Ask, proofread, and switch models without leaving your work.

A lightweight AI palette for macOS and Windows.
Double-tap Option/Alt by default, or choose your own shortcut.

macOS build is Apple Silicon native and runs on Intel via Rosetta 2.
Or view on GitHub · build from source

New in KEN

More models, your shortcut, and proofread in place.

Pick the model that fits the job.

Use your ChatGPT subscription, OpenAI API key, OpenRouter, Z.ai, Gemini, DeepSeek, Groq, Mistral, Ollama, or local GGUF models.

GPT-5.5 GLM-4.6 Claude Gemini DeepSeek Groq Mistral Ollama

Use the shortcut you actually want.

Keep double-tap Option/Alt, switch to Command, Control, or Shift, or record a normal key combo for the palette and proofread action.

Proofread selected text in place.

Highlight text anywhere, trigger your proofread shortcut, and KEN replaces the selection with a cleaner version using your selected model.

First launch on macOS

After dragging KEN into Applications, run this once in Terminal:

xattr -cr /Applications/KEN.app

Then open KEN from Applications. macOS may ask for two privacy permissions:

Input Monitoring lets KEN hear your global shortcut.
Accessibility lets the proofread shortcut copy and paste selected text.

KEN is currently distributed unsigned. macOS Gatekeeper flags anything you download as quarantined, and on Sequoia (macOS 15) it shows unsigned apps as "damaged and can't be opened" — even though they aren't. The xattr -cr command strips that flag so macOS lets KEN run. You only do it once per install. A signed + notarized release is on the way; until then, you can also build from source — local builds aren't quarantined and start cleanly.

GitHub Releases Issues