Set up GBrain Cloud from Claude Code or Codex

This week we launched gbrainio, the command line for GBrain Cloud. It launches a GBrain in the cloud, connects your terminal to it, fills its memory and gives it work on a schedule.

You don’t have to be the one typing: Claude Code, Codex and any coding agent that runs commands can run it too. The quickest start is to hand it to one:

# Claude Code
$ claude "Read gbrain.io/llms.txt and set me up"

# Codex
$ codex "Read gbrain.io/llms.txt and set me up"

What it reads

gbrain.io/llms.txt is a page written for an AI. It walks your assistant through:

1. Installing the command line.
2. Signing you up. A page opens in your browser, where you sign in and start a workspace. You paste one command back.
3. Filling your memory from the notes folder you name.
4. Scheduling work, like a morning brief, if you want it.
5. Working from the memory: searching before it answers, saving what you decide.
6. Setting up GBrains for others: a team, a project or a customer.

Signing in, paying and deciding what anything may reach stay with you, in your browser.

The command line or MCP, not both

An AI that runs commands, like Claude Code, Codex or Cursor’s agent, uses the command line. One that can’t, like ChatGPT or Claude in a chat, connects over MCP, as the works-with pages show. The page has your assistant pick one, so the same memory never has two ways in.

Or by hand

$ curl -sSL https://gbrainio.terminalwire.sh | bash
$ gbrainio terminal connect

Then run gbrainio to see what’s next.