deep-research

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

• Third-party content exposure detected (high risk: 0.90). This skill explicitly instructs child processes to perform networked retrievals from public sources—prioritizing installed skills, then MCP tools (firecrawl, tavily), and if necessary curl/wget—to fetch news, webpages, datasets and cache them under .research//raw/, so the agent will read and interpret untrusted third‑party content.