Hey everyone, maker of Orbit here. I built this because I was living in the terminal running Claude Code and Codex all day, and it got messy fast. Multiple agents going at once, no way to see what any of them were actually doing, constant tab-switching to check git, and copy-pasting context between tools. I wanted one window where I could just watch the work happen. So Orbit started as a simple idea: what if every agent was an orb on a canvas? You spawn one, drag it anywhere, point it at your code, and see each step in real time. Then I added the things I kept reaching for: built-in git with diffs, @ file tagging, usage tracking for both Claude and Codex, and the ability to wire one agent's output into the next. It's native Mac, no Electron, no browser. Claude and Codex get equal footing because I use both and didn't want to pick a side. I'd genuinely love your feedback. What's your current setup for running coding agents? What feels broken about it? Happy to answer anything about how it works under the hood. Thanks for checking it out.

Orbit

Understand Your AI Better, Move Faster, Build More

Understand Your AI Better, Move Faster, Build More