search-specialist

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly requires searching and ingesting open/public third-party sources—e.g., the "Tools & Platforms" section lists "web scraping and content extraction tools" and examples reference Reddit, GitHub, LinkedIn and public websites—and instructs the agent to read and synthesize that user-generated/web content to produce recommendations, so untrusted third-party content can materially influence actions.