How does the orchestrator track the status of work?

While it maintains a minimal local state file, it primarily uses Git branches and PRs as the source of truth for task completion and progress.

What happens if one agent fails during a swarm execution?

Task failure is isolated. Independent tasks continue to run, while only dependent tasks are blocked. You can then use retry or skip commands to handle the specific failure.

What is Swarm Orchestration in Claude Code?

Swarm Orchestration is a skill that allows you to distribute multiple development tasks across several agents or VMs, managing them as a single coordinated workflow.

How are task dependencies handled?

The skill uses a Directed Acyclic Graph (DAG) defined in a YAML manifest. Tasks only begin once all their specified dependencies have successfully submitted a pull request.

Is Swarm Orchestration suitable for single tasks?

No, it is designed for batches of three or more tasks. For single tasks, using the standard /autotask command directly is more efficient.

Swarm Orchestration

Name: Swarm Orchestration
Author: Light-Brands

byLight-Brands

0•

Productivity & Workflow

Orchestrates parallel development tasks across multiple agents using dependency-aware work manifests and automated recovery.

The Swarm Orchestration skill enables large-scale development automation by distributing multiple coding tasks across a network of autonomous agents or virtual machines. By utilizing a declarative YAML manifest, it builds a Directed Acyclic Graph (DAG) to manage complex task dependencies, ensuring that parallelizable work runs simultaneously while sequential tasks wait for required pull requests. This skill is particularly effective for high-volume workflows like sprint execution, multi-module refactoring, and batch feature development, providing robust failure recovery and stateless agent management where Git branches serve as the ultimate source of truth.

Key Features

01Automated failure recovery with task-level isolation and retries

02Real-time agent coordination and progress monitoring via HTTP

030 GitHub stars

04DAG-based dependency resolution for optimized parallel execution

05Declarative task manifests for simplified batch workflow definition

06Branch-based state persistence ensuring Git is the source of truth

Use Cases

01Parallelizing a full sprint of independent feature development tasks

02Automating complex CI/CD workflows that require multi-stage parallel tasks

03Executing large-scale codebase refactors across disconnected modules

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