The biggest Kepler update since public preview is live. If you tried Kepler early, it’s worth another look.

Starting agents is easy. But staying on top of them is the hard part. 

Once you have a handful running at once, it gets harder holding onto what you’re actually trying to get done while they go. This release is built to calm that chaotic feeling: one place to run the agents and models you already use, and take a change from idea all the way to merged.

Bring any agent, any model

Choose the agent and model for each Task, and run several at once. When something better ships next month, swap it in per Action, and everything you’ve configured carries over. Kepler stays agnostic, so your stack stays yours.

Choose an Action. Skip the prompting.

You shouldn’t have to write a prompt to get an agent moving. Every issue and PR carries built-in Actions: implement an issue, review a PR, address feedback. 

Click one and Kepler hands the agent the issue, the diff, and the repo state, already attached. You can edit the prompt, model, and skills behind any Action, point it at a skill you maintain, or build your own.

From idea to merged, without losing the thread

Start from an idea, an issue, or a PR, and carry it all the way through: explore, implement, review, address feedback, and merge, without rebuilding context across tools and sessions. 

A Task holds all of it in one place. And an agent finishing isn’t the same as work that’s ready to ship, so Kepler puts each change on its own branch and lets you review, commit, and open the PR. When you rely on a raw agent, it stops at “done.” Kepler takes you the rest of the way to merged code.

Watch every agent work

Agent Graph draws your agent work as it happens. Every Task, session, turn, tool call, and subagent appears live as the work runs. Click any node to see its status, its exact input, and what it did last. When a lot is running at once, this is how you keep the thread.

Try it, and tell us what you think

Kepler is in public preview and free. Download the latest version of Kepler, or relaunch to pick up the auto-update if you already have it. This release came straight from developer feedback, and that loop is still open:

• The feedback button in the top right of Kepler goes straight to the product and engineering team.
• Our public Slack channel, where they’re already in the conversation. 
• Our public GitHub repo at github.com/gitkraken/gk-ade/issues.

Try Kepler, and tell us what’s working and what isn’t. 

FAQ

Is Kepler free? Yes. Kepler is free to use during public preview.

What is Agent Graph? Agent Graph is a live, animated view of every Task, session, turn, tool call, and subagent in Kepler, drawn as the work happens.

Which issue trackers and Git hosts does Kepler support? GitHub, GitLab (including self-hosted), Jira, Azure DevOps, and Bitbucket, all built on the provider library that’s run inside GitLens for years.

Can I change what an Action does? Yes. An Action’s prompt, agent, model, mode, and effort level are all editable, and you can build new ones from scratch or point them at a skill you already maintain.

Does a Task need a worktree? No. A Task works without one. When you do want a worktree, Kepler creates it and can run saved setup commands automatically.

Is Kepler ready for team-wide standardization? Kepler is in public preview. Team-level standardization, grouping the repos, integrations, and Actions for one body of work, is a direction we’re actively exploring, not a shipped feature.

How do I send feedback? Use the feedback button in the top right of the Kepler window, join our public Slack channel, or open an issue at github.com/gitkraken/gk-ade/issues.