act-or-decide

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

• Third-party content exposure detected (medium risk: 0.30). The skill’s required runtime reads outsider-authored free text indirectly via query-window, which returns task fields like summary, signal_url, and external presumably populated from user/community/vendor-submitted backlog items, and then uses that text for clustering/ranking and to craft actions/decisions without first selecting a specific trusted item.