Claude Code & Codex · Mac · Windows · Linux
Manage a fleet of agents in a beautiful UI, see which need your attention and monitor usage better.
free · unlimited agents
Light & legible
More natural than working with a terminal, and better at helping you decide what needs your attention. See the full comparison →
Usage & limits
Better analytics to monitor and manage your token usage.
At this pace your weekly limit runs out in about 1d 20h, before it resets.
Every design decision has been taken with performance in mind. The app is only a few megabytes and lightning fast, which means you can manage a fleet of agents efficiently.
Questions
No - it's free to use Abralo. You will need to log into either Claude Code or Codex with your existing Claude or OpenAI logins though.
Yes - you can use different coding agents working on different projects in different columns. If you run out of tokens in one column, you can even switch to another coding agent without losing context.
We show you sparklines and session predictions for Claude Code, but Codex is sadly more opaque so there's less we can show you.
Your conversations, code and prompts are between you and whichever model provider you're using. Abralo can't see any of it. We do track anonymous usage of features (to see what we need to build next), but you can opt out of this at any time in the in-app menu (bottom right).
I originally used Claude Code in a terminal, and found myself using a split terminal to manage multiple at the same time.
I quickly found it hard to keep up with multiple agents as the text-heavy UI of the terminal made it hard to parse.
I started using an Anthropic Claude Code extension in VS Code, but it was difficult to switch between conversations and it crashed my laptop a few times when I had more than 3 running at a time.
So I built Abralo, which is lightweight and fast, but which makes these agents easier to understand and read, which means I can manage several agents at a time. It also makes it painfully obvious if your input is needed to unblock an agent.
There are a few good tools for running several Claude Code agents at once. Most of them focus on running many agents on a single repo, isolated in git worktrees. Abralo is lighter and takes a different angle: one agent per project folder, all in view, with a nudge the moment one needs you — and it isn't tied to one vendor's CLI.
I wrote an honest comparison, including where the other tools are the better pick. See how they stack up →
Yes. The whole app is free, with no limit on how many agents you run at once. You don't need an account beyond the Claude or OpenAI one you already have, and there's no time limit or nagging.
macOS, Windows, and Linux. A real native build on each, not the same web page in three wrappers.
Email me at chris@abralo.com with any questions, issues or suggestions. Thanks!
Free. No signup. Installs in a few seconds.
macOS, Windows, or Linux, 64-bit.