screen-recordinglisted

# screen-recording Script-driven product capture — **iOS Simulator, website, or Figma prototype** — with auto-zoom and cursor navigation **derived from measured interaction times**, never hand-guessed. The synthetic cursor comes from the compositor, so it works identically on every surface. **The user's script must specify, per scene: start state, the interaction to show, the payoff to read, and the data that must exist.** If it doesn't, get those four fields per scene before capturing anything (the companion product-film skill ships a full fill-in template in its `references/writing-the-script.md`). **Read `references/capture-protocol.md` before rolling anything.** It has the staging checklist, the per-take protocol, and the derivation rule. **COMPANION for editing/compositing:** the [product-film](https://github.com/Malik1942/product-film) skill supplies the downstream tools (`diffscan.swift`/`diffscan2.swift` for measuring takes, `composite2.swift` for rendering zoom + cursor). This skill produces their inputs; capture itself works standalone. ## The loop 1. Parse the user's script → shot list (start state, actions, end state, readable moments). Seed any missing product data first. 2. Pick the capture surface (protocol §0: simulator / website / Figma proto — each has its drive+record pair) AND the presentation path by form factor: portrait/phone content → mobile path (mockup, dot cursor); landscape/desktop content → desktop path (`FILL=1` full-bleed, arrow cursor, **