literature-review

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

• Third-party content exposure detected (high risk: 0.90). This skill's SKILL.md explicitly instructs the agent to search, fetch, and log content from public third-party sources (e.g., PubMed, arXiv, Semantic Scholar) and to screen/read abstracts and full texts as part of its required "Search and Log Evidence" and "Screen Sources" workflow, so untrusted third-party content could directly influence inclusion, synthesis, and next actions.