copy-editing

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

• Third-party content exposure detected (high risk: 1.00). The skill's required workflow explicitly tells the agent to consult and incorporate questions from public forums and Reddit ("Step 4: ... (in forums, on Reddit, in 'people also ask' boxes)"), and it accepts linked/pasted copy to review, which means the agent will fetch and interpret untrusted third-party, user-generated content that can change its editing decisions.