biorxiv-database

Pass

Audited by Gen Agent Trust Hub on Jun 27, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches preprint metadata and PDF files from bioRxiv's official API and web endpoints (api.biorxiv.org and biorxiv.org).
  • [COMMAND_EXECUTION]: Provides and documents the use of a Python script (scripts/biorxiv_search.py) that performs network requests and local file operations to save PDFs.
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface due to its data processing model.
  • Ingestion points: scripts/biorxiv_search.py retrieves research titles and abstracts from the external bioRxiv API.
  • Boundary markers: No clear delimiters or instructions are used to separate the external data from the agent's internal instructions.
  • Capability inventory: The script can perform network GET requests and write files to the local disk.
  • Sanitization: The retrieved metadata is not validated or sanitized for potential injection patterns before being returned to the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 27, 2026, 11:42 AM
Security Audit — agent-trust-hub — biorxiv-database