Dual Brain: two AI agents on one job

What is Dual Brain for AI agents?

Many people already run Claude Code and Codex side by side in two terminals and carry the results across by hand. Dual Brain turns that into one setup: Sno Station gives the two AI agents one shared memory and a mailbox, so either can hand work to the other without you copying anything.

Sno Station runs on your own machine, with no daemon and no server. The product name for the paired setup is Duo Agent.

Why should a second AI agent review the first one's work?

AI agents from one lab share blind spots, so a review from the other lab catches what the first would miss. The reviewer reads the change itself, not the name on it, and it is always an agent from the other harness.

A wrong plan caught at step three costs three steps, not forty. Two agents cost less than one that has to redo the work.

What happens when one AI agent runs out of quota?

Tell the other agent to take over (say sno reach call from the other one). It picks up the task from the shared memory and the mailbox, on the same line of work, with the same memory and the same branch, so the job does not stop. Sno Station calls this Rotation.

Solo or Duo Workstation: which one do I need?

Start with the smaller setup if you run one agent today, and add the second agent when you want review and takeover.

How do I start with Dual Brain?

Install Sno: tell your AI agent "install sno.ai", or run curl -fsSL https://sno.ai/install | sh. Then ask either agent to review the other's work.