What problem does it solve? Automated screen recordings look like test runs: no cursor, instant clicks, soft text, and clips that miss their narration timing. This Skill turns storyboard actions into demo footage that reads as a person using the app, with measured durations that keep narration and picture in sync. ## Core Features & Use Cases - Two-pass capture workflow: Rehearse every storyboard section unrecorded to resolve targets and measure real timings, then record only after narration exists so each clip fills its time budget. - Multiple driver and recorder combinations: Drive web apps through the Playwright MCP, launch Electron apps per section, or attach over CDP to an already-running app; record with Playwright video, OBS over obs-websocket for native-pixel 60 fps capture, or a macOS ffmpeg screen-capture fallback. - Human-feeling motion: A synthetic cursor overlay with eased movement, click ripples, dwell timing, per-character typing at 5-7 characters per second, and eased scrolling, plus OBS pausing through long waits so agent runs become clean jump cuts. - Use Case: Record a product demo where section three shows an Electron app with native menus: attach over CDP to the running app, capture the window through OBS at native pixels, pause recording through a 30-second agent run, and hand the reconciler a measured clip duration. ## Quick Start Record the demo sections defined in demo/storyboard.json, rehearsing each one first and capturing the app window with OBS so the text stays sharp.
