CCAT Chapter 1: Introduction to Claude Code Agent Teams

  1. Home
  2. Blog
  3. CCAT Chapter 1: Introduction to Claude Code Agent Teams

CCAT Chapter 1: Introduction to Claude Code Agent Teams

Learning Objectives

By the end of this chapter, you will:

  • Understand what Claude Code is
  • Learn what Agent Teams are and why they matter
  • Know when to use Agent Teams vs single sessions
  • Understand the difference between Agent Teams and Subagents

What is Claude Code?

Claude Code is a command-line interface (CLI) tool powered by Claude AI that helps developers with software engineering tasks. Think of it as having an AI assistant right in your terminal!

Claude Code Terminal Interface

What Are Agent Teams?

Agent Teams let you run multiple Claude Code instances working together on the same project. Imagine having a team of AI assistants, each with their own specialty!

Simple Analogy: A Restaurant Kitchen

Think of Agent Teams like a restaurant kitchen:

Agent Teams as a Restaurant Kitchen Analogy

Why Use Agent Teams?

The Problem with Single Sessions

When you have a complex task, a single Claude session must do everything one step at a time:

Sequential vs Parallel Task Execution

The Agent Teams Solution

With Agent Teams, multiple agents work simultaneously:


Agent Teams vs Subagents: What’s the Difference?

This is a common point of confusion. Let’s clarify:

Subagents Vs Agent Teams

Subagents vs Agent Teams Comparison

Comparison Table

FeatureSubagentsAgent Teams
ContextShared with main agentEach has own context
CommunicationReport to main onlyTalk to each other
You can talk toMain agent onlyAny teammate
Best forSimple, focused tasksComplex collaboration
Token costLowerHigher
SetupSimplerMore setup needed

When Should You Use Agent Teams?

GREAT Use Cases for Agent Teams

  1. Research and Review

– Multiple teammates investigate different aspects simultaneously

– They can challenge each other’s findings

  1. Building New Features

– Each teammate owns a different module

– Frontend, Backend, Tests – all at once!

  1. Debugging Complex Issues

– Different teammates test different theories

– They debate and find the truth faster

  1. Cross-Layer Changes

– Frontend teammate, Backend teammate, Database teammate

– Each owns their layer

When NOT to Use Agent Teams

  1. Simple, sequential tasks – Use single session
  2. Same-file edits – Causes conflicts
  3. Tasks with many dependencies – Use subagents
  4. Quick fixes – Overkill for simple tasks

The Components of an Agent Team

Here’s the complete architecture:

Agent Team Architecture Diagram

Component Breakdown

ComponentWhat It Does
Team LeadThe boss! Creates team, assigns tasks, coordinates work
TeammatesWorkers! Each is a full Claude Code instance
Task ListShared to-do list everyone can see and update
MailboxHow agents send messages to each other

Real-World Example: Building a Web App

Let’s see how Agent Teams would help build a simple e-commerce website:

E-Commerce Site Agent Team Example

Key Terminology

TermDefinition
AgentA Claude Code instance that can work on tasks
Team LeadThe main session that creates and coordinates the team
TeammateA separate Claude Code instance spawned by the lead
SpawnTo create a new teammate
TaskA unit of work assigned to a teammate
MailboxThe messaging system for agent communication
Context WindowThe “memory” an agent has for conversation
TokenThe unit of text Claude processes (roughly 4 chars)

Chapter Summary

In this chapter, you learned:

  1. Claude Code is an AI assistant in your terminal
  2. Agent Teams let multiple Claude instances work together
  3. Team Lead coordinates, Teammates do the work
  4. Agent Teams ≠ Subagents – Teams can communicate with each other
  5. Use Agent Teams for parallel, independent work
  6. Avoid Agent Teams for simple or sequential tasks

Knowledge Check Quiz

Answer these questions to test your understanding:

  1. What is the main difference between Agent Teams and Subagents?
  2. Name three good use cases for Agent Teams.
  3. Who coordinates the work in an Agent Team?
  4. Can teammates talk to each other?
  5. Why would Agent Teams be faster than a single session for some tasks?
Click to see answers
  1. Agent Teams allow teammates to communicate with each other, while Subagents can only report back to the main agent.

2. Good use cases: Research/Review, Building new features, Debugging complex issues, Cross-layer changes.

3. The Team Lead coordinates all work.

4. Yes! Teammates can message each other directly.

5. Agent Teams work in parallel (simultaneously), so multiple tasks complete at the same time instead of one after another.


Exercise 1.1: Identify the Right Tool

For each scenario below, decide: Single Session, Subagents, or Agent Teams?

ScenarioYour Answer
Fix a typo in a README file
Review a PR for security, performance, and tests
Research 5 different API options and compare
Add a console.log for debugging
Build a feature with frontend, backend, and tests
Sequential database migration
Click to see answers | Scenario | Answer | Why | |———-|——–|—–| | Fix a typo in a README file | Single Session | Simple, quick task | | Review a PR for security, performance, and tests | Agent Teams | Parallel review with different focuses | | Research 5 different API options and compare | Agent Teams | Each can research independently | | Add a console.log for debugging | Single Session | Very simple task | | Build a feature with frontend, backend, and tests | Agent Teams | Each layer can work in parallel | | Sequential database migration | Single Session or Subagents | Must be done in order |

Next Chapter

In Chapter 2: Setting Up Your Environment, you’ll learn how to install and configure Claude Code for Agent Teams.


Let's Share

Write a Reply or Comment

Show Buttons
Hide Buttons