Skip to content
Kaba

Getting Started

New here? Start with What is Kaba? and Why Kaba?. This page gets it installed and running.

Install Kaba

Section titled “Install Kaba”

Kaba runs on your own hardware. Installing the client also installs kabactl, the engine it talks to, so there is nothing else to set up for a single machine.

Download the build for your distribution from kaba.ai/download, or install from a shell:

Terminal window
curl -fsSL https://kaba.ai/install.sh | sh

RPM

Section titled “RPM”

Fedora, RHEL and derivatives:

Terminal window
sudo dnf install ./kaba-<version>.rpm

DEB

Section titled “DEB”

Debian, Ubuntu and derivatives:

Terminal window
sudo apt install ./kaba_<version>.deb

AppImage

Section titled “AppImage”

Any distribution. No install step:

Terminal window
chmod +x Kaba-<version>.AppImage
./Kaba-<version>.AppImage

Flatpak

Section titled “Flatpak”

[todo: confirm the Flatpak is published, and add its install command and app ID.]

[todo: confirm the exact package file names and download URLs for RPM, DEB and AppImage.]

Other platforms

Section titled “Other platforms”

Builds also exist for macOS (.dmg) and Windows (installer). Linux is the first-class platform; see platform support.

To run only the engine on a server, see kabactl and deployment.

Build from source

Section titled “Build from source”

The client needs Node 22.4 or newer, plus a C/C++ toolchain.

Terminal window
# Debian / Ubuntu
sudo apt-get install libtool m4 make g++ autoconf


# Fedora
sudo dnf install libtool m4 make gcc-c++ libXScrnSaver
Terminal window
git clone https://git.djcas9.com/kaba-labs/kaba.git
cd kaba/
make          # build
make dev      # run the app
make watch    # rebuild on change

First run

Section titled “First run”

  1. The engine starts. A boot screen shows kabactl being prepared and started. The client copies the bundled binary to <user data>/bin/kabactl, links it onto your PATH, and waits for it to answer on port 28832.

    [todo: add screenshot of the boot screen and of parts engine status steps, status notes, error card with "Restart kabactl".]

  2. Create your account. Enter a first name, last name, email address, language and password. Kaba currently supports one local account. Your password unlocks the key that encrypts your memories and saved logins, so it cannot be recovered for you. See accounts.

    This is Kaba Personal, where the account is local. On a fleet-managed device your organisation controls sign-in; see what fleet control changes.

    [todo: add screenshot of the Welcome to Kaba account screen and of parts name fields, email, language, password, Create Account.]

  3. Land on the desktop. A new frame opens at kaba://desktop/, a widget dashboard you can edit. See desktop.

  4. Take the tour. Window → Get Started Tour (Shift+F1) walks through the omnibar, frames and the Kaba menu in about a minute.

The three things you will reach for

Section titled “The three things you will reach for”

| Control | Shortcut | What it does | |---|---|---| | Omnibar | Ctrl+Space or Ctrl/Cmd+L | Navigate, search, jump to a frame, or run a /command. It appears when summoned and then gets out of the way. | | Frames | Ctrl/Cmd+` | Kaba's take on tabs. Each frame holds one or more tiled panes. | | Avatar menu | avatar, top right | Terminal, files, Hippocampus, downloads, settings, and the ad-blocker switch. |

It has its quirks and features, but fundamentally it's kind of like a browser. Just get in and drive.

Turn on the parts you want

Section titled “Turn on the parts you want”

Browsing, the terminal, files and the privacy layer work out of the box. Two things are opt-in or adjustable under Settings:

  • Enable AI is off by default. Turning it on adds the context-menu text actions, Projects, Toolbench, and the Policies and Models pages. The first model downloads when you ask for it.
  • Enable Memory is on by default. Pages you visit become searchable memories in Hippocampus. Turn it off to stop all automatic capture.

Add a second device

Section titled “Add a second device”
  1. On the first device open Settings → Devices and choose Invite a device, or run kabactl cluster invite.
  2. On the new device paste the ticket into Join a cluster, or run kabactl cluster join <ticket>.
  3. Both devices now appear under Devices. You can open a terminal on the peer, browse its files, run inference on it, or route traffic through it.

See cluster & mesh.

Next steps

Section titled “Next steps”