AIThis post was created with the assistance of artificial intelligence (AI).

TL;DR

Thorsten Meyer says he moved his Claude Code and Codex sessions into T3 Code, which he uses to manage work across eight Macs and a Linux server from one workspace. The open-source app connects to coding agents already installed and signed in on those machines. Meyer reports a simpler way to track and hand off tasks, but says he has not measured a productivity gain; some remote and multi-machine features remain in beta.

Thorsten Meyer says he has moved his Claude Code and Codex sessions into T3 Code, a workspace he uses across eight Macs and a first Linux server. In an article published November 18, he said the setup lets him follow work from one thread list on his laptop or phone, while the coding agents continue running on the machines doing the work.

Buying for a business?Offer from Amazon

Get business pricing on tech for your team

  • Business-only prices and quantity discounts
  • Tax-exempt purchasing
  • Multiple users, one account, clear invoices
As an affiliate, we earn on qualifying purchases.

Meyer describes his previous setup as a mix of terminal windows, a terminal multiplexer and the Claude Code desktop app. Claude Code sessions ran on individual Macs, while Codex sessions used to check changes sat elsewhere. With many sessions active, he said, it took effort to find which agent was working, which was waiting for an answer and which machine held the relevant session.

In his new workflow, each task has a thread. Meyer said he uses Claude Code to build and Codex to check the work, then tracks both through the workspace. T3 Code does not replace those agents: according to the article’s description of the project documentation, it runs coding agents that users already have available and signed in. The app is described by its makers as a “control plane for coding agents”; Meyer says it is free and open source under the MIT license.

The project documentation cited by Meyer lists several ways to connect machines, including a local network, Tailscale HTTPS, SSH and T3 Connect. The article says T3 Connect is free “for now” and supports up to three machines per account; a paid T3 Chat plan is listed as one route for connecting more, with Pro supporting up to 10. Meyer also says the desktop and web apps can search threads across connected environments, and that iOS and Android apps can provide remote access and notifications. These are descriptions of the product and its documentation, not independent tests of every connection method.

At a glance
reportWhen: Published November 18, 2026; the articl…
The developmentIn a November 18, 2026 article, Thorsten Meyer described switching to T3 Code to manage Claude Code and Codex sessions across several Macs and a Linux server.
Board titled Three reasons, and what T3 Code offers for each. One: several Macs, one surface; agents keep running where they are, one app connects over the local network, Tailscale, SSH or T3 Connect, and iOS and Android apps act as remotes. Two: Claude and Codex in one place; switch models mid-thread, let Claude hand a task to Codex, or start a second agent in the same worktree to check the change. Three: many agents at a glance; one thread per task, an Agents view for subagents, and a Working section that sends a thread back to the top when it needs you. Highlighted note: no tokens resold; T3 Code drives the agents you are already signed in to, and your plans, terms and limits stay as they are
My three reasons for switching, and the matching T3 Code features. Sources: t3.codes and the T3 Code README and user docs on GitHub, read November 18, 2026. Board: Thorsten Meyer AI.

Tracking Agents Across Machines

For developers running agents in parallel, the practical issue Meyer identifies is coordination across sessions and machines. A shared thread list can make it easier to see which task is active or waiting, and to move work from a coding agent to a checking agent. Remote access may also let a user respond to an agent without sitting at the machine where it is running.

The claimed benefit is a more manageable workflow, not a measured increase in output. Meyer explicitly says he has not measured a productivity gain and offers no before-and-after figures. His account is a description of one user’s setup, so it does not establish how well the app performs across other teams, networks or coding projects.

The setup also depends on machines remaining available to do the work. Meyer’s article says the Mac must stay on while an agent continues running remotely. Users still rely on their existing coding-agent providers, and the article says provider terms and plan limits continue to apply. T3 Code’s interface changes how sessions are organized; it does not remove those underlying constraints.

Products Worth Considering

Apple AirPods Pro (2nd Generation) Wireless Earbuds with USB-C MagSafe Charging Case

Apple AirPods Pro (2nd Generation) Wireless Earbuds with USB-C MagSafe Charging Case

As an affiliate, we earn on qualifying purchases.

