luma-vision

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

• Third-party content exposure detected (low risk: 0.10). SKILL runtime path node <skill_dir>/scripts/vision.js <图片来源> <问题描述> ingests outsider-authored free text only as an explicitly provided prompt argument (second CLI arg) and then sends it to the vision API, rather than ingesting unchecked text from a feed/queue/search without selection.