pubmed-search
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches data from well-known and trusted medical research databases, specifically NCBI PubMed (eutils.ncbi.nlm.nih.gov) and ClinicalTrials.gov (clinicaltrials.gov).
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from search results, presenting a standard indirect injection surface.
- Ingestion points: XML and JSON data retrieved from PubMed and ClinicalTrials.gov APIs.
- Boundary markers: Absent in the provided code snippets.
- Capability inventory: The skill utilizes network GET requests and data parsing; it lacks sensitive file access, file-writing, or shell execution capabilities.
- Sanitization: Extracted fields (such as PMID, title, and abstract) are parsed into structured formats, reducing the risk of malicious payload interpretation.
Audit Metadata