tooluniverse-clinical-trial-matching
Pass
Audited by Gen Agent Trust Hub on Mar 15, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill ingests and processes untrusted data from external sources, which creates a surface for indirect prompt injection. \n
- Ingestion points: The skill retrieves clinical trial summaries and medical abstracts through tools like 'get_clinical_trial_descriptions' and 'PubMed_search_articles'. \n
- Boundary markers: The agent is guided by a rigid multi-phase workflow, a quantitative scoring system, and a completeness checklist, which help maintain task focus and objective analysis. \n
- Capability inventory: The skill has the ability to query network APIs and write markdown-formatted reports to the filesystem. \n
- Sanitization: The skill employs regular expressions to extract structured biomarker data, but it does not specify a sanitization process for the general narrative text retrieved from external sources.
Audit Metadata