openscreen

OpenScreen is a free, open-source desktop application for recording your screen and turning the capture into a polished product demo or walkthrough. It is the community-maintained continuation of the original OpenScreen project, which its creator archived after v1.5.0; development moved to this repository with his approval, keeping the same name and MIT license. It runs on macOS, Windows and Linux and is under active development, so rough edges and occasional breaking changes (including to the .openscreen project format and the CLI) should be expected. Core capabilities Recording: capture a specific window or the whole screen, plus microphone and system audio. A webcam overlay supports picture-in-picture, drag-to-position, mirroring and shape options. Editing: auto or manual zooms with adjustable depth, duration, easing and pixel-precise position, where auto-zoom follows the cursor. Cursor size, smoothing and click effects are configurable, with cursor themes and post-recording path smoothing. The timeline supports crop, trim and per-segment speed control, snapping guides and an audio waveform. Text, arrow and image annotations are available with text animation presets, along with wallpapers, solid colors, gradients or custom background images, and motion blur. Captions: automatic captions for voiceovers are transcribed on-device with no upload, so they work offline; the transcript is editable and can be cut from, with optional subtitle translation. AI editing assistant: you describe the edit in chat and it is applied to the timeline — cuts, zooms, speed ramps, annotations, camera framing. It is bring-your-own-key (Claude, OpenAI, Gemini, Mistral, OpenRouter, MiniMax or any OpenAI-compatible endpoint) and nothing is enabled by default. Export: MP4 or GIF in multiple aspect ratios and resolutions, rendered and encoded on the GPU (Metal on macOS, D3D11 on Windows, Vulkan on Linux) with an automatic CPU fallback. The interface is available in Arabic, Czech, English, Spanish, French, German, Italian, Japanese, Korean, Portuguese (Brazil), Russian, Turkish, Vietnamese, Simplified Chinese and Traditional Chinese. Command-line interface A headless CLI targets scripts, CI and AI coding agents: it can record the screen, edit the .openscreen project JSON programmatically (zooms, annotations, trims) and render MP4/GIF through the full export pipeline, with no visible windows and NDJSON output when passed --json. Installation and platform notes Windows users are pointed to the Microsoft Store (also installable via winget), with a standalone .exe as an alternative; the .exe is not code-signed, so SmartScreen warns about an unknown publisher. macOS builds from 1.9.0 onward are signed and notarized, and require Screen Recording and Accessibility permissions; macOS 15 and later periodically re-asks for screen-recording permission, and permissions do not carry over from pre-1.9.0 builds. Linux packages are published as .deb, .rpm, .pacman and .AppImage, plus a Nix flake with NixOS and Home Manager modules. System requirements list Windows 1903+ with Intel 8th Gen / AMD Ryzen 2000 or newer minimum, macOS 13 (Ventura) or later, and 8 GB RAM minimum (16 GB recommended). All three platforms record through a native capture pipeline (ScreenCaptureKit, Windows Graphics Capture, PipeWire via the ScreenCast portal), with a browser-capture fallback on Linux. System audio works on macOS and Windows out of the box; on Linux it needs PipeWire. Click effects on Linux require the user to be in the input group because Wayland exposes no portal for mouse buttons. Community and license The project is community-driven, with a Discord server, GitHub Issues for bugs and feature requests, a roadmap, and sponsorship via GitHub Sponsors or Ko-fi. It is licensed under the MIT License. The README advises downloading only from the official GitHub Releases linked from the repository, noting that third-party sites using the OpenScreen name are not affiliated.