tooluniverse-metabolomics-analysis
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted metabolomics data (peak tables and spectra) through file loading functions, which serves as a potential entry point for indirect instructions embedded in data fields or metadata.
- Ingestion points: The
load_metabolomics_datafunction incode_examples.mdreads user-supplied CSV/TSV files using pandas. - Boundary markers: There are no explicit delimiters or instructions to the agent to treat input data as non-executable content or to ignore embedded instructions.
- Capability inventory: The skill possesses capabilities to call external bioinformatics tools (
MetabolomicsWorkbench_search_by_mzandMetaboAnalyst_pathway_enrichment) using the ToolUniverse interface, which could be targeted by malicious payloads in the input data. - Sanitization: The code examples do not demonstrate sanitization or validation of metabolite names or other string metadata before they are passed to tool functions.
Audit Metadata