literature-review

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md workflow (Phase 2: "Search & Discovery") explicitly instructs fetching from public third‑party sources—arXiv (via arxiv-search), Semantic Scholar API, Google Scholar web searches, and Connected Papers—and Phase 4 requires reading paper pdf_url full texts for extraction, so the agent ingests and interprets open web content that can materially influence its synthesis and next actions.