ai-avatar-video

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

• Third-party content exposure detected (high risk: 0.80). The SKILL.md explicitly instructs running infsh apps with arbitrary external URLs (e.g., "image_url", "audio_url", "video_url" such as "https://portrait.jpg" or "https://video.mp4"), meaning the skill fetches and ingests open/public user-provided media (images/audio/video) that will be read/transcribed and used to drive downstream actions, enabling indirect prompt injection from untrusted third-party content.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The README instructs users at runtime to run "curl -fsSL https://cli.inference.sh | sh", which fetches and immediately executes remote shell code from https://cli.inference.sh (and then downloads binaries from dist.inference.sh), so the skill depends on and executes external code fetched at runtime.