p-video

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md examples explicitly show supplying arbitrary external URLs for image and audio inputs (e.g., "image": "https://your-image.jpg", "audio": "https://your-audio.mp3"), which the inference.sh app would fetch and ingest as part of generation, exposing the agent to untrusted third‑party content that can influence model behavior.