How does Swarm Orchestration handle task dependencies?

It builds a Directed Acyclic Graph (DAG) from your YAML manifest. Tasks with no dependencies start immediately in parallel, while dependent tasks wait until their prerequisites have successfully generated a Pull Request.

When should I use Swarm Orchestration instead of a standard autotask?

Use Swarm Orchestration when you have 3 or more independent or semi-dependent development tasks, such as during sprint planning or large refactors. For single tasks, the standard /autotask command is more efficient.

Where is the state of the swarm stored?

The primary source of truth is stored in Git branches. However, a local .swarm/state.json file tracks manifest progress, allowing you to resume the orchestration after an interruption.

What happens if one agent fails during a swarm execution?

Task failures are isolated. Independent tasks will continue to run, while only the tasks specifically depending on the failed worker will be blocked. You can then use retry or skip commands to recover.

Swarm Orchestration

Name: Swarm Orchestration
Author: Light-Brands

byLight-Brands

0•

部署与 DevOps

Coordinates distributed AI agent task execution through declarative manifests and dependency-aware scheduling.

Swarm Orchestration enables developers to scale their Claude Code workflows by distributing multiple coding tasks across a network of agents or virtual machines. It utilizes a YAML-based manifest to define task dependencies in a Directed Acyclic Graph (DAG), ensuring that independent features are implemented in parallel while dependent tasks wait for their prerequisites. By using Git branches as the source of truth and maintaining minimal state, it provides a robust framework for large-scale refactoring, sprint-based parallel development, and automated failure recovery without overloading local resources.

主要功能

010 GitHub stars

02Automated failure isolation and recovery protocols

03Declarative YAML task manifests

04DAG-based dependency resolution and scheduling

05Real-time progress monitoring and state persistence

06Parallel task execution across stateless agents

使用场景

01Automating sprint ticket implementation by distributing independent features to parallel agents

02Integrating parallel AI coding workflows into CI/CD pipelines for high-throughput development

03Executing large-scale refactors across multiple codebase modules simultaneously

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add light-brands/brand-factory swarm-orchestration

For use in Claude.ai and ChatGPT

Download Skill