seedance-ark-experience

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

• Third-party content exposure detected (high risk: 0.80). The skill instructs users to upload and reference arbitrary user-provided images, videos, and audio (e.g., "@图片1", "@视频1", "@音频1") which the agent ingests and is expected to read/interpret as part of prompt construction, exposing it to untrusted third-party content via the uploaded materials.