video-watermark

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

• Third-party content exposure detected (high risk: 0.80). The skill's Quick Start and multiple examples require passing arbitrary external URLs via the "image_urls" parameter (e.g., https://example.com/my-video.mp4) which the each::sense API will fetch and process, meaning the agent ingests untrusted third‑party media as part of its workflow and could have its processing/decisions influenced by that content.