Claude Code Tutorial: Learn from Scratch

Welcome to your Claude Code learning project! This tutorial teaches you how Claude Code works and how to harness its power using the .claude folder. Whether you're new to AI-assisted development or looking to master Claude Code, start here!

Watch The Full Tutorial On YOUTUBE

https://youtu.be/lkc-F5I8gFE?si=MFLyX0RVkcHrM2nr

What is Claude Code?

Claude Code is an intelligent coding assistant that helps you write, review, debug, and understand code. It learns from your project's rules and works the way you want it to work.

Think of it like having an expert developer as a pair programmer—but one that understands your project's unique style and requirements.

Key Capabilities:

💬 Ask questions about your code
🐛 Debug issues with detailed explanations
✨ Generate code following your standards
📚 Learn best practices for your project
🤖 Automate tasks with custom skills and agents

The `.claude` Folder: Your Command Center

The .claude folder is where you teach Claude Code about your project. It's like a configuration hub that tells Claude:

How to code (your standards and style)
What it can do (custom skills for tasks)
How to specialize (different agents for different jobs)
What it remembers (persistent memory across sessions)

📁 Folder Structure Overview

Here's what's inside .claude and what each part does:

.claude/
├── CLAUDE.md              # 📖 Your project's "rulebook"
├── skills/                # 🛠️ Custom tools for specific tasks
│   ├── fetchAPI/          # Fetch data from APIs
│   ├── migrate/           # Move data between locations
│   └── visualize/         # Create visualizations
├── agents/                # 🤖 Specialized Claude personalities
│   ├── code_reviewer/     # Expert code reviewer
│   └── orchestrator/      # Workflow coordinator
├── agent-memory/          # 🧠 What Claude remembers
├── hooks/                 # 🪝 Custom automation triggers
├── settings.json          # ⚙️ Global configuration
├── settings.local.json    # 💻 Local machine settings
└── scheduled_tasks.lock   # 🔒 Task scheduling file

📚 Core Components Explained

1️⃣ CLAUDE.md – Teaching Claude Your Standards

This file is like an instruction manual for Claude Code. It defines:

What it covers:

🎨 Coding Style – PEP 8 for Python, naming conventions
🏷️ Type Annotations – Using type hints for clarity
📝 Documentation – How to write good docstrings
📦 Dependencies – Managing libraries and versions
📂 Project Organization – How files should be structured
📋 Best Practices – Error handling, version control, etc.

Why it matters: When you write code or ask Claude to write code, it reads this file and ensures everything follows your project's rules. It's like having a style guide that Claude automatically applies!

Example: If your CLAUDE.md says "use snake_case for functions," Claude will always suggest function names like fetch_data() instead of fetchData().

2️⃣ Skills – Giving Claude Superpowers

Skills are specialized modules that teach Claude how to do specific tasks in your project. Instead of Claude giving generic answers, it can perform targeted operations.

Skill #1: fetchAPI 📡

Purpose: Fetch data from external APIs
What it can do: Connect to APIs, handle responses, store data
Location: .claude/skills/fetchAPI/
Example use: "Fetch weather data from OpenWeatherMap"

Skill #2: migrate 🚀

Purpose: Move and transform data
What it can do: Migrate data between locations, track operations
Location: .claude/skills/migrate/
Example use: "Move the fetched data to the database"

Skill #3: visualize 📊

Purpose: Create visual representations
What it can do: Generate charts, graphs, analytics dashboards
Location: .claude/skills/visualize/
Example use: "Create a chart showing today's sales"

How Skills Work (Step-by-Step):

You ask Claude to do something
Claude checks available skills
If a relevant skill exists, Claude uses it
The skill executes the task and returns results

Learn by Example: Request: "Get weather data from an API and show me a chart"

Claude invokes fetchAPI skill → fetches weather data
Claude invokes visualize skill → creates a chart
Results saved automatically!

