I came to Styx from Atom and SourceTree, pushing to Bitbucket/GitHub and deploying over SSH by hand. Bringing AI agents into that setup meant a mess of terminals, and I never liked having agents run commands on a machine that holds my keys.
Styx fixed both.
All my projects and agents live in one window, and each agent works on its own branch, so I can have Claude and Codex on different tasks without them touching my code. Reviewing their changes hunk by hunk and undoing a bad turn in one click is much easier than untangling it myself.
My favourite part is deploying. My SSH deploy script runs from a button, agents have to ask before touching anything live, and I approve with Touch ID. The agent never sees my keys, and there's a log of everything it was allowed to do.
That's what finally made me comfortable letting agents near production.
The only downside is that it's Mac-only for now. Windows is on the way.
Atlassian Sourcetree