What problem does it solve? Building ComfyUI workflows requires manually wiring nodes, checking parameter ranges, and looking up model files, which is slow and error-prone. This Skill turns a plain-language request into a valid LiteGraph-format workflow JSON that imports directly into ComfyUI. ## Core Features & Use Cases - 34 built-in templates: Covers txt2img, img2img, inpaint, ControlNet, LoRA, upscale, video generation (Wan 2.2, HunyuanVideo, LTXV, Mochi, Cosmos), Stable Audio, and Hunyuan3D across SD 1.5, SDXL, SD3, and FLUX. - 360+ node definitions on demand: A categorized node registry (42 files) ensures correct input types, widget values, and min/max constraints without loading everything into context. - Automatic model download metadata: Every generated workflow includes a top-level models array with HuggingFace URLs so ComfyUI prompts downloads for missing models on import. - LLM integration: Supports comfyui_LLM_party nodes (OpenAI, Claude, Gemini, DeepSeek, Ollama) for script-to-video and prompt-enhancement pipelines. - Use Case: Ask for a FLUX text-to-image plus Wan 2.2 image-to-video pipeline, receive a complete JSON file, drag it into ComfyUI, and run it after the auto-prompted model downloads finish. ## Quick Start Ask the AI to generate a ComfyUI workflow for your task, for example by saying: create a Wan 2.2 image-to-video workflow with camera control at 720p.
