clinicaltrials-database-search
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes clinical trial data retrieved from an external API, which could theoretically contain instructions intended to influence the agent.
- Ingestion points: The skill retrieves data from
https://clinicaltrials.gov/api/v2viarequests.getas shown in the code examples inSKILL.md. - Boundary markers: The instructions do not define explicit boundary markers or "ignore" warnings when processing the
briefTitle,eligibilityCriteria, oroutcomeMeasuresfields. - Capability inventory: The skill includes file writing capabilities, specifically for exporting results to
heart_disease_trials.csvas shown in Core API Section 7. - Sanitization: The skill uses standard JSON parsing but does not implement additional sanitization or filtering for CommonMark Markdown content retrieved from the API before presenting it to the agent context.
Audit Metadata