Jev Voice
Floating macOS bar for voice or text commands where Jev picks the next action from live accessibility controls in a loop.
Overview
Jev Voice (jev-cua) is a small floating bar for macOS that takes voice or text commands. Jev picks the next action from the Mac's live controls and installed apps, the app executes it, and after each action a fresh observation feeds another Jev decision until the request is done. Commands queue while work is in progress, text commands work with the mic off, and the bar stays available across apps and full-screen Spaces.
The demo shows opening Chrome, searching, using Calculator's controls to compute 90 plus 7, and triggering Photo Booth's shutter; the README notes this is a demonstration, not a benchmark or guarantee. A Jev activity view in Settings shows the model's inputs, choices, errors and observed results. There is no confirmation queue; a stop button or cancel task interrupts execution. It requires an Apple silicon Mac on macOS 14+, Xcode command-line tools and your own funded OpenRouter or TypeSafe key, and has a macOS CI workflow.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which live accessibility control or app action is the next step toward the request? | — |
How it works
- User speaks or types a command
- Observe live accessibility controls and installed apps
- Jev picks the next action; app executes
- Re-observe and repeat until done; stop button interrupts
Strengths
- Multi-step tasks through an observe-act-verify loop
- Activity log of model inputs and choices
- Works with TypeSafe or OpenRouter keys
Caveats
- No confirmation step before actions
- Demo is not a performance benchmark
- Apple silicon and macOS 14+ only
At a glance
- Type
- Project
- Status
- prototype
- Language
- Swift
- Runtime
- macOS app
- Has code
- yes
- Has tests
- yes
Use cases and integrations
How curated lists describe it
Voice and text control for macOS. One floating bar, live UI action selection with Jev, and a continuous observe–act–verify loop.
hellogumbo/awesome-jevFloating bar for macOS that takes a voice or text command and picks the next action from the Mac's live accessibility controls, looping until the goal is met.
MrJev/awesome-jev
Frequently asked questions
Can it do multi-step tasks?
Yes, each action is followed by a new observation and another Jev decision.
Does it ask before acting?
No; there is no confirmation queue, but a stop button and cancel task interrupt execution.
Which keys work?
Your own funded OpenRouter or TypeSafe API key.
Related projects
macOS menu-bar Swift app that turns on-device transcribed speech into Jev typed decisions and macOS actions, confirming hard-to-undo steps.
AppbetaSwiftVoice assistant daemon for Hyprland Linux desktops that acts through tool modules, with Jev gating which requests and actions proceed.
ProjectbetaPythonTalk to your Mac: local whisper.cpp transcribes, one Jev request picks the action and typed arguments, and code executes macOS automation.
ProjectprototypePythonmacOS voice and typed computer-use app where Jev picks the next on-screen action from the Accessibility tree, with no screenshots.
AppprototypeSwiftPre-alpha Rust and Tauri voice control layer for macOS that reads the accessibility tree; Jev is a fallback rather than the router.
ProjectprototypeRustElectron app for controlling Windows 10/11 by Turkish or English speech, using OpenAI Realtime, Jev decisions, UI Automation and Playwright.
ProjectprototypeTypeScript