literature-search
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes a local bash wrapper (scripts/search) to execute its logic. It includes specific instructions for the agent to find the script path using the find command.
- [EXTERNAL_DOWNLOADS]: Connects to api.valyu.ai to perform semantic searches and retrieve full-text scientific content from PubMed, arXiv, bioRxiv, and medRxiv.
- [PROMPT_INJECTION]: The skill ingests large amounts of external scientific text, which represents an indirect prompt injection surface if the fetched papers contain adversarial instructions. No explicit safety delimiters are provided in the ingestion logic.
- [SAFE]: The credential handling process is transparent, asking the user for an API key and providing a setup command for storage, which aligns with standard security practices for external API integration.
Audit Metadata