Yappy quickstart
From nothing to your first dictated sentence takes about two minutes, most of which is macOS asking permission.
1. Download and install
Get the notarized disk image from yappy.biz/download, open it, and drag Yappy to Applications. The direct URL for the current release is also available programmatically:
curl -s https://yappy.biz/api/v1/releases/latest | jq -r .download_url
Yappy is signed with a Developer ID and notarized by Apple, so Gatekeeper opens it without a right-click dance.
2. Launch it
Yappy lives in the menubar — look for the mic icon, not the Dock. There is no main window; Settings… from the menubar opens a native window with tabs for Dictation, Engine, Agent, and Files.
3. Grant Microphone
macOS prompts the first time you hold the dictation key. Allow it. Without the microphone there is nothing to transcribe.
4. Grant Accessibility
System Settings → Privacy & Security → Accessibility → enable Yappy.
This is the permission that lets Yappy see the global hotkey and insert text into whatever app you are using. macOS only re-reads the grant at launch, so quit and relaunch Yappy after enabling it.
5. Say something
Put your cursor in any text field. Hold Right ⌥ (Option), speak, release. Clean, punctuated text appears where the cursor was.
Double-tap the key instead of holding it to lock recording on for hands-free dictation; tap again to stop.
6. Optional — try Agent Mode
Hold Right ⌘ (Command) and ask for something: "reply to Sarah and say I'll be ten minutes late." The agent reads the screen, acts in the app you are in, and narrates each step in the notch. It stops before anything irreversible — a send, a purchase, a login — and waits for you to confirm out loud.
Agent Mode runs through the claude CLI. If you do not have it:
curl -fsSL https://claude.ai/install.sh | bash # per-user, no sudo
claude # sign in, then /exit
7. Optional — go fully offline
brew install ollama && ollama pull qwen2.5:3b
Then set the cleanup provider to ollama in Settings → Engine. Dictation, cleanup, and screen context now all run on your Mac with the network off.
What next
- Dictation — personas, hands-free, spacing
- LLM backends — pick the right cleanup model
- Troubleshooting — if the hotkey does nothing