# Record and Edit a Screen Demo (Auto Screen Recorder) Source: [Record and Edit a Screen Demo (Auto Screen Recorder)]() Documentation index: [llms.txt]() Last updated: 2026-11-04 Auto Screen Recorder turns a software workflow into an edited, editable tutorial. You describe the steps to show. The Agent can record the demo itself, help you record it, or edit footage you already have. It then cuts out dead time and frames the recording on a background you choose. It suits product demos, onboarding videos, software tutorials, and internal training. Auto Screen Recorder runs in ChatCut Desktop on a Mac. Demo ![Video poster]() [Video]() From the Auto Screen Recorder Skill page: a recorded ChatCut session, framed on a background with step titles. ## When should I use Auto Screen Recorder? Use it when you need one of these: - **An agent-recorded demo.** The Agent operates the app on your computer, records what happens, and edits it. It needs desktop control and screen capture. - **A demo you record yourself.** You perform the steps. The Agent edits the take. - **Existing screen footage.** Reuse a recording already in your chat or project and edit it into a clear procedure. Do not use it to invent a product screen. The Agent records what the app actually shows. ## What do I need before I start? - **A clear brief.** Say what the tutorial should teach, who it is for, the key actions, and a target length if you have one. - **An app that is ready to demo.** Sign in and prepare sample data first. You resolve login and access. - **Operating-system permissions (agent-recorded demos).** macOS asks you to allow Screen Recording and Accessibility access. Only you can grant them. - **A way for the Agent to operate your screen.** Inside ChatCut Desktop, the built-in Agent and embedded Claude and Codex agents can. An external Claude or Codex app must supply its own computer control; connecting ChatCut does not add one. - **Credits for paid parts only.** Recording does not use credits. Agent messages and generated voiceover do. See [Credits and billing](https://pr-790-preview.chatcut.dev/docs/credits-policy). ## How do I run it? 1. Open **Skills** in the AI composer and select **Auto Screen Recorder**. In ChatCut Desktop you can also type `/` to search for it. Confirm it is attached to your draft. 2. Fill in the Skill’s short form (shown below), or describe the same choices in your message. Selecting the Skill alone does not start anything. 3. Approve the permissions macOS asks for. If you denied Screen Recording once, macOS will not ask again. Turn on ChatCut in System Settings › Privacy & Security › Screen & System Audio Recording (called “Screen Recording” before macOS 15), reopen ChatCut if asked, and tell the Agent to continue. 4. Let the Agent work. Within the first steps, you should see your chosen aspect ratio and background in the editor. Avoid switching windows while it records. 5. Review the finished timeline and ask for changes. If you record the demo yourself, the Agent shows a record button in the chat. You click it, choose the screen or window to share, and control the countdown and finish. In an external agent app, use the Timeline’s Record control instead: open its menu, choose **Screen**, and start. The Agent edits only after your take is saved. See [Record Voiceover, Camera, and Screen](https://pr-790-preview.chatcut.dev/docs/recording). ### What does the form ask me to choose? Try the form ![](https://d3kirzd750m18a.cloudfront.net/media/admin/0be53036-5be6-4148-8a43-3aa5bea7e16b-poster.webp) Loading… [Auto Screen Recorder]() [Open in ChatCut](https://app.chatcut.dev/skills/screen-demo-editing?from=landing) The form has two tabs. **General** covers what to record and how it is framed: - **What to demonstrate:** the steps to show, in your own words. For example: “In ChatCut, import a clip, split it on the timeline, then export.” - **Aspect ratio:** 16:9, 4:3, 9:16, or 1:1. **More settings** holds the window shape, padding, and corner radius. - **Background:** a wallpaper, or **View all** for more. A solid color or a blurred copy of the recording also works. - **AI extras:** None, AI voiceover, MG step callouts, or both. **Cursor** covers the pointer: the system pointer or a smooth, halo, or spotlight pointer, an optional click effect and click sound, and zoom (off by default, or 1.25x, 1.5x, or 1.8x). Select **Create tutorial** to send it. ## Prompts you can copy Prompt ```text Use Auto Screen Recorder. Record a demo of how to invite a teammate in our web app, for new customers. Use 16:9, add step callouts, and keep it silent. Do not show real email addresses. ``` Prompt ```text Edit @signup-recording into a short onboarding tutorial. Remove failed attempts and waiting time, zoom in around clicks, and add step callouts and voiceover. Tell me before any voiceover is generated. ``` Prompt ```text I will record the demo myself. Help me start the screen recording, then edit my take into a 9:16 walkthrough with a blurred background. ``` ## What does ChatCut do for me? **The Agent does these automatically:** - Checks your project, any footage, and what it can actually control and record. - Sets the ratio and background first, then rehearses uncertain actions off-camera. - Brings only the target window to the front. It does not minimize, hide, or move your other windows. It avoids risky shortcuts such as select-all and quit. - Chooses a window capture for a self-contained app and a display capture for menus, native dialogs, or multiple windows. - Checks the saved video, then edits with hard cuts only. It removes failed attempts and idle time and speeds up long loading waits. It adds no transitions. - Frames the UI so it stays readable, and adds callouts, pointer effects, zooms, and captions only as you chose. **The Agent asks you when needed:** the audience, scope, format, or narration choices that would change the result. **You provide or approve:** the brief, a ready-to-demo app, macOS permissions, any narration choice, and the final review. Choosing voiceover lets the picture wait for speech, up to about three seconds per step. Voiceover generation uses credits. ## What do I get? An editable ChatCut timeline. The background, recording, callouts, pointer effects, click sound, zooms, voiceover, and captions sit on separate tracks, so you can adjust each one. The original raw take is kept. A voiceover is written one clip per step and matched to the on-screen actions. Captions come from that narration. Ask for an export when you want a file. ## Limits and tips - If the Agent cannot control or capture your screen, it says which is missing and offers your own recording or existing footage. - A window capture records only that window. Separate pop-ups can be missed. The Agent switches to a display capture, or re-records the step. - If the target window keeps losing focus, the Agent stops and asks you to keep the desktop still. - Length is whatever the cut produces. To shorten it, the Agent asks which steps to drop. - A drawn pointer needs a current Desktop release. Otherwise the Agent records with the system pointer and tells you first. - Say what must stay private, and review the video before sharing. - When demonstrating ChatCut, use a separate clean project window and hide notifications. ## What related pages can I read? - [Skills and Skill Marketplace](https://pr-790-preview.chatcut.dev/docs/skills-overview) - [Browse all Skills](https://pr-790-preview.chatcut.dev/skills) - [Record Voiceover, Camera, and Screen](https://pr-790-preview.chatcut.dev/docs/recording) - [Edit a Desktop Project from Your Agent](https://pr-790-preview.chatcut.dev/docs/desktop-agent-workflows) - [Credits and billing](https://pr-790-preview.chatcut.dev/docs/credits-policy)