Wan2.1/2.2 I2V ComfyUI

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

• Third-party content exposure detected (high risk: 0.90). The skill's required workflow (SKILL.md and references/comfyui-setup.md) explicitly instructs fetching models and code from public third-party sources (e.g., huggingface-cli downloads from HuggingFace repositories like Wan-Video/QuantStack and git clone of ComfyUI-GGUF from GitHub) and then loading those models/nodes into ComfyUI at runtime, meaning untrusted external content is ingested and can materially alter tool behavior.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill explicitly instructs cloning and installing external code that will run as ComfyUI custom nodes (e.g., https://github.com/city96/ComfyUI-GGUF and https://github.com/comfyanonymous/ComfyUI.git), which fetches remote code that is executed at runtime and is a required dependency, creating a clear remote-code execution risk.