What is video-growth-producer?

This is a short-video production skill for Codex, an AI coding assistant. You give it a piece of text or a script, and it works through a series of production steps to generate a vertical short-form video along with a cover image. The idea is that you can also train it on your own content style, so that in future sessions you can say something like "make me a video" and it produces one matching your channel's established direction. The production pipeline covers topic selection, scripting, visual planning, AI-generated image assets for scenes, subtitle timing, voiceover (either AI-generated or provided by the user), video rendering, cover art, and a pre-publication quality check. The actual video rendering is handled by Remotion, a code-based video framework. The output format is vertical at 1080 by 1920 pixels, which matches the standard short-form video format used on most mobile-first platforms. The tool has two modes. The default is called strict mode, which is intended for publication-quality output. This mode requires access to an image generation model to produce the visual scenes, cover backgrounds, and key explanation frames. Without image generation capability, strict mode will not run, because the visual quality depends on those AI-generated assets. The second mode, called remotion-only preview, generates a draft using text cards, logos, and basic animations, without requiring image generation. It is clearly labeled as a preview, not a finished video. Your content preferences, winning patterns, forbidden patterns, and visual style are stored in local profile files that you update over time. These files are not committed to the public repository, so personal account data and voice samples stay private. Installation is a single git clone command that places the skill in the correct directory for Codex to discover it. The project is licensed under MIT.