3.1 IDE Extension: VS Code, Cursor, JetBrains

Everything so far has been the terminal, and the terminal is genuinely the right place for a lot of work — but it's not the only place Codex lives, and knowing when to switch surfaces is its own skill.

The IDE extension puts the same Codex agent directly inside VS Code, Cursor, or JetBrains, with one real advantage over the terminal: it can see what you're already looking at. You're mid-edit on a function, you highlight it, you ask Codex to refactor it, and it's working with the exact file and cursor context you're already staring at instead of you describing "the function in UserService around line 40." Local changes preview inline the same way any other edit in your editor would, so reviewing a suggestion feels like reviewing a colleague's inline comment rather than reading a diff block in a separate window.

This is the surface for tight edit-compile-test loops — the kind of back-and-forth where you're actively driving and Codex is more of a fast collaborator sitting in the same file than an independent agent running off on its own. Compare that to the terminal CLI, which is better suited to a task you're willing to describe once and let run — "add this feature," "fix these failing tests" — without you sitting there watching every keystroke land.

Neither surface is strictly better; they're suited to different moments in the same workday. A rough heuristic that holds up in practice: reach for the IDE extension when you're already in the file and want a fast collaborator, reach for the terminal when you're handing off a self-contained task and going to do something else while it works.