video-analysis

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

• Third-party content exposure detected (high risk: 1.00). This skill explicitly accepts videos by URL ("启动检查点: 视频来源：本地文件 / URL" and the case JSON "url": "视频链接或本地路径") and uses analyzed content to drive a feedback loop that updates the prompt-matrix/agent behavior, so untrusted third‑party media could materially influence tool decisions.