A skill for your AI agent. Give it a topic and it writes the script, builds a free local voice-over, finds cleared pictures and renders a narrated explainer: 16:9 for YouTube, 9:16 for Shorts. The demo was rendered by the kit, not edited after.

The look

• A real picture on screen almost all the time. A new one with every line.
• 1 to 5 word labels instead of text slides.
• A paper canvas, bold lowercase titles, and a morph from title to picture at each chapter.
• A red ✕ over the myth, split screens, grids, a struck-through word.

In the zip

• SKILL.md: a 14-step workflow from topic to checked render.
• The engine: 7 shot types, 16:9 and 9:16 from one shot list.
• Voice builder: Kokoro, runs offline, commercial use allowed. Cuts land on the words.
• Picture tools: Wikimedia Commons search that keeps only public domain, CC0 and CC BY, with a credit file. Share-alike pictures are blocked.
• Render, music mix to -16 LUFS, a thumbnail template, QA scripts.
• A working demo, 7 reference docs, two example videos, 13 stills.

You need

• Node 20+, ffmpeg, Python 3.10 to 3.12, Chrome. About 1 GB of free disk.
• An AI coding agent that runs shell commands and reads images, such as Claude Code.
• No API keys, accounts or video editor.

Good to know

• Tested on Apple Silicon macOS.
• The 36 s demo renders in about two minutes.
• Uses HyperFrames and GSAP (own free license; read it before commercial use). Fonts: SIL OFL.