Ask by voice or text, select any area of your screen, and let it click for you. Runs on free APIs like Gemini and Groq, or fully local models.
Always the newest release · All versions on GitHub
| Provider | Good for | Watch out |
|---|---|---|
| Gemini | Vision and audio | Small daily quota, and Pro is paid-only |
| Groq | Very fast text, generous free use | Text-only, so use transcribe mode for voice |
| OpenRouter | Many :free models | Daily caps on free models |
| Ollama / LM Studio | Unlimited, private | Depends on your hardware |
Free tiers change often, so check each provider’s current limits.
No. It works with free API tiers and local models. Free quotas are limited, so you can chain several providers.
Open ⚙ Settings → API keys. Each provider has its own hidden field, plus one for text-to-speech. Keys are stored in a file on your PC.
Most likely a free daily quota ran out. The overlay shows which provider answered and how much is left, and it falls through to the next provider in your list.
Your messages, plus a screenshot only when you use Select area, Summarize or Screen control. They go to the provider you chose.
Screen control can ask before every action, and a stop hotkey ends it immediately.
Yes. Settings → Display lets you pick the monitor, or follow your cursor.
Settings, memory and the encrypted vault are saved next to the app on your computer.
Unsigned apps that use global hotkeys and mouse automation are sometimes flagged by SmartScreen or antivirus. If it came from the official download link, choose More info, then Run anyway.