Releases: joelburlin/mac-rec
Release list
mac-rec 0.5.0 — settings, updates, one-line install
macOS 15+ · universal (Apple silicon + Intel) · 2.2 MB
Install
curl -fsSL https://screenrecording.dev/install.sh | bashInstalls to /Applications, clears the quarantine flag macOS puts on downloads, links the mac-rec CLI, and launches the app.
Downloading the .dmg directly? macOS will block it. This build is unnotarized (notarizing requires a paid Apple Developer account). Open System Settings → Privacy & Security, scroll down, click Open Anyway. On macOS 15 the old right-click → Open bypass no longer works. Or:
xattr -dr com.apple.quarantine /Applications/Mac-Rec.app
New
- Settings window (⌘,) — a real home for the ElevenLabs API key, with a live key test, voice picker, one-click whisper model download, and the recordings folder. No more CLI flags for setup.
- Check for Updates… in the menu, plus a quiet once-a-day check at launch that only speaks up when there's a new version (with skip-this-version).
config.jsonis nowchmod 600— it can hold an API key.
Fixed
- Documented the correct Gatekeeper flow for macOS 15. The previous instructions said right-click → Open, which Apple removed in Sequoia.
mac-rec 0.4.0 — first public build
macOS 15 (Sequoia) or later · universal (Apple silicon + Intel) · 2.2 MB
Download Mac-Rec-0.4.0.dmg below, drag it to Applications, then right-click → Open the first time.
⚠️ First launch needs a right-click. This build is open source and unnotarized (notarizing requires a paid Apple Developer ID), so macOS quarantines it on download. Right-click → Open → Open — once. If macOS still refuses:xattr -dr com.apple.quarantine /Applications/Mac-Rec.appPrefer to build it yourself?
git clone+./make-app.shtakes about a minute and skips Gatekeeper entirely.
On first record, macOS asks for Screen Recording (and Microphone, if you want narration). The app spawns the recorder daemon, so one grant covers the menu-bar UI, the mac-rec CLI, and any agent.
What's in it
- Live rewind — record into 2-second fragments; rewinding deletes trailing fragments, so it's instant and lossless
- Capture a display, a single window, or a dragged region, with system audio + microphone on separate tracks
- Captions — whisper.cpp with word timings, five burn-in templates (classic / boxed / bold / karaoke / minimal), rendered with CoreText so Hebrew and Arabic come out right
- AI re-narration — replace your recorded voice with an ElevenLabs voice, sentence-anchored so captions and picture stay in sync
- Mic that survives reality — auto-gain for quiet webcam mics, denoise, live mute, level meter on the floating pill
- Agent-drivable — every verb is a hotkey, a
--jsonCLI command, and a localhost REST call. Dropskills/mac-recinto~/.claude/skills/and an agent can record itself working.
Needs
ffmpeg for the finalize pipeline (brew install ffmpeg), whisper-cpp for captions, gcloud only for uploads.
Full docs: https://screenrecording.dev