sn-deep-research

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md explicitly requires an initial web_search probe and mandates multi-source evidence collection and cross‑validation during "分维度取证" and "质量守门", which means the agent will fetch and interpret public/untrusted web content (search results/pages) that can materially influence planning, synthesis, and report generation.