zysilm-ai/ai-video-producer-skill

Integrates with ComfyUI to orchestrate **WAN 2.2** and **Qwen Image Edit 2511** video generation models using GGUF quantization, enabling 10GB VRAM operation on consumer GPUs. Implements a philosophy-first workflow where users define visual style and narrative before generation, then iteratively refines through scene breakdown, asset creation, and two pipeline modes (Video-First for continuity, Keyframe-First for precise control). Designed as a Claude Code skill that automates server management and guides users through structured production steps with approval gates at each stage.