video-extend
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references model and LoRA weights hosted on Hugging Face repositories (e.g., RaphaelLiu/PusaV1 and Kijai/WanVideo_comfy). These are standard assets for video generation workflows.
- [COMMAND_EXECUTION]: Instructions include a recommendation to install the imageio-ffmpeg package via pip to resolve media processing errors, which is a standard procedure for configuring video handling environments.
- [DYNAMIC_EXECUTION]: The documentation mentions the use of torch.compile (via the WanVideoTorchCompileSettings node) to improve performance by compiling model kernels. This is a common and safe optimization in deep learning.
- [INDIRECT_PROMPT_INJECTION]: The skill describes a workflow that processes external video files through the VHS_LoadVideo node. * Ingestion points: Video file loading via VHS_LoadVideo in SKILL.md. * Boundary markers: Not explicitly defined within the video data stream. * Capability inventory: Includes file writing (VHS_VideoCombine) and system-level media processing (ffmpeg). * Sanitization: Standard binary video decoding is performed by the underlying libraries.
Audit Metadata