manage-submission

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

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime workflow ingests “decision letter and the reviews verbatim” into papers/<slug>/correspondence/ (“Decision received” event), and those reviews/letters are outsider-authored free text from the venue/reviewers, which the agent then reads/summarizes and uses to draft responses—an indirect prompt-injection risk.