academic-research

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

• Third-party content exposure detected (high risk: 0.90). The skill's required L4 pipeline explicitly harvests and scrapes open/public third‑party content (e.g., Stage 1b Exa web_search_advanced_exa calls for research papers, Researcher A's arXiv/Google Scholar scraping, and Researcher B/C targeting community sites like habr.com, cyberleninka.ru and other web sources) and then reads and synthesizes those external texts into the report and downstream agent decisions, creating a clear channel for untrusted user‑generated content to influence actions.