- Who it helps
- Developers creating repeatable product demonstrations.
- Availability
- Public MIT repository checked November 4. It documents Node 20+, ffmpeg and a browser or supported desktop display.
What remains editable
A scenario describes the actions to perform. The saved take contains frames and action timing; rendering can change styling, captions and voice without repeating those actions. Changing the demonstrated behavior still requires another capture.
Screenplay separates screen capture from presentation edits ↗
Before trying it
The recorder has no built-in AI model; an external agent supplies the reasoning. Setup downloads Chromium and registers agent tools. Voice generation may require a paid provider, or you can supply audio. Desktop capture needs permissions and a real display; Linux support specifies X11.
Screenplay separates screen capture from presentation edits ↗
What you can do
- Inspect the scenario and capture actions before running a demonstration. Open reference ↗
Where the evidence stops
README and creator profile read; no software installed or recording produced. Recording executes real actions, so use fictional data and a staging account. No demonstration video was watched.
Sources read. Primary sources read November 4, 2026 UTC. No installation, model inference or complete workflow executed.
Sources & further reading
GitHub · official
Screenplay separates screen capture from presentation edits ↗
Identifies the subject, documented capability or specific verification limit.
Georges Petrov · page checked 2026-11-04
41 stars · observed 2026-11-04 13:06 UTC
Coverage note: Independent attention and end-to-end execution remain unverified.
The record over time (1)
Dated source checks and published explanations. An observation date can differ from the original release date.
- · Added to Current
Screenplay lets an agent capture a software demonstration and then revise its presentation from the saved recording.