Running one coding agent is manageable. Running four of them across three repos, each at a different stage, each waiting on something different from you, is where most developers lose the thread. That’s the problem Kepler is built to solve, and it’s also why we put together a full tutorial series instead of a single walkthrough. Parallel agent work has more moving parts than a five-minute demo can cover.

We pulled together every video in the Kepler tutorial playlist below, in order, so you can watch the whole series or jump to the piece you need right now.

Welcome to Kepler

Coding agents are fast becoming part of the software development workflow, but agent conversations easily get spread out and disjointed as you work. Tracking sessions from multiple agents across multiple surfaces becomes a pain.

This first video sets up the problem Kepler solves: how to leverage parallel agents working on multiple issues and PRs while still keeping track of the work. Kepler is an agentic development environment (ADE) built to manage multiple sessions in parallel using any agent and any model, scoped to any number of repos, issues, or pull requests.

How Do You Start a Task in Kepler?

A task is Kepler’s container for agent sessions, changes, and resources. This video walks through the three ways to start one: from an issue, from a pull request, or from scratch when you’re just exploring an idea.

Connect your issue tracker and your to-do list populates with assigned issues. Pick an action like Plan from the dropdown, and Kepler kicks off a session with the agent and model you’ve set up for that action, feeding the issue’s context straight into the task. The same pattern works for reviewing assigned PRs. And if you don’t have a ticket yet, you can start a blank task, type out your idea, and attach work trees, issues, or files as you go.

Tour the Kepler Workspace

Before you can move fast in Kepler, it helps to know where everything lives. This video is a straight tour of the main UI: your tasks, your agent sessions, and your project context, all in one place.

What’s covered:

• Task lists
• Task view
• Kanban view
• Agent Graph
• Settings
• Task panels and tools

Follow Agent Work with the Agent Graph

Once you have more than one agent session running, the question stops being “is it working” and starts being “what is it actually doing right now.” That’s what the Agent Graph is for.

The graph shows every in-progress agent session for your tasks, including which ones are waiting on your permission and which ones have hit an error. You can view it top-down or left-to-right, and control how many edges (the connecting lines between steps) show at once, from a simplified view up to a full breakdown that includes sub-agent activity. Each task also has its own Agent Graph, so you can watch a single session update turn by turn: a permission request comes in, you hit allow, and you can see the agent move from opening a terminal to closing it out, live.

Use Actions to Launch Work Faster

Actions are the fastest way to kick off a task in Kepler. From the to-do tab, each issue or PR gets a button, usually Plan, sometimes Review or Address Feedback, that launches a session with your default agent, model, and mode already applied.

You can change the agent and model per action, attach extra context like other repos or files, or use the dropdown for alternate actions on the same item. The Manage Actions settings (the gear icon) let you set defaults globally: what happens for Jira issues versus GitHub issues, what happens for PRs you authored versus PRs from others. You can also build fully custom actions with their own prompts, including calling a specific skill your coding agent already supports.

Start Work From Issues and Pull Requests

This video walks through the full flow of turning a real GitHub issue into agent work inside Kepler. Click into an issue, hit Plan (from the top corner, the list view, or the dedicated start button), and Kepler jumps you straight into a chat with the agent, the issue open alongside it.

The task view is where you track everything happening on that piece of work: the running agent session, any external sessions like Claude Code, and the ability to launch a second agent (say, Codex) on a different part of the same problem. You’re not limited to one task at a time either. While one session is in progress, you can open another issue, hit Plan again, and track both from the Task in Progress tab, grouped by activity so you can see what’s active and what’s disconnected and needs your attention.

Try Kepler Yourself

Kepler is in Public Preview now for Windows, Mac, and Linux. If you’re already running parallel agents across your repos, or thinking about it, this is the point to start tracking that work in one place instead of five terminal tabs.

Download Kepler

FAQ

What is Kepler? Kepler is GitKraken’s agentic development environment (ADE), built to manage multiple coding agent sessions in parallel across any number of repos, issues, or pull requests.

Do I need to use a specific coding agent with Kepler? No. Kepler works with any agent and any model. The tutorials above show sessions running with Claude Code and Codex side by side on the same task.

What is the Agent Graph? The Agent Graph is a live visual view of what an agent session is doing at each step, including permission requests, tool calls, and sub-agent activity. You can view it per task or across all your in-progress tasks at once.

Can I connect Kepler to my issue tracker? Yes. Once connected, your assigned issues and PRs populate in the to-do tab, and you can launch agent sessions directly from there using Actions.