biorxiv-search
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- COMMAND_EXECUTION (LOW): The skill instructions direct the agent to dynamically resolve the path to its own internal scripts using the
findcommand. This is used to ensure compatibility with various installation environments but involves executing commands from computed paths. - EXTERNAL_DOWNLOADS (LOW): The skill performs network requests to
api.valyu.ai. While this is the primary purpose of the skill (semantic search), the domain is not on the pre-approved whitelist. No sensitive data exfiltration was detected. - CREDENTIALS_UNSAFE (SAFE): The skill implements a setup flow to collect a user-provided API key. This follows standard CLI practices for configuration and does not include hardcoded secrets.
- Indirect Prompt Injection (LOW): The skill ingests full-text article content from external search results. While this creates a surface for indirect prompt injection, the output is structured in JSON, providing a clear boundary for the agent.
Audit Metadata