biorxiv-database
Pass
Audited by Gen Agent Trust Hub on Jul 30, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill retrieves preprint metadata and downloads full-text PDF files from bioRxiv's official domains (api.biorxiv.org and www.biorxiv.org). These are well-known resources for life sciences research.
- [COMMAND_EXECUTION]: The skill includes a Python script (
scripts/biorxiv_search.py) intended to be executed via the command line to perform search queries and handle file downloads. - [PROMPT_INJECTION]: The skill processes external data (preprint abstracts and titles) which serves as a potential surface for indirect prompt injection. * Ingestion points: BioRxiv API responses in
scripts/biorxiv_search.py. * Boundary markers: Data is structured as JSON objects before being processed by the agent. * Capability inventory: Network requests via the requests library and local file writing for saving PDF documents. * Sanitization: The skill performs standard JSON formatting but does not implement specific sanitization to strip potential instructions from the retrieved paper abstracts.
Audit Metadata