parallel-executor

Orchestrate parallel Claude Code sessions using Git worktrees and detached agents.

Updated Jan 14, 2026
One-click install
npx skills add https://github.com/kingpinzs/claude_parallel_templates --skill parallel-executor
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: parallel-executor
Source: https://github.com/kingpinzs/claude_parallel_templates/tree/main/base/.claude/skills/parallel-executor
Command: npx skills add https://github.com/kingpinzs/claude_parallel_templates --skill parallel-executor

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires jq, git, claude, flock.

What problem does it solve?

This Skill automates the setup and orchestration of parallel Claude Code sessions by creating isolated Git worktrees and launching detached agents, removing manual task distribution and coordination.

Core Features & Use Cases

  • Automated parallel session spawning: creates separate worktrees and agents per task.
  • Persistent planning and crash recovery: uses a shared plan and checkpoint state to resume across sessions and machines.
  • Cross-project orchestration: monitor, merge, and clean up worktrees post-execution.

Quick Start

  1. Spawn tasks: ./spawn.sh "Task A" "Task B" "Task C"
  2. Monitor progress: tail -f logs/*.log
  3. Start orchestrator: ./orchestrate.sh
  4. Merge results: ./merge.sh --cleanup

Frequently Asked Questions about parallel-executor

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

FAQPage Schema
How do I run multiple Claude Code sessions in parallel?

You can run multiple Claude Code sessions in parallel by using an orchestrator that creates isolated Git worktrees and spawns detached agents for each independent task. This removes manual task distribution and coordination overhead.

Can Claude Code resume parallel tasks after a crash?

Yes, parallel tasks can resume after a crash by using a shared persistent plan and checkpoint state. This cross-session continuity allows detached agents to recover progress across different machines and sessions.

What dependencies do I need to automate parallel Claude Code agents?

Automating parallel Claude Code agents requires jq for state management, Git for worktree isolation, the Claude CLI for headless execution, and flock for safe concurrent process locking.

How do I merge results from parallel Git worktrees after execution?

You merge results from parallel Git worktrees by running a dedicated merge script post-execution. This script integrates the isolated branches back together and optionally cleans up the temporary worktrees.

What are the limitations of orchestrating parallel Claude Code sessions?

The primary limitation is task independence; the orchestrator enforces safe task independence and optional scope constraints, meaning tightly coupled tasks with shared dependencies cannot run concurrently.

Is there a way to monitor progress of detached Claude Code agents?

Yes, you can monitor progress of detached Claude Code agents by tailing the individual log files generated for each session, allowing you to track real-time execution across all parallel worktrees.