brand-review

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

• Third-party content exposure detected (high risk: 0.85). The skill ingests “Content to review” from outsider-authored sources at runtime—e.g., pasted text from other people, uploaded/downloaded files, or a user-supplied URL whose page text is fetched and then reviewed—so that free-form outsider prose can be placed into the agent’s LLM context for analysis.