Book 1: Getting Started with Claude Code

Written 2026-10-04.

From installation to your first checked piece of work. No coding experience required.

This book is for anyone starting with Claude Code in late 2026: people who have never opened a terminal, and developers who have used chat assistants but not a coding agent. It follows the second edition of the handbook. The first edition (March 2026) taught you to approve every step an AI took. A lot has changed since then. Auto mode is now how most sessions start, the models can work for long stretches on their own, and the skill that matters most is checking the result, not reading every line. This book teaches the new basics in that order.

If you read the first edition, start with What's New. The first edition is still online under /v1/.

Table of Contents

Part I: What and Why

1. What Is Claude Code?: the agent, where it runs (terminal, desktop, web, IDE), and the tools it competes with
2. Why Claude Code?: an honest comparison with Codex, Cursor and others, including the weak spots
3. How It Works: the agent loop, the harness, the context window, models and effort levels

Part II: Setup

4. Installation: terminal install on Mac, Windows and Linux, signing in, and the desktop and web apps
5. Your First Conversation: start a session, ask a question, watch it make a file
6. Auto Mode and Permissions: what the classifier lets through, what it blocks, deny rules, and when to turn it off

Part III: Daily Workflows

7. Reading and Understanding Code: ask for explanations, summaries and diagrams instead of reading line by line
8. Editing Files: making changes and undoing them
9. Running Commands: terminal work through Claude
10. Git Workflows: a branch per task, pull requests, and a first look at worktrees

Part IV: Your First Project

11. Building a Simple Website: a hands-on project, start to finish
12. Check the Work: tests first, asking the agent to prove it works, screenshots, and debugging
13. Working with APIs: connecting to outside services

Part V: Customization Basics

14. CLAUDE.md and AGENTS.md: standing instructions, and one file that several tools can read
15. Memory: what Claude remembers between sessions and what to store
16. IDE Integration: VS Code, Cursor and JetBrains

Appendix

• Glossary
• Troubleshooting
• Keyboard Shortcuts

How to read this book

Read Parts I and II in order. They take about an hour and leave you with a working setup. After that, go to whichever chapter fits the job in front of you. Chapter 12, Check the Work, is the one to read even if you skip others: an agent that works on its own is only useful if you can tell when it got something wrong.

When you are comfortable here, Book 2: Power User covers skills, subagents, hooks and context, and Book 3: Architect covers running many agents, containing them and measuring what they cost.