Kepler Documentation

Getting Started with Kepler

Last updated: August 2026

Kepler is GitKraken’s Agentic Development Environment (ADE): a place to direct coding agents across your real work, in as many repositories as you need, at the same time.

You bring the agent. Kepler pulls in the issues and pull requests already assigned to you, hands any of them to an agent with the context attached, and takes the result through to a reviewed, mergeable change.

Kepler is in public preview. A free GitKraken account is all you need.


Install Kepler

Platform Installer Minimum OS
Windows 64-bit, ARM64 Windows 10+
macOS Apple Silicon, Intel macOS 12+
Linux (x64) .deb, .rpm, .AppImage, Flatpak, AUR Ubuntu LTS 18.04+ / Debian 10+ / RHEL 8+ / Fedora 39+
Linux (ARM) .deb, .rpm, .AppImage Ubuntu LTS 20.04+ / RHEL 8+ / Fedora 39+
  1. Go to gitkraken.com/kepler/download and pick the installer for your platform.
  2. Run it.
  3. Launch Kepler.

Sign in

Kepler opens on a welcome screen. Click Sign in with GitKraken. If you do not have an account, you’ll create one in the next step.


Kepler's welcome screen with the Sign in with GitKraken button
The welcome screen on first launch.

A Kepler Quick Start Tour video appears on that screen if you’d rather watch first, or watch the Kepler Quick Start Tour playlist on YouTube.


Finish the setup checklist

The Setup entry in the toolbar tracks 5 things, and shows Setup · N/5 until they’re done. Open it for Finish setting up Kepler and a progress readout.

# Item What it does
1 Sign in to GitKraken Done when you signed in above
2 Connect an AI agent At least one coding agent. See Agent Integrations
3 Connect issue & PR trackers Any provider. This is what fills your Todo list. See Issue Tracker Integrations and Pull Request Integrations
4 Set a default repositories folder Where Kepler clones repositories it does not find locally
5 Set a default worktree folder Where Kepler creates one worktree per repository, per Task

Expand More setup options for 3 optional items:

Add repo commands is worth doing early if your project needs a setup step. Save a repository’s install or build command once and Kepler can run it automatically every time it creates a worktree, so an agent never starts in a checkout it cannot build. See Tasks and Resources.

Folder paths and everything else live in Settings. You can hide the checklist from the toolbar and still reach it there.


Put an agent on real work

After you connect a tracker, the Todo segment of the Kepler interface fills with the issues and pull requests assigned to you.

  1. Pick an item in Todo.
  2. Click its Action button.

The Action dropdown open on a Todo row, listing Plan, Implement, Review, and Address Feedback
The Action dropdown, opened from the chevron next to the button.

Nicely done! Kepler automatically does the following:

  1. Creates a Task.
  2. Attaches the repository and the item.
  3. Sets up an isolated worktree.
  4. Starts an agent with the context already in place.

The left half of the button runs the sensible default:

  • Plan for an issue
  • Review for someone else’s pull request
  • Address Feedback for one of yours

The chevron next to the button offers the rest of the Actions.

Every one of those prompts is yours to change, and you can add your own. That’s the part worth reading next: Actions.


Start something that is not tracked yet

Click New task, or press Shift+Alt+T from anywhere. Write a prompt, attach a repository if the task needs one, and go. You can attach nothing at all and turn it into real work later. See Create a Task.


The New task button in the Kepler top navigation bar
The New task button, available from anywhere in Kepler.

Where to go next

If you want to Read
Change the prompts, or write your own Actions
Understand what a Task holds Tasks and Resources
Work inside a Task The Task View
Direct an agent mid-session Agent Sessions
See what every agent is doing at once The Agent Graph
Reshape the list to fit how you work Arranging Your Work
Review and ship what an agent wrote Review Changes
Run agents on another machine or in WSL Remote Environments

Have feedback about this article? Did we miss something? Let us know!
On this page