What triggers the orchestration engine?

The engine triggers on keywords like 'parallel', 'collaboration', or 'cross-layer', through the /team command, or when tasks involve multi-perspective analysis like code reviews.

How does the skill handle file editing conflicts?

It implements a 'File Ownership' principle where each agent is assigned specific directories or modules, preventing multiple agents from concurrently editing the same file.

How many agents should be in a team for optimal performance?

The orchestrator recommends 2-3 agents for simple tasks and 3-5 for medium complexity. Teams larger than 7 are generally discouraged due to increased coordination overhead.

When should I use Agent Teams instead of a single Subagent?

Use Agent Teams when a task requires multi-angle parallel analysis, frequent inter-agent communication/critique, or clear context isolation across different architectural layers.

Is it possible to monitor the progress of individual agents?

Yes, you can monitor status using Ctrl+T and manage agent interactions via the Lead agent to ensure all tasks are synchronized before finalization.

Agent Team Orchestrator

Name: Agent Team Orchestrator
Author: NanmiCoder

byNanmiCoder

•

コラボレーションツール

Orchestrates specialized AI agent teams using an intelligent decision engine to solve complex, multi-dimensional development tasks.

The Agent Team Orchestrator is a sophisticated decision engine for Claude Code designed to optimize multi-agent workflows. It automatically evaluates tasks across five critical dimensions—parallelism, communication needs, context isolation, file conflict risk, and cost-efficiency—to determine whether a single subagent or a coordinated team of specialized agents is required. By providing structured templates for scenarios like multi-angle code reviews, competitive hypothesis debugging, and cross-layer feature development, it empowers developers to scale their productivity while maintaining strict control over agent roles, file ownership, and resource cleanup.

主な機能

01Pre-built prompt templates for PR reviews, debugging, and architectural research

02Automated 5-dimension task complexity analysis and workflow recommendation

03Structured team design framework for optimal role allocation and scaling

04Lifecycle management protocols for agent synchronization and resource cleanup

0536 GitHub stars

06Conflict prevention strategies to manage concurrent file modifications

ユースケース

01Conducting multi-layered security, performance, and logic audits on complex PRs

02Coordinating front-end, back-end, and test engineers for full-stack feature implementation

03Executing systematic 'Competing Hypotheses' debugging for intermittent or complex system bugs

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add nanmicoder/claude-code-skills agent-team-orchestrator

For use in Claude.ai and ChatGPT

Download Skill