How I Turned Scattered Workstations into a Super Agentic Setup

Description and use

Distributing agents across mini‑PCs with Tailscale and Herder increased parallelism but raised cognitive load.

Problem

Managing a high‑throughput agentic workflow on a single machine caused crashes and limited scalability.

Outcome

The setup achieved high agent throughput, scalable compute addition, and centralized monitoring, but increased cognitive load and required rigorous human validation.

Workflow

• Set up a central terminal with Neovim and tmux for architecture review
• Connect mini‑PCs via GL.iNet KVMs and run 4 agents each
• Install Tailscale on all devices to create a private network
• Use Herder to manage tmux sessions and receive notifications
• Review agent code pushes with LazyGit or Hunk in Neovim
• Unblock agents or intervene when Herder alerts
• Add new PCs to the network as needed for scaling

Evidence

I use Herder. It’s team tmux plus notifications. When an agent node finishes its job or needs human intervention, it goes "DING!"

Source

Tailscale turns these scattered closet PCs, my development laptop, and even my phone into one private local network, anywhere.

Source

The Central Hub (Me): I sit in the Terminal (using Neovim and tmux).

Source