What problem does it solve? Turning an approved narration script into a finished explainer video requires coordinating TTS voiceover, forced caption alignment, visual assets, semantic motion planning, rendering, and quality gates; this Skill automates that entire local pipeline with reproducible, debuggable stages. ## Core Features & Use Cases - End-to-end video production: Converts an approved spoken-script JSON into timed audio, word-level captions, visual asset manifests, semantic motion plans, rendered video, and delivery packages with publishing copy. - Externalized configuration: Keeps voice assets, brand, CTA, canvas, and machine runtime in a workspace-level .agent-video/ config root, frozen per project via a resolved profile with SHA validation. - Quality gates and debugging: Enforces staged QC gates (audio boundaries, voice stability, layout occlusion, AV alignment) with a single run_gates.py entry point and a step-ordered debug checklist. - Use Case: A creator has a confirmed Chinese narration script and wants a subtitled, animated 1080p video with consistent cloned voice; the pipeline generates the master audio, aligns captions, plans motion, renders, and validates the final MP4. ## Quick Start Use agent-video-pipeline to process my approved narration script at /path/to/spoken-script.json in workspace /path/to/workspace with project directory /path/to/workspace/my-video, checking the external config and resolved profile first.
