T3 Code: A Free Open-Source Control Panel for Running Multiple Coding Agents
T3 Code is an MIT-licensed interface that lets developers manage Claude Code, Codex and other coding agents from one place. It does not include its own AI model.
Source: Dev.to
Developers who run several AI coding agents at once often end up juggling multiple terminals, with no single view of what each agent is doing or what it has changed. T3 Code, a project from Theo Browne's team, aims to solve that. It is a graphical control layer rather than a coding agent: it drives the agent command-line tools already installed on your machine, including Claude Code, Codex, Cursor, Grok Build, OpenCode and Google Antigravity. It signs in through each tool's own login, so existing subscriptions still apply, and the software itself is free.
The app offers a desktop version for macOS, Windows and Linux, along with a web app and mobile apps for checking on agents remotely. Each task runs as a thread that can use its own Git branch or worktree, and changes appear as a diff that can be turned into a pull request from within the app. The project is still at version 0.0.x, and new threads default to full access, which lets agents run commands without asking. The project's documentation recommends switching to Supervised or Auto mode before starting real work.
T3 Code is also early software with limited contributions accepted for now, so users should expect bugs and stick to the stable release channel. Alternatives include Conductor for Mac users, Claude Squad for terminal-focused developers and Vibe Kanban for ticket-based planning. Pricing details were checked in early November 2026, and readers should confirm current terms on each project's website.
