase 1load_metabolomics_dataandidentify_metabolitesfunctions). - Boundary markers: None defined in the code snippets.
- Capability inventory: File reading (pandas), plotting (matplotlib), and specific analytical tool calls via the
tooluniverseframework. - Sanitization: Not explicitly shown for metabolite identifiers passed to database query tools.
- [EXTERNAL_DOWNLOADS]: The skill references and queries established scientific databases including HMDB, KEGG Compound, PubChem, and Reactome. These are well-known, trusted resources within the metabolomics research community.
Audit Metadata