TL;DR
Gas Town is a Go-based orchestrator enabling developers to manage 20-30 parallel AI coding agents productively using tmux, built on the Beads framework.
Key Points
- Orchestrator manages seven distinct agent roles (Mayor, Polecats, Refinery, Witness, Deacon, Dogs, Crew) across multiple project rigs
- Handles complex multi-agent coordination including merge queue management, work swarming, and graceful degradation
- Solves MAKER benchmark (Tower of Hanoi) with 1M-step workflows; prior research showed LLMs fail after hundreds of steps
- Codebase is 3 weeks old, 100% 'vibe coded', requires Stage 7+ AI-assisted development experience to use effectively
Why It Matters
As AI coding agents become commoditized, orchestration is the next frontier for scaling AI-assisted development. Gas Town demonstrates practical patterns for managing dozens of parallel agents at production scale, addressing the coordination and merge conflict problems that emerge when scaling beyond single-agent workflows.
Source: steve-yegge.medium.com