ai-avatar-video

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.80). Multiple URLs in the list are non‑resolving/placeholder hosts that use file extensions as top‑level domains (e.g., https://portrait.jpg, https://speech.mp3), which are untrusted/invalid and could mask malicious content; official-looking domains (inference.sh, raw.githubusercontent.com, cloud.inference.sh) appear legitimate.

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

• Third-party content exposure detected (medium risk: 0.30). The runtime workflow in SKILL.md is a user-invoked inference.sh CLI that takes outsider-provided free text fields like voice_script or text as direct inputs to the apps (e.g., pruna/p-video-avatar and inworld/text-to-speech-2), so the agent may ingest attacker-authored free text.