Can Gas Town run tasks in parallel?

Yes, through 'Formula Convoys,' Gas Town can execute multiple tasks in parallel and then synthesize the results into a single output.

What is the 'Hook' in Gas Town?

The Hook represents an agent's current assignment. Under the GUPP principle, if an agent finds work on its hook, it must execute it immediately without waiting for human confirmation.

How does Gas Town handle complex workflows?

It uses 'Molecules,' which are multi-step workflow instances built from 'Formulas,' allowing developers to define and execute structured, repeatable AI processes.

What is the primary purpose of Gas Town?

Gas Town is designed to coordinate and manage multiple Claude Code agents simultaneously, allowing for complex, parallel, and resilient development workflows across various projects.

What is the difference between gt and bd commands?

The gt command handles agent-level operations like checking hooks and convoys, while bd manages data and task-level operations like creating and closing 'beads' (tasks).

Gas Town Orchestrator

Name: Gas Town Orchestrator
Author: kyletabor

bykyletabor

0•

Productivity & Workflow

Orchestrates multiple Claude Code agents through a structured multi-agent coordination system for complex, cross-project workflows.

Gas Town is a sophisticated multi-agent orchestration framework designed to scale Claude Code operations across large-scale projects. It introduces a unique "Propulsion Principle" and the "Hook Contract" to ensure continuous progress without constant human intervention. By organizing work into atomic "Beads" and structured "Molecules," and employing specialized roles like Mayors, Polecats, and Witnesses, Gas Town enables parallel execution, automated task handoffs, and resilient workflow management for high-concurrency development environments.

Key Features

010 GitHub stars

02Automated task execution via the 'Hook' system for immediate AI action

03Granular tracking of atomic work units within larger multi-step Molecules

04Role-based architecture including Mayors, Polecats, and Witnesses for distributed oversight

05Scalable parallel processing with Formula Convoys for complex synthesis tasks

06Multi-agent coordination using the gt and bd command-line interfaces

Use Cases

01Automating multi-step R&D workflows that require synthesis from multiple perspectives

02Orchestrating background maintenance and code reviews across a fleet of AI instances

03Managing large-scale code migrations across multiple repositories using parallel agents

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add kyletabor/claude_plugins gastown

For use in Claude.ai and ChatGPT

Download Skill