skills/freedomintelligence/openclaw-medical-skills/tooluniverse-clinical-trial-matching/Gen Agent Trust Hub
tooluniverse-clinical-trial-matching
Pass
Audited by Gen Agent Trust Hub on Jul 30, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill retrieves information from several well-known and authoritative medical and scientific services, including ClinicalTrials.gov, OpenTargets, CIViC, ChEMBL, and the FDA. These services are used according to their intended purpose for clinical research and do not represent a security risk.
- [PROMPT_INJECTION]: The skill processes patient profile data (cancer types, biomarkers, prior treatments), creating an indirect prompt injection surface.
- Ingestion points: Patient characteristics provided as input to the skill as described in SKILL.md.
- Boundary markers: The skill relies on a structured multi-phase workflow to process data, though it does not explicitly define delimiters for untrusted content.
- Capability inventory: The skill invokes search and retrieval tools for medical databases (ClinicalTrials.gov, OpenTargets, CIViC, etc.) via the ToolUniverse interface.
- Sanitization: Patient inputs are parsed using regular expressions and normalized against standard identifiers like Ensembl IDs and EFO terms during the standardization phase.
- [SAFE]: Analysis of the skill instructions and the accompanying test script (test_skill.py) revealed no signs of credential exposure, base64 obfuscation, or persistence mechanisms. The skill operates within a restricted execution environment using authorized tools.
Audit Metadata