# Codex with a desktop that stays on: the MCP setup

*Kubi, cloud desktops for AI agents. https://trykubi.com/integrations/codex, last changed 2026-11-23. The page as Markdown; the same facts as the page itself.*

**What it is.** Codex is good at writing and running code and has nowhere to look at the result. Its own sandbox is made fresh and thrown away, so nothing it signed into is there next time. Two commands point it at a Kubi screen instead: a desktop with Chromium and Firefox, a disk that stays, and a person who can watch and take over.

Codex is OpenAI's coding agent, in the terminal or the cloud, from OpenAI. A screen is $20 a month at the founding price, or $0.035 an hour billed by the minute.

## What Codex gains

- **A machine that is not wiped between runs.** A persistent screen keeps its disk, its packages and its browser profile. Codex spends its first minute working instead of installing what it installed yesterday.
- **A browser for the half of the web that needs one.** Dashboards, consoles and anything behind a sign-in are reachable by clicking, which is often the only way they are reachable at all.
- **Runs you can name and look back at.** Send a run header with the calls and the work groups under one label, which beats reading a flat log of clicks when something went wrong.

## Set it up

1. **Add the server.** 

```
codex mcp add kubi --url https://mcp.trykubi.com
```

2. **Sign in and pick the screen.** The first call opens our sign-in page in your browser. Choose which screens the app may use and what it may do, then go back to the app. You can take it back at any time in the dashboard.

```
codex mcp login kubi
```

3. **Or use a key, for runs with nobody there.** For CI and servers, send a key instead of signing in. The key's own screens and permissions apply. Add a run label so the calls group under one name.

```
Authorization: Bearer kubi_...
Kubi-Run: nightly-checks
```


The server's address is https://mcp.trykubi.com. Each app's setup is kept current at https://docs.trykubi.com/mcp/connect.

## The first thing to try

"On my Kubi screen, open the staging site in Chromium, sign in with the saved profile, and screenshot the billing page so I can see what a new customer sees."

## What to know

- Without a run label, one MCP session counts as one run.
- The screen is Linux with a GNOME desktop. Code that needs Windows or a GPU has no home here.
- An act call carries up to 20 actions, and it is refused if anything changed the screen since the screenshot it names. That is on purpose: it is what stops an agent clicking where the button no longer is.

## Questions people ask

### How is this different from Codex's own sandbox?

Codex's sandbox is for running code and it is disposable. A screen is a desktop: a GUI it can see and click, a disk that persists, an address of its own, and a person who can watch live. You can also keep both, and use the screen only for what needs a screen.

### Does it work in CI with nobody watching?

Yes. Use an API key with the screens and permissions you want it limited to, rather than a sign-in, and name the run so the calls group together.

### What stops a runaway loop costing a fortune?

The price is the machine, not the calls: a persistent screen is $20 a month however hard it is driven. A disposable one is $0.035 an hour, and each organization has a cap on how many screens can exist at once.

## Read next

- [MCP: Connect](https://docs.trykubi.com/mcp/connect)
- [Errors and limits](https://docs.trykubi.com/mcp/errors-and-limits)
- [Billing and history](https://docs.trykubi.com/concepts/billing-and-history)
- [Pricing](https://trykubi.com/pricing)
- [Kubi compared](https://trykubi.com/compare)
- [Claude Code on a screen](https://trykubi.com/integrations/claude-code)
- [Cursor on a screen](https://trykubi.com/integrations/cursor)
- [your own code on a screen](https://trykubi.com/integrations/your-own-code)
