zhangxuefeng-perspective

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

• Third-party content exposure detected (high risk: 0.90). The SKILL.md Agentic Protocol (Step 2) explicitly requires using tools like “WebSearch等” to fetch and ingest public third‑party content (e.g., employment data, school rankings, 校友反馈, 求职论坛) and says the agent must base its judgments on that research, which clearly exposes it to untrusted web content that can influence actions.