Summary

Alisa0808/vox-director automates the creation of Vox-style paper-collage explainer or ad videos from a single-line topic. Designed as an agent skill for coding agents like Claude Code or Codex, it streamlines the entire video production pipeline, from scriptwriting and collage keyframe generation to motion, voice-over, music, and captions, delivering a final MP4 output. This offers a significant benefit by abstracting complex video editing and AI model orchestration into a simple, topic-driven process.

How It Works

The project orchestrates video generation through a defined workflow driven by a beats.json file, which maps narrative beats to specific stages. The core approach emphasizes two critical elements: the visual style is baked into the initial image generation step, where each beat becomes a rich paper-collage poster containing all visual DNA (torn paper, cut-outs, headlines). Subsequently, motion is applied to these static posters, either via an AI image-to-video model for a "living poster" effect or a local engine for piece-by-piece animation. Two human decision gates—approving the beat map and selecting the visual theme—ensure user control within the automated pipeline.

Quick Start & Requirements

• Installation: Clone the repository (git clone https://github.com/Alisa0808/vox-director.git ~/.claude/skills/vox-director) or install via a .skill file.
• Prerequisites: A coding agent (e.g., Claude Code, Codex), an Atlas Cloud API key (obtainable at atlascloud.ai/console/api-keys), ffmpeg and ffprobe (installable via brew install ffmpeg), and Python 3 with the Pillow library (pip install pillow).
• Setup: Primarily involves setting the ATLASCLOUD_API_KEY environment variable.
• Documentation: Links to SKILL.md, AGENTS.md, and references/prompt-guide.md are available within the repository.

Highlighted Details

• End-to-end video generation automated on Atlas Cloud API and local ffmpeg.
• Features two distinct human decision gates for narrative and visual theme approval.
• Supports AI-driven "living poster" animation and an advanced local engine for element-level motion control.
• Integrates multiple AI models for keyframing, animation, narration (xai/tts), and music (minimax).

Maintenance & Community

The repository is maintained by Alisa0808. Specific details regarding community channels, active contributors, or sponsorships are not detailed in the provided README.

Licensing & Compatibility

The project is released under the MIT License. © 2026 Atlas Cloud. This license generally permits commercial use and integration into closed-source projects.

Limitations & Caveats

The system relies heavily on the Atlas Cloud API and specific AI models, whose availability and performance are external dependencies. While an advanced local motion engine exists, the default AI animation path may have content restrictions. The README does not detail known bugs or specific unsupported platforms.