Skip to content
JournalTORIS / FIELD NOTES

← Blog AI insights

Picking Up Browser Work from Claude Code in Codex with ego lite

How ego lite's Space keeps the browser screen alive so Codex can pick up where Claude Code left off, seen from a solo developer's point of view.

3 min read Read the Korean original

Joohwan Yu · Founder, Full Stack Developer

Toris · Published

Contents

Say you are working with Claude Code and hit a usage limit. If you were editing code, you can open Codex and tell it to start with git diff. But what if you were just checking sign-in settings in the Google console? You’d have to explain again which account you signed in with, which menu you opened, and how far you had checked.

What caught my eye about ego lite is the Space, which can be used in exactly this kind of situation. Codex can pick up the work there while keeping the browser screen Claude Code had open. As someone who runs apps and web services alone, I find that a pretty welcome possibility.

An illustration of two AI agents taking over a browser screen while a developer works

An AI-generated illustration of an agent taking over browser work. It is not an actual product screen.

Keep Working While the Agent Works

ego lite is a Chromium-based browser, like Chrome. If you choose to import Chrome data on first launch, you can keep using your existing bookmarks, extensions, sign-in state, and more. For anything you want done on the web, you just tell the Claude Code or Codex you normally use. ego lite official docs

A Space separates the area you use from the area the agent works in, within the same browser. You can hand off several investigations or checks, and keep doing what you were doing in your own space. ego lite’s Space explanation

Seeing this feature reminded me of my blog. Before publishing a post, I need to check whether tables get cut off on mobile and whether the links in the body open properly. If I could hand that to Codex and keep editing the paragraph I was writing, that is reason enough to use it even without running many tasks at once.

Hand Over the Screen, Leave One Line of Context

For example, you can ask Claude Code to check the Google sign-in settings, with the condition that it stop before pressing the publish button. If passkey authentication is needed along the way, I can authenticate myself and then let it continue.

Then say Claude Code’s limit runs out and you switch to Codex. What Codex can pick up is the Space that was open. It can use the console screen and sign-in state to look at the remaining items, but the conversation with the earlier agent isn’t passed along automatically. You have to tell it separately how far you had checked.

If it were me, I’d leave a note like this for the next agent: “This is the screen where I was checking Google sign-in settings. I’ve checked up to the test user list, so just sort out what’s left before going public. I’ll press the publish button myself.” If you can hand it off like this, you don’t have to go looking again for menus you already checked.

For My Own App, I’d Want to Try It Right Before Release

The job I’d want to hand off first is the pre-release check for an Android app. “Check in the Play Console whether anything is blocking this version’s release, and list only what I need to fix.” In the meantime, I can run the last build once more on a real device. If, when I come back from finishing the test, the items to check and their screens are both still there, I think that would be the most practical reason to use ego lite in my own work.