What roles can the Team Run skill create?

The skill can spawn specialized agents for implementation, research, testing, code review, and documentation, while the main Claude instance acts as the design lead and coordinator.

How does it handle dependencies between tasks?

It uses TaskCreate and TaskUpdate tools to define subtasks and establish 'blockedBy' or 'blocks' relationships, ensuring that work follows the correct logical order.

Can it utilize different AI models for specialized work?

Yes, it features fallback logic to leverage Codex for implementation and Gemini for deep research if the respective CLI tools are available in your environment.

How do the spawned agents communicate with each other?

Agents use the SendMessage tool to report progress, share research findings, and notify the lead of blockers or completed tasks.

Is Team Run suitable for small, single-file edits?

While functional, Team Run is optimized for large-scale tasks. For simple edits, standard Claude Code commands are more cost-effective and faster.

Team Run Orchestrator

Name: Team Run Orchestrator
Author: kentanakae

bykentanakae

0•

협업 도구

Orchestrates a multi-agent team of specialized AI roles to execute complex, large-scale development tasks through parallel coordination.

Team Run is a sophisticated orchestration skill designed for large-scale software projects that require more than a single agent's capacity. It transforms Claude into a project lead that spawns and manages specialized sub-agents—including Implementers, Researchers, Testers, Reviewers, and Documenters—to handle complex implementation, deep research, and quality assurance in parallel. By automating task decomposition, dependency tracking, and cross-agent communication, it ensures high-quality results for multifaceted engineering challenges while maintaining a clear, hierarchical workflow.

주요 기능

01Intelligent model fallbacks using Codex and Gemini for domain-specific tasks

02Automated task decomposition and dependency management with Task tools

03Seamless cross-agent communication via SendMessage for progress reporting

04Multi-role agent spawning for specialized task handling (coding, research, testing)

050 GitHub stars

06Parallel execution of implementation and research phases to save time

사용 사례

01Building complete features from scratch including research, implementation, and documentation

02Performing large-scale refactoring where coding and testing must happen simultaneously

03Executing comprehensive security audits and bug fixes across multiple modules

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add kentanakae/claude-code-plugins team-run

For use in Claude.ai and ChatGPT

Download Skill