Keystrokes never leave the Mac
Inference happens locally on Apple Silicon. There is no completion server, so your keystrokes, drafts, and predictions are never uploaded, logged, or shared. Turn Wi-Fi off and it keeps working.
Shadowtype is an AI typing assistant that predicts your next words as inline ghost text in any app — and runs 100% on your Mac. There is no cloud keyboard reading what you type, no telemetry, and no account. Your drafts, your style, and your copyright stay entirely yours.
Most AI autocomplete tools route your keystrokes to a server to predict the next word. Shadowtype doesn’t — the model runs on-device as a local LLM and works fully offline. For privacy-conscious users, legal and compliance-sensitive workplaces, and anyone who doesn’t want a cloud keyboard reading their typing, that’s the whole point. Read our full privacy policy.
Inference happens locally on Apple Silicon. There is no completion server, so your keystrokes, drafts, and predictions are never uploaded, logged, or shared. Turn Wi-Fi off and it keeps working.
Shadowtype ships with zero analytics backend — nothing to opt out of, because nothing is collected. You never sign up or log in. The app just runs locally, the moment you launch it.
Secure and password input fields are ignored entirely. macOS flags them as sensitive, and Shadowtype generates no ghost text and captures nothing there — your credentials are off-limits by design.
The optional style profile that learns your tone is stored locally and encrypted on your Mac — never uploaded. Wipe it any time with one click and start fresh, no questions asked.
Because nothing you write is sent anywhere, there is no clause handing your text to a vendor for training. What you draft is yours — full copyright, full ownership, no fine print.
Continuous inline ghost text in any text field — mail, docs, chat, code. Tab accepts a word, ⌥Tab a whole line, and ⌥⌘K rewrites a selection — all on-device.
An OpenAI-compatible HTTP endpoint runs on 127.0.0.1 so your scripts, editors, and agents can call the same on-device model — no key, no cloud. A built-in MCP bridge plugs into Claude Code and other MCP clients, and the BYOM picker loads any GGUF.
Privacy here isn’t a setting you trust — it’s how the app is built. There’s simply no path for your text to leave the machine.
You pick a GGUF model from the catalog (Qwen3 Base, Gemma 3, MoE variants — or bring your own) and it runs entirely on Apple Silicon. Predictions are computed locally — there is no API call to a completion server, ever. The optional developer-facing API binds to 127.0.0.1 only.
Password and secure input fields are never read. The app respects the system’s secure-field flag and produces no suggestions where you’d type a credential.
Any personalization is stored encrypted on your Mac and never uploaded. One click wipes it. With zero telemetry and no account, there’s nothing about you on any server.
127.0.0.1 with no key required, so your terminal, editor plugins, and AI agents can call /v1/chat/completions against the same on-device model. A built-in MCP server plugs into Claude Code and other MCP clients, and the BYOM picker loads any GGUF you drop in. Everything is bound to localhost — nothing leaves your Mac.Download Shadowtype free, accept your first word with Tab, and know nothing you type ever leaves your Mac.