video-link-extract

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

• Third-party content exposure detected (high risk: 0.75). Required runtime path: the script calls RedFox APIs and then returns data.text / data.stampSents[].textSeg from the API response into the agent’s output/LLM context; this text is derived from outsider-authored video content (public platforms like Douyin/Kuaishou/Xiaohongshu), so it can contain free-form injection payloads.