multi-swarm

Orchestrate parallel Claude Code sessions with independent agent teams and git worktrees.

Updated Aug 27, 2026
One-click install
npx skills add https://github.com/itsgaldoron/multi-swarm --skill multi-swarm
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: multi-swarm
Source: https://github.com/itsgaldoron/multi-swarm/tree/main/plugins/multi-swarm/skills/multi-swarm
Command: npx skills add https://github.com/itsgaldoron/multi-swarm --skill multi-swarm

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires tmux, jq, litellm, gh cli, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill automates the parallel execution of multiple Claude Code sessions, enabling complex tasks to be broken down and tackled concurrently by independent agent teams.

Core Features & Use Cases

  • Parallel Orchestration: Run N independent Claude Code sessions (swarms) simultaneously.
  • Agent Teams: Each swarm can have its own team of agents for specialized tasks.
  • Git Worktrees: Each swarm operates in its own isolated git worktree.
  • LiteLLM Gateway: Manages API token rotation for cost-effective scaling.
  • Use Case: Develop a new feature by assigning different parts (e.g., backend API, frontend UI, tests) to separate swarms, drastically reducing development time.

Quick Start

Use the multi-swarm skill to add user authentication with login, signup, and password reset.

Frequently Asked Questions about multi-swarm

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run parallel Claude Code sessions for concurrent development?

You can run parallel Claude Code sessions by orchestrating multiple swarms with independent agent teams. This Skill automates concurrent execution by assigning separate swarms to isolated git worktrees, drastically reducing development time for complex tasks.

What dependencies do I need to orchestrate Claude Code swarms with git worktrees?

Orchestrating Claude Code swarms requires Claude Code, tmux, jq, LiteLLM, and gh CLI. Tmux manages parallel sessions, while LiteLLM handles API token rotation for cost-effective scaling across multiple agent teams.

Can I use LiteLLM to manage API token rotation for parallel AI agent teams?

Yes, LiteLLM manages API token rotation for parallel AI agent teams. It acts as a gateway for cost-effective scaling, allowing multiple independent Claude Code swarms to execute concurrently without hitting individual token limits.

What is the best way to break down a large feature into concurrent AI agent tasks?

The best way to break down a large feature is assigning different parts to separate swarms. For example, you can allocate backend API, frontend UI, and tests to independent agent teams operating in isolated git worktrees for parallel execution.

Does parallel orchestration of Claude Code sessions work with isolated git worktrees?

Yes, parallel orchestration works by having each swarm operate in its own isolated git worktree. This ensures that concurrent Claude Code sessions can execute complex task decomposition independently without causing code conflicts.

When should I avoid using parallel swarms for task decomposition?

You should avoid using parallel swarms if your environment lacks the required tmux, jq, LiteLLM, and gh CLI dependencies. Additionally, tasks that require strict sequential execution or have heavy interdependencies are not suited for this concurrent approach.