in-depth-research

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly requires web search and web fetch capabilities and its SKILL.md "Protocol → Search" and "Prerequisites" sections (and references to forum mining like Reddit/HN and "web-scraper" in Works Well With) direct the agent to fetch and interpret open/public, user-generated third‑party content as part of decision-critical research, creating clear exposure to indirect prompt injection.