What is the benefit of using a Team Lead agent?

A Team Lead agent acts as an orchestrator, coordinating between implementers and reviewers to ensure clear boundaries and reduce duplicate work.

How can I see what all my agents are doing at once?

Configure your 'teammateMode' to 'tmux' or 'iterm2' in your Claude settings to view each agent's activity in separate panes or tabs.

Which agent types are safe for code exploration?

Use the 'Explore' or 'Plan' agent types. These are read-only and cannot modify your files, making them ideal for research and analysis.

How do I decide how many agents to spawn for a task?

Follow the sizing heuristics: 1-2 agents for simple bugs, 2-3 for medium features, and 4-5 for complex, full-stack migrations or systemic issues.

AI Team Composition Patterns

Name: AI Team Composition Patterns
Author: gwickman

bygwickman

0•

生産性とワークフロー

Orchestrates multi-agent workflows by providing optimized team sizes, specialized agent roles, and preset configurations for complex development tasks.

This skill provides a comprehensive framework for structuring AI-powered agent teams within Claude Code, ensuring optimal performance for varied software engineering tasks. It offers battle-tested sizing heuristics and specialized configurations—such as dedicated Review, Debug, and Fullstack teams—to match task complexity with the right number and type of AI agents. By defining clear roles and communication patterns, it minimizes coordination overhead while maximizing the throughput of parallelized workflows across frontend, backend, and security domains.

主な機能

01Read-only agent configurations for risk-free code exploration and architectural planning.

02Optimized team sizing heuristics ranging from 1 to 5 agents based on task complexity.

030 GitHub stars

04Support for multi-pane display modes like tmux and iTerm2 to monitor agents in parallel.

05Pre-configured team presets for Security Audits, Migrations, and Research.

06Specialized agent types including Lead, Reviewer, Implementer, and Debugger roles.

ユースケース

01Conducting multi-dimensional security audits using specialized reviewer agents for OWASP and access control.

02Debugging complex systemic issues by deploying a team of investigators to test competing hypotheses.

03Scaling from a single assistant to a coordinated multi-agent team for large-scale feature development.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add gwickman/auto-dev-test-target-1 team-composition-patterns

For use in Claude.ai and ChatGPT

Download Skill