Can subagents use different AI models than the main agent?

Yes, you can configure subagents to use specific models like Claude 3 Haiku for speed and cost-efficiency or Claude 3.5 Sonnet for complex reasoning.

What is the main difference between a subagent and an agent team?

Subagents are specialized workers that report back to a single session, whereas agent teams are independent instances that can communicate directly with each other via a shared task list.

How do I enable the experimental agent teams feature?

You must set the CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS environment variable to '1' in your configuration settings.

Where should I store my custom subagent configuration files?

Custom agents can be stored in .claude/agents/ for specific projects, ~/.claude/agents/ for global use across all projects, or passed via the --agents CLI flag.

Agent Teams & Subagents Orchestrator

Name: Agent Teams & Subagents Orchestrator
Author: Aznatkoiny

byAznatkoiny

0•

Collaboration Tools

Orchestrates complex task delegation and parallel processing by configuring specialized subagents and collaborative agent teams within Claude Code.

This skill provides a comprehensive framework for scaling Claude Code's capabilities through advanced delegation patterns. It guides users in defining specialized subagents with restricted toolsets and specific model configurations for focused tasks like code review or documentation, while also facilitating the orchestration of experimental agent teams for large-scale, collaborative engineering projects. By leveraging distinct context windows and inter-agent communication, it allows developers to manage high-volume operations, perform parallel research, and resolve complex architectural challenges without cluttering the main conversation context.

Key Features

01Automated delegation logic based on task descriptions and built-in agents

020 GitHub stars

03Context management through independent windows and result summarization

04Specialized subagent creation with custom system prompts and tool restrictions

05Multi-agent team orchestration with lead/teammate roles and shared task lists

06Support for parallel research, adversarial debugging, and cross-layer development

Use Cases

01Conducting adversarial code reviews where multiple agents challenge each other's security and performance hypotheses

02Parallelizing deep codebase research by spawning multiple Haiku-powered subagents simultaneously

03Isolating high-volume terminal operations or log analysis into background subagents to preserve main session context

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add aznatkoiny/claude-dev-toolkit subagents-and-teams

For use in Claude.ai and ChatGPT

Download Skill