Module 0: Your First AI Agent in 30 Minutes

Install Claude Code

~5 minutes

What is Claude Code?

Claude Code is Anthropic's official command-line tool for building AI agents. It's the same engine that powers some of the most advanced AI systems in production today.

Think of it as your AI co-pilot that can read files, write code, search the web, execute commands, and work with you in real-time.

Installation

Open your terminal and run:

npm install -g @anthropic-ai/claude-code

Don't have Node.js? Install it first:

• Windows: Download from nodejs.org (LTS version)
• Mac: brew install node or download from nodejs.org
• Linux: sudo apt install nodejs npm

Verify Installation

claude --version

You should see something like claude-code v1.x.x. If you see a version number, you're good.

Troubleshooting

"command not found" — Make sure Node.js is installed (node --version). Try closing and reopening your terminal.

Permission errors on Mac/Linux:

sudo npm install -g @anthropic-ai/claude-code

Get Your API Key

~3 minutes

Claude Code connects to Anthropic's AI models via the API. Your API key is how Anthropic knows it's you.

Step 1: Create an Anthropic Account

Go to console.anthropic.com and sign up.

Step 2: Get Your API Key

1. Go to API Keys in the dashboard
2. Click Create Key
3. Give it a name (e.g., "AgentForge")
4. Copy the key — it starts with sk-ant-

Important: Save this key somewhere safe. You won't be able to see it again.

Step 3: Set the API Key

Mac/Linux:

export ANTHROPIC_API_KEY="sk-ant-your-key-here"

Windows (PowerShell):

$env:ANTHROPIC_API_KEY="sk-ant-your-key-here"

Step 4: Verify

claude "Hello, are you working?"

If Claude responds, you're connected.

About Costs

• Anthropic offers a free tier with limited usage
• For the course exercises, typical cost is $5-$20/month
• You control spending via usage limits in the dashboard

Build Your First Agent

~15 minutes

What We're Building

A Personal Task Agent — a team of specialized sub-agents:

1. Summarizer — condenses any text into key points
2. Email Drafter — turns bullet points into professional emails
3. Task Organizer — prioritizes and structures your to-do list

This isn't a chatbot. It's a team of agents with defined roles.

Step 1: Create Your Project

mkdir my-first-agents
cd my-first-agents

Step 2: Create the Project Memory

Create a file called CLAUDE.md:

cat > CLAUDE.md << 'EOF'
# My First Agent Project

## Agents
- Summarizer: Condenses text into clear bullet points
- Email Drafter: Creates professional emails from brief notes
- Task Organizer: Prioritizes and structures tasks

## Conventions
- Keep responses concise
- Use bullet points for lists
- Be professional but friendly in emails
EOF

This file is your agent's long-term memory. Every time Claude Code opens this project, it reads this file.

Step 3: Create Your Agents

mkdir -p .claude/agents

Agent 1: The Summarizer — Create .claude/agents/summarizer.md:

---
name: summarizer
description: Summarizes any text into clear, concise bullet points.
tools: Read, Glob, WebFetch
model: sonnet
---

You are a summarization specialist. When given text or a file:

1. Read the full content
2. Identify the 3-5 most important points
3. Summarize each in one clear sentence
4. End with a "Bottom Line" one-sentence summary

Rules:
- Maximum 5 bullet points
- No jargon unless the original uses it
- Capture the "so what?" — why does this matter?

Agent 2: The Email Drafter — Create .claude/agents/email-drafter.md:

---
name: email-drafter
description: Turns bullet points into professional emails.
tools: Read, Glob
model: sonnet
---

You are a professional email writer. When given notes:

1. Determine the purpose (request, update, follow-up)
2. Write a clear subject line
3. Draft the email with proper greeting, body, sign-off
4. Keep it concise

Tone: Professional but warm. Not robotic.

Agent 3: The Task Organizer — Create .claude/agents/task-organizer.md:

---
name: task-organizer
description: Organizes tasks by priority and category.
tools: Read, Write, Glob
model: sonnet
---

You are a productivity expert. When given tasks:

1. Categorize each task (urgent/important matrix)
2. Assign priority (P0: do now, P1: today, P2: this week, P3: eventually)
3. Group by category
4. Suggest an order of execution
5. Flag tasks that could be automated

Step 4: Test Your Agents

Start Claude Code:

claude

Test the Summarizer:

@summarizer Summarize this:

"Artificial intelligence agents represent a paradigm shift in how
we interact with software. Unlike traditional chatbots, agents can
plan multi-step tasks, use tools, maintain memory, and take
autonomous action. The market is projected to reach $11.78 billion
by end of 2026."

Test the Email Drafter:

@email-drafter Write an email from these notes:
- To: potential client, Sarah from TechCorp
- About: proposing an AI agent for customer support
- Key points: reduce response time by 80%, handles 70% of queries
- Tone: professional but enthusiastic

Test the Task Organizer:

@task-organizer Organize these tasks:
- Fix the bug in the login page
- Write blog post about AI agents
- Prepare slides for Friday's meeting
- Respond to client email about pricing
- Review pull request from teammate

What Just Happened?

You built a team of 3 specialized AI agents. Each one has a defined role, specific tools, and clear instructions. This is the same architecture used to build agents that businesses pay $2,000-$5,000 for.

See It In Action

~5 minutes

Demo 1: Analyze a Real Codebase

Navigate to any code project you have:

cd /path/to/any/project
claude

Then ask:

Explain the architecture of this project. What does it do,
how is it structured, and what are the main components?

Claude Code will read your files, understand the structure, and give you a clear breakdown. This is what businesses pay for — instant codebase understanding.

Demo 2: Summarize a Webpage

@summarizer Fetch and summarize the key points from:
https://docs.anthropic.com/en/docs/about-claude/models

Your agent browsed the web, read a page, and summarized it. 10 seconds vs 10 minutes manually.

Demo 3: Agent Team Chaining

@summarizer Read the file tasks.txt and summarize what I need to focus on

@task-organizer Now take that summary and create a prioritized action plan

@email-drafter Draft an email to my team with this week's priorities

See how they chain? That's the power of agent teams.

What This Means

You just experienced what most AI courses take weeks to get to:

• Agent teams — multiple specialists working together
• Tool usage — reading files, browsing the web, writing output
• Memory — CLAUDE.md gives persistent context
• Real utility — not a demo, something genuinely useful

What's Next

~2 minutes

What You Accomplished

In 30 minutes, you:

• Installed Claude Code
• Connected to the Anthropic API
• Created project memory (CLAUDE.md)
• Built 3 specialized AI agents
• Tested them on real tasks
• Saw agent teams work together

You now know more about building AI agents than 95% of people taking "AI courses" on YouTube.

Module 0 vs Full Program

Module 0 (Free)	Full Program
3 simple agents	10+ production-grade agents
Claude Code CLI	Agent SDK (Python/TypeScript)
Local agents only	Deploy to production (24/7)
Basic tools	MCP Protocol (databases, APIs, browsers)
Personal use	Sellable to businesses ($2K-$5K/project)
Solo practice	Community + support