parallel-worktrees

Coordinate multiple Claude Code agents across isolated git worktrees for parallel development.

12|Updated Dec 23, 2025
One-click install
npx skills add https://github.com/SpillwaveSolutions/parallel-worktrees --skill parallel-worktrees
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: parallel-worktrees
Source: https://github.com/SpillwaveSolutions/parallel-worktrees/tree/main
Command: npx skills add https://github.com/SpillwaveSolutions/parallel-worktrees --skill parallel-worktrees

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

The skill helps you generate and compare multiple independent code solutions for the same task by leveraging git worktrees as isolated execution environments for Claude Code agents.

Core Features & Use Cases

  • Parallel Development with Git Worktrees: Spawn multiple worktree-based branches so each agent works in isolation while sharing the same git history database.
  • Competitive Solutions via LLM Non-Determinism: Run N parallel agents to produce multiple viable implementations, then select the best one.
  • Background Agent Orchestration: Delegate subtasks to Claude Code background agents and coordinate completion using per-task status files and per-agent results documents.
  • Easy Comparison and Merging: Review diffs across worktrees and merge the winner back into your main branch.

Use case example: For a UI component redesign, you can run 3-4 agents in parallel worktrees, compare their implementations, and merge the best approach while reducing the risk of a single flawed solution.

Quick Start

Ask your Claude Code agent to run feature work in separate worktrees by instructing it to create parallel worktrees for the feature and then start Claude in each worktree session with identical instructions.

Frequently Asked Questions about parallel-worktrees

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

FAQPage Schema
How do I run multiple Claude Code agents in parallel for the same feature?

Run multiple Claude Code agents in parallel by spawning isolated git worktrees for the same task, allowing each agent to work on a separate branch while sharing the same git history database. This skill coordinates spawning, syncing, and cleaning up these worktrees for parallel development.

What is the best way to compare multiple code implementations for the same task?

Compare multiple code implementations by running N parallel agents in separate git worktrees to produce multiple viable solutions, then review diffs across worktrees and merge the best approach back into your main branch. This leverages LLM non-determinism to generate competitive solutions.

How do I orchestrate background Claude Code agents and track their completion?

Orchestrate background Claude Code agents by delegating subtasks and coordinating completion using per-task status files and per-agent RESULTS.md documents. The skill uses optional .agent-status JSON conventions alongside RESULTS.md for completion tracking and result collation.

Do I need a specific git setup to use parallel worktrees for development?

To use parallel worktrees for development, you need a root SKILL.md file with discovery frontmatter. The skill uses scripts to spawn, sync, and cleanup worktrees, relying on your existing git history database to allow multiple agents to work in isolation on the same repository.

When should I use parallel git worktrees instead of a single agent for feature implementation?

Use parallel git worktrees instead of a single agent when you want to reduce the risk of a single flawed solution, such as redesigning a UI component where running 3-4 agents in parallel allows you to compare implementations and select the best one before merging.