tooluniverse-clinical-trial-matching

Pass

Audited by Gen Agent Trust Hub on Mar 29, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill architecture is designed for precision medicine research and follows standard procedures for trial matching without malicious behavior.
  • [EXTERNAL_DOWNLOADS]: Accesses clinical and molecular data from well-known scientific sources such as ClinicalTrials.gov, OpenTargets, CIViC, and the FDA. These references are appropriate for the skill's purpose.
  • [COMMAND_EXECUTION]: Provides a test suite (test_skill.py) to ensure tool connectivity. The script utilizes the tooluniverse framework to simulate legitimate user requests.
  • [PROMPT_INJECTION]: Analyzed for potential indirect prompt injection via external trial descriptions. Ingestion point: ClinicalTrials.gov eligibility criteria (MATCHING_ALGORITHMS.md). Boundary markers: Absent. Capability inventory: Read-only medical tool calls and local report generation. Sanitization: Employs regular expressions to strictly extract biomarkers and disease terms, mitigating risk from unstructured data.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 29, 2026, 11:41 PM