OpenAI is giving Codex a much better view of the web apps it is supposed to help build.
The company’s developer account announced a new Chrome plugin for Codex that can test web apps, pull context from open tabs and use Chrome DevTools while the user keeps working. That sounds small, but it changes Codex from something that only reasons about files into something that can also see how a project behaves in the browser.
That matters because modern app work does not live entirely inside an editor. Bugs show up in routes, UI states, responsive layouts, console errors and weird browser behavior. If Codex can inspect those pieces without taking over your session, it becomes more useful for the boring but important work of checking whether something actually runs.
OpenAI brought Codex to macOS earlier this year and has been steadily adding features around multi-agent coding and app workflows. The Chrome plugin also hints at a bigger direction: Codex could become less of a developer-only coding assistant and more of a general browser work agent for people who spend most of their day in web tools.
That future will depend on trust. A browser plugin that can gather tab context and touch DevTools needs clear boundaries. But for developers, this is exactly the kind of practical AI feature that makes more sense than a floating chatbot with no idea what your app is doing.
