joeseesun/capwise

What it is

Capwise is a local-first screen recording and editing studio for macOS. It combines screen, camera, and audio capture with a timeline editor, caption workspace, music and narration, cover generation, a teleprompter, and an AI editing assistant, so a demo or tutorial can go from recording to exported video without moving between separate tools.

It is built on Cap, with a native Rust + GPUI desktop interface and the recording, rendering, and export pipelines kept in the repository's crates (18)(8). The main use cases are product demos, software tutorials, course recording, and knowledge-sharing videos.

Install and start

The published package targets macOS 13.3+ on Apple Silicon. Download Capwise.dmg, drag Capwise to Applications, launch it, and grant Screen Recording permission; grant microphone or camera access only if needed. Record a video or import an existing clip, edit it, then click Export and choose output settings and a destination (3).

Core recording and editing do not require a Cap Cloud account. AI features are optional and need model services configured under Settings → AI Models: speech recognition, content generation, and cover generation are configured separately. You can also connect to an installed and logged-in ACP assistant. AI requests send the relevant audio or subtitle text to the chosen service, while projects and media stay local. API keys are stored in the macOS Keychain (3).

To explore without recording first, download the 24-second editable demo project, unzip it, and open the .cap file in Capwise to adjust titles, translation styles, and the timeline.

What the AI assistant does

You describe a goal (for example, polish these captions, translate them into French and show both languages, or lower the first background track by 6 dB). The assistant returns a proposed change set for preview; you review it, then apply it or undo it. Cover creation calls your configured image service directly and applies the result in the same workspace, with options to preview, edit, replace, or undo. Provider charges may apply (3).

Limits and licensing

General caption rewrites use at most 80 captions of context per request, so long videos must be processed in batches. Hiding filler words changes displayed captions only; removing the corresponding speech or video is a separate operation that must be explicitly requested and reviewed. AI output can still get proper nouns, tone, or translations wrong, so review before applying and exporting (3).

Only Apple Silicon builds are provided; Intel Mac, Windows, and Linux packages are not verified for this fork. The desktop app is primarily AGPLv3, with some recording and camera crates under MIT and other third-party exceptions documented in the license. The build instructions require macOS, Xcode Command Line Tools, Rust, Bun, and native media dependencies; a development build is not the same as a signed, notarized release (3).

HotRepo's take

This fork appeared on the Rust-language new-project snapshot for 2026-11-10 with 110 stars, 21 forks, and 5 stars gained in a one-hour growth window, so the observation is brief and does not by itself show a sustained trend. The repository is more substantial than a thin wrapper around Cap: it adds a separate GPUI editor, a review-and-apply assistant flow, bilingual caption styling, and an in-editor cover workflow, which is the part worth watching for people producing Mac tutorial or demo videos.