notion-infographic

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

• Third-party content exposure detected (high risk: 0.90). The skill's required workflow instructs sub-agents to perform web_search and web_fetch on public URLs (reference/subagent-prompt.md: "执行 2-4 次 web_search... 对最佳 2-3 个结果，使用 web_fetch 阅读全文") and writes those external Sources/URLs into research-notes which the Lead Agent must read and use to drive decisions (SKILL.md P1/P2), so untrusted third-party content from open websites can directly influence the agent's actions and outputs.