What problem does it solve? Short video generation often fails because prompts are vague, shots drift visually, and the wrong generation mode is chosen. This Skill enforces a script-first workflow: it drafts a shot-by-shot video script for user confirmation, selects the correct RedBox video mode, and only then calls the video generation API. ## Core Features & Use Cases - Script-First Workflow: Drafts a Markdown shot table (Time, Picture, Sound, Shot) with explicit duration and aspect ratio, and blocks generation until the user confirms. - Mode Routing: Maps requests to the official RedBox models: text-to-video (wan2.7-t2v-video), reference-guided (wan2.7-r2v-video), and first-last-frame (wan2.7-i2v-video). - Storyboard-First Planning: For multi-shot or continuity-sensitive videos, generates a core environment reference image and stable keyframes before video generation. - Video Project Packs: Organizes briefs, scripts, keyframes, clips, and outputs under media/video-projects/<id>/ for multi-shot or revision-heavy work. - Use Case: A creator asks for a 15-second product promo with three shots and a consistent character. The Skill drafts the shot table, asks whether storyboard keyframes should be generated first, and after confirmation calls app_cli video generate with the correct mode and reference assets. ## Quick Start Ask the assistant to create a short video from your idea or reference images, and it will draft a shot-by-shot script for your confirmation before generating with the RedBox video API.
