The promised news that blew my own mind.
Remember how I made Gennady and Fyodor communicate via Discord? Well — I threw out all the unnecessary stuff.
Deleted OpenClaw. Deleted nanobot. All the routine work with exchanges now goes directly through Claude Code — without AI intermediaries. It works with the browser itself, more reliably and faster than anything before, and objectively makes better decisions.
But there was one problem: Claude Code during complex debugging requires my involvement. I have to sit and make obvious decisions. Got tired of it.
And here's what I came up with.
On the local laptop — Claude Code ("Fyodor"). Works with exchanges and websites, learns, saves knowledge.
On the main work laptop — second Claude Code. I taught it to log into Fyodor's laptop and write directly to its console. As if I'm doing it.
I literally watch as one writes to the other, reads responses, thinks — and completes the task to the end. Without me.
They communicate not with routine commands — with full-fledged text, like people. They find the problem themselves, make decisions themselves, solve it themselves.
Two Claude Code. SSH as a bus. No middleware. Works.
