Introducing Styx

A few months ago I had four coding agents running across five projects. One of them had been waiting on me for twenty minutes, and I had no idea which one. It was sitting in a terminal, on a branch, in a window I wasn’t looking at.

That was the moment I stopped looking for a tool for it and started building one.

Who I am

I’m Nic. I build software from South Africa, usually several things at once, and over the last year most of that work has moved to AI coding agents: Claude Code, Codex, Gemini and Cursor, often all in the same week. They’re very good. Keeping track of them wasn’t.

What Styx is

Styx is a free app for the Mac. It puts every project you work on, and every agent working in them, in one window. People have started calling this kind of tool an agentic development environment, or ADE: not a place to write code, but a place to run and supervise the agents that do.

It does three things.

1. Everything in one window

Every project, every agent, and the ones waiting on you first. One keystroke takes you to the agent that needs you. Each agent works in its own copy of the repo, on its own branch, so they never trip over each other or over you, and when the work is ready you land it on main once your checks pass.

2. It doesn’t replace anything

Styx sits on top of the editor you already use. It picks up your recent projects, keybindings and theme, and opens any file back in your editor with one key. It runs on the Claude, ChatGPT, Gemini and Cursor subscriptions you already pay for. I don’t resell model access, and there’s nothing new to buy.

3. Your agents never hold your keys

The useful agents need to touch real things: a deploy, a database, a server. Until now that meant pasting in a token and hoping, or sitting there watching. In Styx your credentials stay in your Mac’s Keychain. When an agent wants production, it stops and asks. You approve it for a set time, with Touch ID for anything live, and the access ends on its own. Every request and every approval goes into a log that nobody can edit: not the agents, and not me.

How it was built

Most of Styx was written by the agents it now runs, several at a time, each on its own branch. That’s how I found almost every problem it solves: the agent I lost track of, the token I shouldn’t have pasted, the one that quietly waited all afternoon.

It also taught me that “run an agent” isn’t one thing. Claude Code, Codex, Gemini and Cursor each talk to the outside world differently, and getting all four to stop and ask before touching something real was most of the work.

Where it stops

Styx is a guardrail, not a sandbox. Agents run as you, on your Mac. The commands that go through Styx’s tools are gated; an agent that deliberately goes around them isn’t stopped. For production I’d still use credentials with the least access you need. I’d rather say that here than have you find it.

Your code, prompts and agent conversations stay on your Mac. Styx counts a small, fixed set of things, like launches and agents started, so I can see what people use, and you can turn that off in Settings. The privacy policy has the details.

What’s next

Styx is on Mac today and free. Windows is next. After that, it depends on what you tell me, which is the real reason I’m launching now.

If you run more than one agent, I’d love to know how you keep track of them today, and what you’d need to see before you let one near production. There’s a Feedback button in the app that comes straight to me, and I read every message. Or write to hello@heystyx.com.

Nic