Just tried the new Mac app for Codex

OpenAI shipped a native macOS app for Codex and I've been poking around with it today. They're calling it a command center for managing multiple AI agents on your codebase and I can see why.

OpenAI is clearly going after Cursor and Claude Code here. Their pitch is that the bottleneck now isn't what agents can do but how we supervise a bunch of them at once.

It's included with Plus, Pro, Business, Enterprise, and Edu plans. Free and Go users get temporary access too so you can try it out without committing. Rate limits got doubled across paid plans which is nice.

Windows and Linux folks are on a waitlist for now since it's built on Electron for cross platform later.

Parallel agents feature: run multiple agents on the same repo using Git worktrees so they don't step on each other, but that's nothing groundbreaking since this is something I was doing before anyway.

They also added Skills which are basically reusable instruction bundles that have been gaining more and more interest in the dev community. These extend Codex beyond just code stuff, things like Figma integration, Linear, Vercel deployments, doc generation.

What I was curious the most about are the Automations for scheduling background tasks like issue triage and CI failure summaries. These could be super useful if you've got a lot of noise to cut through.

Has anyone else taken it for a spin yet? I'm wondering how the multi agent workflow feels compared to just running one agent at a time in something like Cursor or Claude Code.

I'm going to write a longer blog post about this, hopefully soon, as soon as I get to play with it more.