literature-search

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md explicitly directs the agent to fetch and synthesize content from open/public third-party sources (e.g., Semantic Scholar, arXiv, Tavily/web search, Exa, Gemini deep research with web grounding, and social platforms like Twitter/X and Reddit — see "Engine Selection" and "Step 3: Execute Search"), and that fetched untrusted/user-generated content is read and used to rank/synthesize results and drive next actions, creating clear risk of indirect prompt injection.