gr-social-distill

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

• Third-party content exposure detected (high risk: 1.00). The skill's SKILL.md explicitly accepts "1 个已发布博客 URL" as input and requires the agent to read and rewrite that external published blog into platform-specific variants (see "输入" and the templates), meaning arbitrary public blog pages would be fetched/ingested and could thus supply untrusted instructions that influence the agent's outputs and actions.