Fundamentals
The foundation. When a demo video is the right choice (vs. an explainer), the guiding principle—"capture first, animate later"—and the 100% local stack with no API key: agent-browser, HyperFrames, and Kokoro.
Capture of the actual app
The part most likely to go wrong if rushed. Drive the app with agent-browser/Playwright, describe steps in actions.json, target coordinates and bounding boxes, and handle scrolling, React-controlled inputs, and asynchronous multi-state flows.
Composition & Rendering
Where the capture becomes a video: browser frame, animated cursor that clicks the target, and highlight zoom; the dark premium house style; local TTS narration with Kokoro; and rendering in HyperFrames with lint, inspect, and the INEMA.CLUB CTA.