Which spawn backends are supported for orchestration?

Claude Code auto-detects the environment and supports 'in-process' (invisible), 'tmux' (separate panes), and 'iterm2' (split panes) backends for running teammates.

How do agents in a swarm communicate with each other?

Agents communicate through structured JSON messages stored in dedicated inbox files (~/.claude/teams/{name}/inboxes/), allowing for status updates, data sharing, and shutdown requests.

What is the difference between a Subagent and a Teammate?

Subagents are short-lived, focused workers that return a result and exit. Teammates are persistent agents that join a named team, appear in the team configuration, and can claim multiple tasks from a shared list.

Can I use specialized agents for different parts of a swarm?

Yes, the orchestration system supports various agent types like 'Bash', 'Explore', 'Plan', and specialized plugin agents like 'security-sentinel' or 'architecture-strategist'.

What is a Claude Code Swarm?

A swarm is a group of specialized AI agents, known as teammates, coordinated by a leader agent to perform complex, parallel tasks using a shared task management system.

Swarm Orchestration

Name: Swarm Orchestration
Author: The-Rabak

byThe-Rabak

•

협업 도구

Orchestrates multi-agent swarms using Claude Code's TeammateTool and Task system for complex, parallel engineering workflows.

The Swarm Orchestration skill provides a comprehensive framework for managing multi-agent environments within Claude Code. It enables developers to coordinate a lead agent with multiple specialized teammates, leveraging a structured task system with dependencies and automated unblocking. By utilizing integrated inboxes and standardized message formats, this skill facilitates sophisticated divide-and-conquer patterns, allowing for parallel code reviews, automated migration pipelines, and self-organizing work queues that dramatically increase development velocity and thoroughness.

주요 기능

01Support for persistent teammates versus short-lived subagents based on workload requirements.

02Comprehensive multi-agent lifecycle management from team creation to automated shutdown.

03Structured inter-agent communication via dedicated JSON inboxes for seamless coordination.

041 GitHub stars

05Shared task system with JSON-based tracking, owner assignment, and dependency mapping.

06Compatibility with multiple execution backends including in-process, tmux, and iTerm2.

사용 사례

01Coordinating parallel security, performance, and architectural reviews for large pull requests.

02Building complex software migration workflows where tasks are blocked by multi-step dependencies.

03Automating large-scale codebase research where specialized agents scan for different patterns simultaneously.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add the-rabak/compound-engineering-plugin orchestrating-swarms

For use in Claude.ai and ChatGPT

주요 기능

01Support for persistent teammates versus short-lived subagents based on workload requirements.

02Comprehensive multi-agent lifecycle management from team creation to automated shutdown.

03Structured inter-agent communication via dedicated JSON inboxes for seamless coordination.

041 GitHub stars

05Shared task system with JSON-based tracking, owner assignment, and dependency mapping.

06Compatibility with multiple execution backends including in-process, tmux, and iTerm2.

사용 사례

01Coordinating parallel security, performance, and architectural reviews for large pull requests.

02Building complex software migration workflows where tasks are blocked by multi-step dependencies.

03Automating large-scale codebase research where specialized agents scan for different patterns simultaneously.