goto

Lost a window in a sea of windows? Hunting for it with the mouse and Alt+Tab is slow and annoying. goto brings it to the front, just say where you want to go. On Linux, fully offline.

🔒 100% private 🔌 works offline 🗣️ hands-free 🤖 works with Claude Code

⬇️ Download (.deb / .rpm / .tar.gz) View on GitHub

you: "goto vscode myproject" → focuses the VS Code project

you: "goto slack john doe" → opens the DM

you: "goto chrome github" → jumps to the right tab

How it works

1. Say it

Hold a key or just speak, the way you'd tell a person: "goto slack john doe".

2. goto finds it

It figures out which window, tab or conversation you mean, across everything you have open.

3. You're there

The right window snaps to the front. No mouse, no endless Alt+Tab.

Your voice never leaves your computer. Everything runs locally, nothing is sent to the cloud.

What you can do

Windows & apps

"goto terminal logs", "goto slack", "goto browser" (focuses the only open browser).

Chrome tabs

"goto chrome github" uses Chrome's native tab search to jump straight to the right tab.

Slack DMs

"goto slack john doe" opens the conversation via Jump to (Ctrl+K).

VS Code files

"goto vscode app config" opens the file via Quick Open (Ctrl+P).

Accent-tolerant

"goto", "go to", "good to", "gotchu"... recognition forgives the variations.

Extensible

Each program is an isolated adapter. Adding a new app is one file, no core changes.

The tray menu

goto lives in the system tray. Click the icon and everything is one click away:

Start / Stop listening

Turn voice capture on or off whenever you want.

Mode

Wake word "goto" (hands-free, always listening) or Hotkey (hold a key, speak, release).

Language

Português (BR), English or Automatic. Pick the language you actually speak for the best accuracy.

Precision

Normal is light and fast. High downloads a larger, more accurate model (~1.5 GB, once) and then runs fully offline.

Calibrate mic

Records 8s and auto-tunes and saves the voice sensitivity for your microphone. One click, no terminal. Do this first if recognition is poor.

Show logs

Opens the recent-activity log: what goto heard, errors and model downloads.

Recognition not great? Two steps, in order:
1. Calibrate mic — auto-tunes detection to your microphone (start here).
2. Precision → High — downloads a bigger, more accurate model (~1.5 GB, once, then offline).
Also make sure Language matches what you actually speak.

Install

One line, that's it:

curl -fsSL https://espigah.github.io/goto/install.sh | bash

Or grab a package:

# Debian / Ubuntu
sudo apt install ./goto_<version>_linux_amd64.deb

# Fedora / RHEL
sudo dnf install ./goto_<version>_linux_amd64.rpm

# Portable (tar.gz)
tar xzf goto_<version>_linux_amd64.tar.gz && ./goto

Prefer one portable file that just runs? Grab the .AppImage from the releases.

Snap (experimental):

sudo snap install goto-window --edge
sudo snap connect goto-window:audio-record   # allow the microphone

The Snap is still in testing and may have trouble installing or running. If it gives you grief, use the one-liner, a package, or the AppImage above.

A one-time voice pack downloads on first run. goto lives in the system tray and can start with your login.

Becomes an arm of Claude Code

goto exposes desktop actions as MCP tools. Connect it and Claude Code can focus windows, switch tabs and open conversations for you:

claude mcp add --transport stdio goto -- /path/to/goto mcp

Beta. The hands work; voice recognition sometimes misreads proper names. Feedback is very welcome in the issues.

Stop hunting. Just say where you want to go.

Your windows, one sentence away. Free, open source and private, for Linux.

⬇️ Get goto