comment-strategy

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly instructs the agent to engage with and respond to Xiaohongshu post comments—public, user-generated content—requiring the agent to read, categorize, and act on those third‑party comments (e.g., "Use when engaging with post comments" and the many reply/flow examples), so untrusted inputs can materially influence actions.