AURIX puts AI coding agents in one Mac workspace

A coding task can quickly become a coordination problem: one agent makes a change, another needs the decision behind it, and you still have to check the result. AURIX is aimed at that layer of work. It does not write code itself or replace the coding agents. It offers a Mac workspace for running them together and keeping some of their shared context in view.

That distinction matters. If you want a new coding model, AURIX is not the answer. If you already use AI coding agents and find yourself moving between terminals, tools and notes, its pitch is to give those parts a common place to work.

A workspace around the agents

AURIX arranges agents in split terminal panes. The developer says you can see which agent is working and which is waiting for you, and bring a pane forward when you need it. A right-click menu is intended to handle practical pane actions such as splitting, resizing, opening full screen or sending a saved prompt. The interface is presented as a way to manage concurrent work rather than as a new coding environment with its own agent.

You are also meant to set the model and effort for each workspace in advance. The stated aim is to keep an agent from changing those choices on its own. That could make a collection of agent sessions easier to supervise, but it does not remove the need to decide what work to delegate or to review what comes back. “Run the show” still means that the operator is responsible for the work.

AURIX is described as working with coding agents rather than replacing them. The landing page names Claude, Codex and pi, but it does not explain compatibility in detail or show the setup process. That leaves an important practical question for anyone relying on a particular agent: the broad idea is clear, while the exact boundaries of support are not.

Shared context, tools and checks

The central problem AURIX claims to address is work getting lost between agents. It says decisions and tasks are written down and checked off so that another agent can start with those facts. This is a shared record, not proof that agents will interpret every note correctly. Teams would still need to provide clear decisions and check that the handoff makes sense.

The same shared approach applies to tools. The developer says an MCP plugin is installed once and then made available to every agent. That makes AURIX relevant to people looking for an MCP plugin manager as part of a multi agent terminal setup. The page does not describe which plugins are supported or how permissions are handled, so those details remain unknown.

There is also a browser pane. Agents can open pages and interact with them while you watch and step in. The landing page adds that you can ask about the current page or have AI fill a form, with you sending it yourself. This points to browser-based checking and interaction, not just code generation. It also keeps the human in the loop for the final form submission, a useful boundary where an automated action could have consequences.

Visibility, with a waitlist caveat

AURIX says it tracks token use by day, workspace, model and task. For people running several AI coding agents, that could help attribute usage instead of treating it as one undifferentiated bill. The claim is about visibility; the page gives no example screen or detail about how the ledger handles different providers, so it is too early to judge how useful the accounting will be in practice.

A voice shortcut called Friday is meant to send a job to an agent and notify you when it is done. This is another way to delegate, not a substitute for choosing the task or checking the result. In combination with the shared notes and panes, the product’s shape is a control surface for coordinating work, rather than an autonomous team that can be left unattended.

AURIX is not open yet. The current route is a waitlist, and the app is specified for Apple Silicon Macs with M1 or newer, not Intel Macs or other platforms. The listing classifies it as paid, while its page says pricing has not been decided. That makes it hard to assess the eventual cost or try the workflow today.

AURIX is for Mac developers already using multiple coding agents who want shared task context, tools and usage visibility in one place. It is not for someone looking for a standalone coding assistant, a Windows or Intel Mac app, or a product they can evaluate now. The AURIX waitlist is the only concrete next step until access opens.