Setup GitKraken Kepler: The Complete Guide to Agent-Driven Development

What Is Kepler and What Problem Does It Solve?

Teams often struggle when they hand feature work to AI agents. Normal coding tools do not keep work apart well. They also miss the full picture across many repos at once. When you Setup GitKraken Kepler, you get a full Agentic Development Environment, or ADE for short. It brings tasks, repos, and safe sandboxes into one system. This stops agents from writing over each other's work.

Why Kepler Was Built This Way

Old Git tools just show you commit trees after you make changes by hand. Kepler works in a new way. It runs tasks, manages safe worktrees, and checks agent output before anything reaches a pull request.

Who Should Use This Guide

This guide fits any team that runs, or plans to run, AI coding agents on real work. It also helps solo coders who want one clean setup, not loose scripts and open terminal windows.

How to Install Kepler: The Setup Checklist

To Setup GitKraken Kepler, you install the app and finish a short checklist. Here is what your system needs before you begin.

Get the installer for your system, then run it. The setup steps check your current tools and set up file paths on their own, with no manual edits needed.

Signing In and Setting Up Your Profile

Open the sign-in screen and add your name and email. This links your commits to your real account, so each agent-made change traces back to you correctly, every time.

Checking Your System Is Ready

Before you launch your first task, check your Git version and shell setup match the table above. This small check now saves you a stuck install later on.

Touring the Kepler Workspace

The workspace gives you full control over active tasks and branches. Learning this layout is the first real step toward using Kepler well each day. You can see active sessions, changed files, and new commands, all from one dashboard. This keeps your project clear, without you digging through raw Git commands each time.

Key Panels to Know

Look for the Todo list, live terminal docks, and the main commit graph. These three spots cover most of what you use each day, from tasks to agent checks.

Setting Up and Connecting Your Repositories

Connecting your repos the right way stops broken links later on. You can link local folders, set up remotes, and prep projects for agent work, all from one setup screen. Here is the basic flow. Your remote code, on GitHub, GitLab, or Bitbucket, syncs with your local project files. Kepler then builds safe task worktrees on top of that synced base, so agent changes stay sandboxed until you approve them.

Cloning or Starting a New Repository

To clone one, paste your project URL and pick an empty folder. Kepler pulls branch history and open pull requests right away. To start fresh, pick "new repository" from the menu and set your root folder.

Linking GitHub, GitLab, and Jira

Connect your accounts through normal sign-in steps. This links your issues and pull requests straight into Kepler, so tickets sync with your team's daily work with no extra setup.

How to Start and Manage Tasks

A task in Kepler groups your goal, target branch, and agent session into one spot. This keeps work tidy, instead of spread out across loose branches and chat messages. Kepler can also pull tickets straight from your project board. This cuts down on vague instructions, since the agent gets real ticket context right from the start.

Starting From Existing Tickets

Pick a ticket from the side panel to launch a task right away. Kepler pulls the description, makes a branch, and sets up the agent session, all in one single step.

Creating Untracked Tasks for Quick Tests

For quick, private tests, use an untracked task. This gives you a sandbox to try ideas without putting a branch on your team's shared board before it is ready.

Organizing Your Daily Todo View

The Todo view shows active work, open reviews, and finished tasks in one place. You can drag cards to adjust order, so your team always sees what matters most that day.

Putting AI Agents to Work

Once your task is set, hand it to an AI agent like Claude Code, Codex CLI, or another connected tool. Kepler passes the right file paths and instructions on its own. Keep your API keys stored in Kepler's locked vault. This lets agents run safely, with no keys ever sitting out in plain text anywhere in your project.

Key principle: handing real coding work to an AI agent takes a clear task, a safe sandbox, and a real human check before anything ships.

Using Quick Actions

Kepler offers one-click steps like Plan, Review, and Fix. These handle common jobs fast, so you are not typing the same notes over and over for routine work.

Steering the Agent With Custom Prompts

You can guide an agent mid-task through the built-in terminal. If it takes a wrong turn, add a note or fix, and it adjusts its next step based on your input.

Following Agent Work in Real Time

When several agents work at once, you need a clear view of what each one is doing. Kepler shows active branches and sessions right inside its live commit graph. Watch for the WIP marker. It flags work still in progress and live agent sessions, so you always know what is not yet final before it turns into a real commit.

Managing Multiple Repos Without Conflicts

Big projects often share code across several parts at once. Kepler tracks linked branches across these repos, so agents do not step on each other's work by accident.

Reviewing and Committing Agent Code

Always check AI code before it becomes part of your project. Kepler's diff view shows each new, changed, or removed line, in green and red, so nothing hides from view. Check each change at the line level. This catches small bugs early, well before they reach your main branch or touch your live product in any way.

Staging Changes and Writing Clear Commits

You can stage just part of a file if needed, and undo staged changes if you spot a mistake fast. When you commit, write a clear note that says what changed and why.

Managing Branches and Pull Requests

Keep your main branch stable. Keep agent work on its own separate branch instead. Kepler lets you drag one branch onto another to merge, so joining finished work stays quick and simple. When branches conflict, Kepler shows both versions side by side. You pick the right lines, confirm the result, and move on, with no digging through raw conflict markers.

Pushing Changes and Opening Pull Requests

One click pushes your checked commits and drafts a pull request on GitHub or GitLab. This keeps your review process fast, while still giving your team a clear paper trail.

Best Practices for Using Kepler Well

A few simple habits keep your repos clean as your team scales up agent-driven work. These matter even more once several people run tasks at the same time.

Common Setup Mistakes to Avoid

Skipping a proper .gitignore file is one of the most common slips teams make. This lets build files, and worse, private API keys, end up in your commit history by accident. Set this up before your first real task, not after. A few minutes here saves you real cleanup work later, once agents start generating files across your project.

Rushing the Initial Repo Setup

Some teams rush through checks when they first set up Kepler, just to save time. This often costs more time later, once a task fails midway due to a missing tool or wrong Git version.

Setting a Realistic Timeline

Plan for a short learning stretch right after you Setup GitKraken Kepler. Most teams feel at ease with daily use within the first week, once they run a few real tasks through the full check flow. Do not expect to master every feature on day one. Start with simple, single-repo tasks. Move to bigger, multi-repo work only once your team trusts the review and merge process.

Final Thoughts

To fully Setup GitKraken Kepler is to gain a real system for safe, AI-driven work. It keeps agent tasks separate, checked, and clearly tracked from start to finish. Start with one small task, get used to the review flow, then scale up to bigger, multi-repo work once your team trusts the process.