From Terminals to Threads

Meyer’s move follows a workflow in which Claude Code and Codex served different roles: one built changes and the other checked them. Before adopting T3 Code, he kept sessions in separate terminal windows on each Mac and also used the Claude Code desktop app. He said the agents themselves worked, but carrying context between the building and checking stages and locating sessions across computers had become difficult.

The article points readers to an earlier feature tour titled “T3 Code: One Workspace for Your Claude Code, Codex and Other AI Coding Subscriptions.” The November 18 piece instead focuses on Meyer’s reasons for switching and his own configuration. He cites the T3 Code README and user documentation, read November 18, and identifies version 0.1.10, released November 17, as the version current at the time of writing.

Meyer characterizes T3 Code as a fast-moving, young project. He says several multi-machine features are still in beta and that new threads start in full-access mode. Those details matter for readers considering whether to put the app into a workflow that handles code or requires approval before an agent acts.

“T3 Code runs the coding agents on your machine and gives them one fast interface. Start a thread at your desk, then approve the next step from your phone.”

— T3 Code documentation, as quoted by Thorsten Meyer

Beta Features and Open Questions

The article does not report independent testing of T3 Code’s connection options, mobile notifications or performance across multiple machines. Meyer says several multi-machine features remain in beta, but does not identify every affected feature or describe their reliability in specific network conditions.

It is also unclear how much time the setup saves compared with Meyer’s previous workflow. He provides no measured comparison, and his experience with eight Macs and a Linux server may not represent users with fewer machines or different provider plans. The article says new threads start in full-access mode, but gives no detailed account of the permissions available in each setup or what safeguards users can configure.

T3 Connect’s three-machine allowance is described as free “for now,” leaving its future pricing and terms uncertain. The article also does not provide a date for when beta features may become generally available.

What Users Should Check

Meyer’s immediate next step is to continue using T3 Code across his connected machines; his article does not announce a product launch, a planned release date or a change to the service. For readers evaluating the app, the project’s current README and user documentation are the relevant sources for checking supported connection methods, machine limits and beta status, since those details can change as the project develops.

Users considering a similar setup will need to account for their own machines, network access, agent-provider plans and approval preferences. Meyer’s account suggests starting with the workflow problem—such as finding waiting sessions or handing work between agents—and then checking whether the current T3 Code features address it. No broader productivity result or general recommendation is established by this single account.

Key Questions

What is T3 Code?

It is a free, MIT-licensed open-source app that Meyer describes as a shared interface for coding agents already available on a user’s machines. Its makers call it a “control plane for coding agents.”

Which agents does Meyer use with it?

He says he uses Claude Code to build and Codex to check changes. His article does not provide independent compatibility testing for other agents.

How many machines does his setup include?

Meyer reports using T3 Code with eight Macs and a first Linux server. He says additional machines can join, though connection options and limits depend on the method used.

Has Meyer shown that T3 Code makes him more productive?

No. He says he has not measured a productivity gain. His report describes easier session tracking and handoffs, but includes no before-and-after data.

What should users know before trying it?

Meyer says the project is young, some multi-machine features remain in beta, and new threads start in full-access mode. He also says users remain subject to their providers’ terms and plan limits.

Source: Thorsten Meyer AI

You May Also Like

Secure Your AI Agents: Essential Security Measures For Infrastructure

New security measures are being developed to protect AI agent infrastructure, including a proxy for MCP servers with access controls, audit logs, and safeguards.

What AI Means For Learning To Code: Evidence For Students And Professionals

Evidence from developer studies, hiring data and learning experiments suggests coding skills still matter, but the work is changing.

A Look At Git Infrastructure At OpenAI

OpenAI has published a page titled “Scaling Git at OpenAI,” but the available material does not include its technical details or publication date.

Why Meta And Microsoft Pulled Back From Claude—and What A Switch Costs

A report says Meta and Microsoft are shifting employee AI use toward in-house tools and Copilot. Their ability to switch reflects the cost of building alternatives.