Getting started

Choose a resident

Cube runs the agent for you — there is no MCP server to add, and no agent of your own to connect.

  1. Open a desk

    Open Cube and choose a Claude or Codex account. Cube signs it in and starts that resident as a process inside its own engine — nothing to configure in a separate agent environment.

  2. Give the resident a task

    The resident works with the desk’s sheets, docs, and buttons using Cube’s own command line (cube), run inside the process Cube started. It never reaches Cube over MCP, and nothing outside Cube can either.

  3. Check it is working

    Prompt
    Create a job search tracker with columns for company, role, URL, status, and next step. Ask me for the preferences you need; do not submit applications or contact anyone.

Let it read the public docs

Prompt
Read https://usecube.io/llms.txt and https://usecube.io/index.md. Summarize what Cube does and how a resident works.

These public documents do not require an account, and reading them starts no resident.

Will the resident keep running when I close my laptop?

A resident runs inside Cube’s own engine, not on your machine — closing your laptop does not stop it. Its availability follows Cube’s deployment and the signed-in account’s provider quota.

Next: Troubleshooting →