My Agentic Coding Setup, July 2026
Domenic Denicola
A developer shares their Linux VM and Tailscale-based setup for running AI coding agents autonomously, combining remote VM access with ChatGPT's native SSH support for seamless work across devices. The setup uses a disposable Ubuntu VM on an always-on desktop with Tailscale for private networking, allowing agents to run with minimal approval prompts and continue work in the background. Key tools include git worktrees for parallel work, the ChatGPT desktop app as a thin client, and elevated agent permissions (sudoers, GitHub CLI login) that trade safety for convenience.
Why it matters
Running agents unattended with work that survives closing your laptop and follows you to your phone requires getting development off your local machine entirely, onto a disposable Ubuntu VM joined to a Tailscale tailnet. Safety is handled by blast radius rather than prevention: push to GitHub constantly, treat the VM as expendable.