clinicaltrials-skill
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEPRIVILEGE_ESCALATIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- User-Specified File Write Path: The
_save_rawfunction inscripts/clinicaltrials_client.pywrites data to a path provided in theraw_output_pathfield. If the execution environment has broad permissions, this could potentially be used to write files to unintended directories. - External Data Interaction: The skill retrieves information from ClinicalTrials.gov (
https://clinicaltrials.gov/api/v2). This is a well-known service, and the data fetched is used to provide study summaries. - Indirect Prompt Injection Surface:
- Ingestion points: Data enters the skill via API responses in
scripts/clinicaltrials_client.py. - Boundary markers: No explicit delimiters are used to wrap external content.
- Capability inventory: The skill includes file-writing capabilities through
Path.write_text. - Sanitization: The
_compactutility function limits the depth and length of data returned to the agent to mitigate large-scale injection.
Audit Metadata