What is a git worktree in the context of this skill?

A git worktree allows you to have multiple branches checked out simultaneously in separate directories. This skill uses them to give each AI agent its own isolated workspace, preventing them from overwriting your current work.

Can I monitor what the spawned agents are doing?

Yes, you can use the 'agent-cli dev status' command to see all active worktrees and 'agent-cli dev run cat .claude/REPORT.md' to view specific task summaries.

Does this skill work with agents other than Claude?

Yes, the skill supports various agent backends including Aider and others via the --with-agent flag, allowing you to choose the best tool for the specific task.

Why is the --prompt-file option recommended over --prompt?

Using a file avoids common shell quoting issues with special characters like exclamation marks, dollar signs, and backticks that often cause errors in ZSH or Bash when passing long instructions.

Parallel AI Agent Orchestrator

Name: Parallel AI Agent Orchestrator
Author: basnijholt

bybasnijholt

•

103

生产力与工作流

Spawns and manages multiple AI coding agents in isolated git worktrees for high-concurrency development workflows.

关于

The agent-cli-dev skill empowers developers to scale their productivity by delegating independent tasks to multiple AI agents simultaneously. By leveraging git worktrees, it creates isolated environments for each agent, preventing code conflicts and allowing for massive parallelization of feature development, refactoring, and testing. It automates the entire lifecycle of agent orchestration—from workspace setup and dependency installation to status monitoring and standardized report generation—enabling a true 'manager-of-agents' workflow.

主要功能

Utilize prompt files to bypass shell quoting issues with complex instructions
103 GitHub stars
Spawn parallel AI agents in isolated git worktrees to prevent workspace conflicts
Centralize status monitoring for all active agents and their respective branches
Automate project setup and dependency installation for every new agent task
Generate standardized progress reports in .claude/REPORT.md for easy review

使用场景

Conducting parallelized Test-Driven Development where separate agents handle tests and implementation
Implementing multiple independent features or modules concurrently to accelerate delivery
Executing large-scale refactoring by splitting the workload across different architectural layers

关于

主要功能

Utilize prompt files to bypass shell quoting issues with complex instructions
103 GitHub stars
Spawn parallel AI agents in isolated git worktrees to prevent workspace conflicts
Centralize status monitoring for all active agents and their respective branches
Automate project setup and dependency installation for every new agent task
Generate standardized progress reports in .claude/REPORT.md for easy review

使用场景

Conducting parallelized Test-Driven Development where separate agents handle tests and implementation
Implementing multiple independent features or modules concurrently to accelerate delivery
Executing large-scale refactoring by splitting the workload across different architectural layers