tooluniverse-multiomic-disease-characterization
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface due to its core functionality of ingesting and processing data from numerous external APIs. * Ingestion points: Untrusted data enters the agent context via multiple external tools and databases, including OpenTargets, GWAS Catalog, ClinVar, ExpressionAtlas, Human Protein Atlas, STRING, Reactome, KEGG, and WikiPathways. * Boundary markers: The instructions do not define any specific delimiters or markers (such as XML tags or unique string wrappers) to isolate data retrieved from these tools from the agent's internal logic. * Capability inventory: The skill allows the agent to create and write to markdown files and execute various computational biology tools across eight different analysis layers. * Sanitization: There are no instructions for validating, escaping, or sanitizing the content retrieved from external sources before it is used to populate the multi-omics report or influence downstream analysis steps.
Audit Metadata