3️⃣ Agents – Claude's Specialized Personas

Agents are like different "versions" of Claude, each specialized for a specific job. Each agent has its own:

Custom instructions and expertise
Memory of past interactions
Specific workflow optimizations

Agent #1: code_reviewer 👀

Specialization: Reviews your code for quality, bugs, and improvements
Memory: Remembers patterns and feedback from previous reviews
Location: .claude/agents/code_reviewer/
When to use: "Review this code" or "Check for issues"

Agent #2: orchestrator 🎼

Specialization: Manages complex multi-step workflows
Memory: Tracks task dependencies and execution order
Location: .claude/agents/orchestrator/
When to use: Complex projects with many interconnected tasks

Agent Memory (The Smart Part!):

🧠 Each agent learns from interactions
📝 Memories are saved in .claude/agent-memory/
🔄 Information persists across sessions
📈 Improves quality over time as agent learns your preferences

4️⃣ Settings – Fine-Tuning Claude's Behavior

Two configuration files control how Claude Code behaves:

settings.json (Team Configuration)

Global settings for everyone working on the project
Check this into version control
Example: "Always use Python 3.10+"

settings.local.json (Personal Preferences)

Your local machine preferences
Don't share with the team
Example: Your preferred IDE theme or notification settings

🔄 How It All Works Together

Here's the flow when you use Claude Code in this project:

You ask Claude to do something
        ↓
Claude reads CLAUDE.md (your standards)
        ↓
Claude checks available skills
        ↓
Claude selects the right agent
        ↓
Claude applies your standards + uses relevant skills
        ↓
Claude saves to agent-memory for next time
        ↓
You get smarter, contextual results!

Real Example Walkthrough:

You: "Review my new Python function for bugs and style issues"

Claude Code does this:

✅ Reads CLAUDE.md → learns your PEP 8 style rules
✅ Loads code_reviewer agent → uses its expertise
✅ Reviews function against standards
✅ Saves feedback to agent memory → improves next time
✅ Returns detailed review with fixes

🎓 Tutorial: Learning Path

Beginner Level

Choose one of these to start:

Challenge 1: Understand CLAUDE.md

Open .claude/CLAUDE.md
Read the project standards
Ask Claude: "Explain the coding standards in this project"

Challenge 2: Explore Skills

Look at .claude/skills/ folder
Pick one skill (like fetchAPI)
Read its SKILL.md file
Ask Claude: "What can the fetchAPI skill do?"

Challenge 3: Meet the Agents

Check .claude/agents/
Look at code_reviewer configuration
Ask Claude: "Review this code for me"

Intermediate Level

Challenge 4: Request Automated Tasks

Ask Claude: "Fetch data from [API] and create a visualization"
Claude will use both fetchAPI and visualize skills
Understand how skills work together

Challenge 5: Agent Integration

Say: "I'm working on a complex feature—help me orchestrate the workflow"
Claude uses the orchestrator agent
Observe how specialized agents differ from standard Claude

Challenge 6: Create Your First Custom Skill

Design a new task you want to automate
Create a folder in .claude/skills/
Write a SKILL.md file with instructions
Ask Claude to use it!

Advanced Level

Challenge 7: Train Agent Memory

Have multiple conversations asking Claude the same type of question
Notice how responses improve over time
Check .claude/agent-memory/ to see what Claude learned

Challenge 8: Customize Settings

Edit .claude/settings.json and .claude/settings.local.json
Experiment with different configurations
Observe how Claude behavior changes

Challenge 9: Create a New Agent

Define a specialized agent for your workflow
Create a new folder in .claude/agents/
Configure its unique instructions
Use it for specialized tasks

💡 Quick Examples

Example 1: The Standard Developer

Scenario: You keep asking Claude to fix code, but the formatting is wrong

Solution: Claude reads CLAUDE.md and now automatically:
✓ Uses PEP 8 formatting
✓ Adds type annotations
✓ Includes docstrings
✓ Follows naming conventions

Example 2: The Automated Workflow

Scenario: "Fetch user data from API, migrate to database, show me a report"

What happens:
1. fetchAPI skill → gets the data
2. migrate skill → moves it to database
3. visualize skill → creates a report
All following your project's standards!

