vox-director
worthwhileTurn a topic into a finished Vox-style explainer video with automated collage keyframes, animation, voice-over, music, and captions.
What it is
A multi-step agent skill that transforms a one-line topic into a finished Vox-style paper-collage explainer video. It uses Atlas Cloud APIs for image generation (nano-banana-2), video animation (gemini-omni-flash), narration (xai/tts), and music (minimax/music-2.6), plus ffmpeg for assembly. Human approval gates control the beat map and visual style.
How it differs from vanilla Claude
A vanilla Claude agent can call the same APIs individually but lacks the curated prompt structure for the collage look, the multi-stage orchestration, and the embedded human-in-the-loop gates. This skill bakes in years of trial-and-error for prompt engineering and asset timing.
Skill, plugin, or workflow shift?
Integration as a skill is appropriate because it's a structured workflow for a coding agent to execute—not a library or standalone app. The AGENTS.md and SKILL.md define the exact steps, and the agent runs scripts under user supervision.
Devil's advocate — is this just complexity?
A vanilla capable agent with access to the same APIs could replicate this pipeline with a few hundred lines of Python and some prompt iteration. The 'collage' look is essentially a prompt style, not a technical breakthrough. The human gates add overhead that a fully custom script could skip. Vendor lock-in to Atlas Cloud and dependency on specific model IDs further limit portability. Any technical user could build this themselves in a day or two.
What would make it better
Add local model fallbacks (e.g., Stable Diffusion, Coqui TTS) to remove the Atlas Cloud dependency. Allow full customization of collage elements (textures, cut-out styles) beyond the 8 themes. Provide an option to export individual assets for manual editing. Offer a one-command install without requiring a coding agent.
The honest case for it
For content creators who lack the time or expertise to orchestrate multiple AI models and craft specific prompt styles, this skill provides a turnkey pipeline that consistently produces visually appealing explainer videos. The human approval gates offer creative control while automating the heavy lifting. It removes weeks of experimentation and API integration work.
Who it's for
Audience fit
Depth and leverage for a technical engineer who wants to understand it and level up their workflow — not just offload work.
Value for someone who wants a more capable tool without the technical depth — accessible, does-it-for-you.
Vibe coders get a one-shot video generator without deep technical overhead; engineers benefit from the structured pipeline but the value lies in convenience, not novel architecture.