Answering Research Questions

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs the agent to search PubMed, fetch/save abstracts and full-text PDFs (via PubMed/Unpaywall) and traverse citations (e.g., Semantic Scholar), and to read and extract findings from those public third-party papers which directly influence scoring, queuing, and follow-up actions (see "Phase 3: Search Literature", "Phase 4: Evaluate Papers" and Unpaywall/download PDF steps), so untrusted external content can alter the agent's behavior.