ComfyUI plug-in based on DeepSeek-Harness

Key features

• comfyui_run - Submit a workflow in ComfyUI API format, or choose a built-in template, and return the generated media.
• Media loader imitating the ComfyUI LoadImage node, located at the top of the workflow page: in addition to images, the selection window also lists the video and audio files available for the ComfyUI load node (LoadVideo / LoadAudio).
• Classify runnable workflows with preset categories (Tusheng pictures / Vincent pictures / Vincent videos / Tusheng videos / Reference videos / Vincent audios / Reference audios) + custom labels.
• Two modessync (wait for and return media) and async (background task, use job_output to collect results - highly recommended for video generation).