Descriptions:
David Ondrej walks through tmux — the terminal multiplexer — as a foundational tool for anyone running AI coding agents at scale. Unlike casual vibe-coders who run agents locally on a laptop, Ondrej argues that serious agentic engineers need tmux on a VPS so agents can operate for hours or days without depending on a connected machine. The video covers installation via Homebrew, the three core abstractions (sessions, windows, and panes), key shortcuts for detaching and reattaching sessions, and how tmux enables agents to programmatically read and write across terminal panes — something impossible with separate standalone terminals.
The tutorial then moves to the real-world setup: spinning up a VPS, SSH-ing in, and installing both Codex CLI (OpenAI) and Claude Code inside tmux sessions. Ondrej demonstrates splitting panes to run multiple agents simultaneously and shows how to configure mouse support for easier navigation. He references how prominent AI developers — including Peter Steinberger (creator of OpenClaw) and Mario Zakner (creator of Pi) — already rely on this exact workflow.
Viewers will come away with a step-by-step setup they can replicate immediately, whether on macOS, Linux, or Windows (with a tip to use Claude or ChatGPT for platform-specific variations). The video is especially relevant for developers running long-horizon tasks with tools like Codex CLI’s `/go` mode or Claude Code’s autonomous refactoring flows.
📺 Source: David Ondrej · Published May 25, 2026
🏷️ Format: Tutorial Demo
