OpenAI has launched a new macOS app for its Codex platform, introducing a desktop interface designed to help users manage multiple AI agents at once. According to
OpenAI, the app is built to run tasks in parallel, supervise long-running work, and provide a central space to collaborate with agents across coding and other computer-based tasks.
The company said the Codex app functions as a command centre for agents and builds on Codex’s existing availability through the Command Line Interface (CLI), Integrated Development Environment (IDE) extensions, and cloud-based tools.
What is OpenAI’s Codex app
The Codex app is a desktop interface that allows users to work with multiple AI agents simultaneously. Instead of interacting with a single agent, users can delegate tasks across separate agents organised by project, with each agent running in its own thread.
OpenAI said the app is designed to support complex, long-running tasks that may span hours, days, or weeks. Users can switch between agent threads without losing context, review changes made by agents, comment on diffs, or open files in their editor to make manual edits.
The app also carries over session history and configuration from the Codex CLI and IDE extensions, allowing users to continue working with existing projects without reconfiguration.
How it works
According to OpenAI, each agent in the Codex app works in an isolated environment, which is intended to prevent conflicts when multiple agents operate on the same codebase. Built-in support for worktrees allows agents to work on separate copies of a repository, enabling parallel exploration of different approaches without affecting the main code.
Beyond code generation, Codex supports “skills,” which allow agents to perform tasks such as gathering and synthesising information, writing content, creating documents, or interacting with external tools. Skills bundle instructions, resources, and scripts, and can either be explicitly selected by users or automatically used by Codex based on the task.
OpenAI demonstrated this capability by having Codex independently build and test a 3D racing game using a combination of skills, with the agent handling design, development, and testing over a long-running session.
The app also includes Automations, which let users schedule recurring tasks to run in the background. When an automation completes, its output appears in a review queue, allowing users to inspect results or continue work as needed.
On security, OpenAI said the Codex app uses system-level sandboxing by default. Agents are limited to editing files within their assigned folders or branches and can only access cached web search unless given permission. Commands requiring elevated permissions, such as network access, prompt user approval, with configurable rules available for teams or projects.
Why it matters to you
Although the Codex app is primarily aimed at developers, it reflects how AI tools are being designed to handle more complex and long-running tasks with less direct human involvement. OpenAI’s focus on parallel agents, background automations, and task-specific skills shows how software can increasingly be built, tested, and maintained with AI systems working continuously behind the scenes.
For consumers, this shift can influence how quickly apps and digital services are updated, fixed, or expanded, as development teams are able to offload routine and time-consuming work to AI agents.
Availability
The Codex app is available on macOS. Users with ChatGPT Plus, Pro, Business, Enterprise, or Edu subscriptions can access Codex across the app, IDE extensions, CLI, and the web using their ChatGPT login.
OpenAI also said that, for a limited period, Codex will be available to ChatGPT Free and Go users. During this time, rate limits for paid plans are being doubled, with the option to purchase additional credits if required.
 "OpenAI releases Codex app for Mac users: Here's what it is and how it works")
