bioservices
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests large volumes of biological data from over 40 external web services (UniProt, KEGG, ChEMBL, etc.). Maliciously crafted data entries in these databases could theoretically contain instructions intended to influence the agent's behavior.
- Ingestion points: Data is fetched via the
bioserviceslibrary in all included scripts (protein_analysis_workflow.py,pathway_analysis.py,compound_cross_reference.py,batch_id_converter.py). - Boundary markers: No specific delimiters or instructions to ignore embedded content are provided when the agent processes retrieved biological descriptions or metadata.
- Capability inventory: The skill possesses the ability to execute Python scripts, perform network operations to scientific APIs, and write results to the local filesystem.
- Sanitization: There is no evidence of filtering or sanitizing the retrieved biological data for potential natural language instructions before it enters the agent's context.
- [EXTERNAL_DOWNLOADS]: The skill relies on the
bioservicesPython package to interact with well-known bioinformatics platforms. These services are legitimate scientific resources. - [COMMAND_EXECUTION]: The provided Python scripts perform network requests and manage local data files (CSV, SIF). This is standard functionality for the skill's intended purpose.
Audit Metadata