Example 3: The Smart Reviewer

Scenario: First code review vs. 5th code review by code_reviewer agent

First time: Generic feedback
Fifth time: Specific to YOUR patterns and preferences
Why? Agent memory learned what matters to you!

🎯 What You'll Learn

By exploring this project, you'll understand:

✅ What is Claude Code – An AI assistant for developers ✅ The .claude folder – Control center for customization
✅ CLAUDE.md – Teaching Claude your standards ✅ Skills – Automating specific tasks ✅ Agents – Creating specialized versions of Claude ✅ Memory – How Claude learns and improves ✅ Settings – Fine-tuning behavior ✅ Real workflows – Putting it all together

🚀 Next Steps

Ready to Dive In?

Start Small: Read .claude/CLAUDE.md (5 min)
Explore: Browse the .claude/skills/ folder (5 min)
Ask Claude: Open VS Code and ask Claude questions about this project
Experiment: Try one of the beginner challenges above
Build: Create your own skill or agent!

Common Questions:

Q: How do I use Claude Code? A: Open any file, press Ctrl+I (or Cmd+I on Mac), and start asking questions!

Q: Can I change CLAUDE.md? A: Yes! Edit it to reflect your actual project standards and Claude will follow them.

Q: How do I create my own skill? A: Create a folder in .claude/skills/, add a SKILL.md file with instructions, and Claude will recognize it.

Q: Does Claude really remember things? A: Yes! Check .claude/agent-memory/ to see what Claude learns across conversations.

📖 External Resources

🎓 Project Goals

This tutorial project demonstrates:

How to structure .claude folder for education
Best practices for teaching Claude Code
Real-world skill and agent examples
Practical learning path for beginners

Built with the goal: Help you master Claude Code from scratch! 🚀

Happy learning! Now open VS Code and start exploring Claude Code! 💻

Name		Name	Last commit message	Last commit date
Latest commit History 14 Commits
.claude		.claude
Airflow_Project		Airflow_Project
Data		Data
Notes		Notes
.gitignore		.gitignore
.mcp.json		.mcp.json
.python-version		.python-version
1_llm_call.ipynb		1_llm_call.ipynb
MIGRATION_REPORT.md		MIGRATION_REPORT.md
README.md		README.md
URLs.md		URLs.md
create_dataframe.py		create_dataframe.py
create_sample_dataframe.py		create_sample_dataframe.py
fetch_api_data.py		fetch_api_data.py
generate_dataframe.py		generate_dataframe.py
pyproject.toml		pyproject.toml
uv.lock		uv.lock

Folders and files

Latest commit

History

Repository files navigation

Claude Code Tutorial: Learn from Scratch

Watch The Full Tutorial On YOUTUBE

What is Claude Code?

Key Capabilities:

The .claude Folder: Your Command Center

📁 Folder Structure Overview

📚 Core Components Explained

1️⃣ CLAUDE.md – Teaching Claude Your Standards

2️⃣ Skills – Giving Claude Superpowers

Skill #1: fetchAPI 📡

Skill #2: migrate 🚀

Skill #3: visualize 📊

3️⃣ Agents – Claude's Specialized Personas

Agent #1: code_reviewer 👀

Agent #2: orchestrator 🎼

4️⃣ Settings – Fine-Tuning Claude's Behavior

🔄 How It All Works Together

🎓 Tutorial: Learning Path

Beginner Level

Intermediate Level

Advanced Level

💡 Quick Examples

Example 1: The Standard Developer

Example 2: The Automated Workflow

Example 3: The Smart Reviewer

🎯 What You'll Learn

🚀 Next Steps

Ready to Dive In?

Recommended Reading Order:

Common Questions:

📖 External Resources

🎓 Project Goals

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

The `.claude` Folder: Your Command Center

Packages