A faster car doesn’t get you home faster if the freeway is still jammed. That is the problem most teams run into once they add a second, third, or fourth AI coding agent to the mix. More agents generate more code. They do not automatically generate more finished work, because someone still has to track which agent is waiting on input, which one just opened a pull request, and which one has been quietly stuck for twenty minutes.

Kepler is GitKraken’s answer to that traffic jam. It is now in public preview, free for anyone with a GitKraken account, and available for Windows, Mac, and Linux.

What is Kepler?

Kepler is an agentic development environment (ADE) built to organize AI-driven work across multiple repositories at once. Where GitKraken Desktop is built to go deep on a single repo with the help of an agent, Kepler is built to go wide: one task, several repositories, several agents, tracked in a single place.

Three ways to start a task

Kepler gives developers three entry points, and each one meets the work where it already lives:

• From scratch. Define a task, attach any repositories it touches, and add context by typing, pasting a document, or talking (Kepler includes voice input).
• From an issue. Kepler connects to Jira, Trello, Linear, GitHub Issues, GitLab Issues, and Azure DevOps. Since most teams already write requirements into their issues, Kepler pulls that context in directly instead of asking a developer to retype it.
• From a pull request. Bring an agent in to either address review feedback line by line, or to review a PR before a teammate ever sees it.

A single task can span several repositories at once. Attach one set of shared instructions and every repo in that task inherits it, rather than pasting the same brief into four separate conversations.

Choose your agent, choose your mode

Kepler works with any agent that supports the Agent Client Protocol (ACP), which today includes Claude Code, Codex, Cursor, Gemini, Copilot, OpenCode, and Augment (Augie). A team rationing credits on one provider can shift a task to another without changing tools.

Each task also has a mode and an effort level. Plan mode has the agent talk through its approach before it touches any code, useful the first time through. The default mode prompts for permission before anything risky. A more autonomous mode lets the agent move without asking at each step, with every change still gated behind a commit before it goes anywhere. Effort level is a dial for cost: turn it down to keep token spend in check, turn it up when the task calls for deeper reasoning.

See the whole board, from anywhere

Work in progress shows up in three views: a list, a Kanban board sorted by status (exploring, in development, needs review, done), and a console view for scanning everything at once. Kepler also supports remote access, so a task kicked off from a desktop can be checked from a phone or another browser without leaving it running unattended.

Kepler and GitKraken Desktop, together

Kepler hands off cleanly to GitKraken Desktop. Open a worktree an agent is actively editing, and GitKraken Desktop recognizes it and offers a direct shortcut in, live diff included. That is the point where a developer stops managing an agent’s output from a distance and starts reviewing it up close, ready to commit and push toward a pull request.

Frequently asked questions

Is Kepler free? Yes, during public preview Kepler is included with any GitKraken account.

Which AI coding agents does Kepler support? Any agent built on the Agent Client Protocol, including Claude Code, Codex, Cursor, Gemini, Copilot, OpenCode, and Augment.

Can Kepler work across more than one repository at a time? Yes. A single task can span multiple repositories, each running in its own worktree, with shared context attached once instead of repeated per repo.

Kepler is in public preview, which means feedback shapes what ships next.

Try Kepler in Public Preview →