T3 Code — One App for Every Coding Agent I Run

Things I'm Trying

T3 Code — One App for Every Coding Agent I Run

T3 Code is an open-source (MIT) control plane for coding agents from the t3.gg team, orchestrating Claude Code, Codex, Antigravity, OpenCode, Cursor, and Grok from one surface with your own subscriptions.

T3 Code is an open-source (MIT) control plane for coding agents from the t3.gg team. It orchestrates Claude Code, Codex, Antigravity, OpenCode, Cursor, and Grok from one surface. The pitch is simple: bring your own subscription, fork the whole thing.

The part that sold me is the billing model, or rather the lack of one. T3 Code does not resell tokens. You plug in Claude Code, Codex, Antigravity, OpenCode, Cursor, or Grok with the credentials you already have. No keys resold, no quota caps, and you can switch models mid-thread. More harnesses ship weekly.

The workflow detail that matters day to day: every agent thread writes to its own branch. When it is good, one button opens the PR on GitHub with a generated title, body, and changelog. You get auto-generated PR titles and bodies, inline diff review before you push, and draft, stacked, and amended PRs, all working with your existing GitHub auth. No terminal dance required.

And because it is MIT licensed with 24k+ stars, if I do not like something, I can fork it. Change the UI, wire in my own tools and models, ship my own build. That is the whole deal: one app for every agent I already pay for, and I own the surface it runs on.

I run more than one coding agent every day: Grok, OpenCode, Codex, Claude. If you run more than one coding agent, this is the one-app version of that.

/things-im-trying/2026-10/t3-code-one-app-every-coding-agent-i-run austin.amento.dev