tooluniverse-clinical-guidelines
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute Python code via a Bash tool to perform statistical analysis, statistics, and data processing on retrieved clinical data.
- [DYNAMIC_EXECUTION]: The instructions mandate the runtime generation and execution of Python scripts using standard libraries such as pandas, scipy, statsmodels, and matplotlib to analyze clinical guideline data as part of its 'COMPUTE, DON'T DESCRIBE' directive.
- [INDIRECT_PROMPT_INJECTION]: The skill aggregates clinical recommendations from numerous external authoritative sources, including NICE, WHO, GIN, and various medical societies, which constitutes an attack surface for indirect prompt injection. Ingestion points include search results and full-text retrieval from these external databases. While a synthesis template provides structure for output, the skill lacks explicit instructions for sanitizing or filtering these external contents to prevent potential instruction injection.
Audit Metadata