Vercel's Coding Agent fx Is Now a Built-In T3 Code Provider

Vercel's open-source coding agent fx is now a built-in T3 Code provider on web, desktop and mobile, with Supervised and Auto modes and your existing models.
Coding agents keep multiplying, and the hard part is managing them. Vercel says its open-source agent framework, fx, is now a built-in provider in T3 Code, available on web, desktop and mobile.
One control panel, one more choice
T3 Code is an open-source control plane that lets you run several coding agents from one interface. It currently supports Claude Code, Codex, Cursor, Grok Build, OpenCode and Google Antigravity, using your own subscriptions and without reselling tokens. With fx, the list gets one more name, and unlike most of the others it is fully open source.
Vercel says the built-in fx provider supports Supervised and Auto modes, along with the models and skills you already use with fx. Getting started is simple: install fx and set up your model provider, then download T3 Code and add fx under Settings > Providers > Add provider.
What fx is about
fx positions itself as a minimal, Unix-style coding agent. It is a single native binary written in Zig, needs no runtime, and prints output like a shell rather than an IDE, with the conversation going into your terminal scrollback. The project says its CI fails a pull request if any benchmarked command takes longer than 2 ms, so "small and fast" is taken seriously.
It is also built to be embedded. Scripts call it with fx ask, editors connect over the Agent Client Protocol, and JavaScript apps can run the same agent core through libfx. For models, it can use Vercel AI Gateway, a ChatGPT or Grok subscription, or an OpenAI-compatible endpoint such as Ollama or OpenRouter. It is licensed under Apache-2.0, so teams can inspect it, fork it and embed it in their own tooling.
For developers, the point is being able to swap. In one project and one T3 Code interface, you can run fx alongside other agents, compare the results and decide who gets the job. As coding agents become standard equipment, that freedom to switch inside a single console may matter more than any one agent's benchmark score.