biomedical-search

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly instructs the agent to ask the user for their Valyu API key and to run commands like scripts/search setup <api-key> (and shows embedding the key in example code), which requires the LLM to receive and output secret values verbatim, creating an exfiltration risk.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The skill calls the Valyu API (see scripts/search.mjs and SKILL.md) to retrieve full-text content and URLs from public third-party sources (PubMed, bioRxiv/medRxiv, ClinicalTrials.gov, FDA drug labels), and those returned untrusted/public contents are consumed and exposed as part of the normal workflow for clinical/research decision-making, creating a clear avenue for indirect prompt injection.