tooluniverse-clinical-trial-protocol
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied trial concepts and feasibility data from other skills to generate document sections. This represents a potential surface for indirect prompt injection where instructions could be hidden in clinical descriptions. * Ingestion points: Clinical trial concepts and feasibility inputs (SKILL.md). * Boundary markers: No specific delimiters or safety instructions are used when interpolating external data into the protocol template. * Capability inventory: Access to informational tools (FDA, ClinicalTrials, PubMed) and execution of a local Python script. * Sanitization: No evidence of input validation or sanitization for the clinical data used.
- [DYNAMIC_EXECUTION]: The skill invokes a local script 'scripts/sample_size_calculator.py' to perform statistical sample size calculations. * Evidence: The agent is instructed to execute the script using python3 with parameters established in the workflow (SKILL.md). * Implementation: The script uses standard libraries and scipy for mathematical operations without using unsafe dynamic functions like eval or exec.
Audit Metadata