Claude Code Handbook

A companion for students learning to build real software with an AI coding agent.

This is not a transcript of a video series and it is not a replacement for one. It is the book you keep open next to the videos: it explains each idea a second time in different words, shows a different worked example you can run, warns you about the mistakes students actually make, and fills in topics the videos do not have time for (permissions, hooks, security, troubleshooting).

Who this is for

You can write some JavaScript and have used a terminal. You have heard that AI agents can write code, and you want to use one responsibly — to move faster and still understand what ships.

How the book follows the video series

The nine parts below follow the same order as the nine sections of the Claude Code video series, so you can watch a lesson, then read the matching chapter.

Part	Topic	Chapters
1	Orientation	Start here
2	Claude Code fundamentals	Coding agents · Install · Memory · Prompting & plan mode · Context & cost · MCP & skills · Safety nets · Permissions & hooks · Future
3	Plan a real project	Idea to scope · Stack & plan · Project setup
4	Authentication, security, testing	Authentication · Forms, validation, roles · Audits & E2E
5–6	Building features	CRUD · Lists, details, safe output
7	AI features	AI features, jobs, metrics
8	Email	Email integration
9	Ship it	Production · Claude in GitHub & final thoughts
—	Appendices	Cheat sheet · Troubleshooting · Glossary · Prompt library · Exercises

One running example: Shelf

Every chapter uses the same small project so ideas build on each other: Shelf, a reading-list REST API (add books, track to-read / reading / done, rate finished books). It was chosen because it is small enough to read in ten minutes yet real enough to need validation, tests, authentication, background jobs, an AI feature, email, and deployment.

• Zero dependencies — plain Node 20+ built-ins, so nothing to install.
• 47 automated tests — npm test is green, and chapters tell you which test proves which idea.
• Claude configuration included — a real CLAUDE.md, a skill, a subagent, hooks, permissions and an MCP server.

Download Shelf (zip)

unzip shelf.zip && cd shelf
npm test      # 47 passing tests
npm start     # API on http://localhost:3000

How to use each chapter

Every chapter has the same shape, so you always know where to look:

1. Video map — which lessons it pairs with.
2. The idea in plain words — the concept without jargon.
3. Worked example — real, tested Shelf code or configuration.
4. Common mistakes — what trips students up.
5. Check yourself — a short checklist before you move on.

Your output will differ — that is normal

Language models are non-deterministic. Ask the same question twice and you may get two different (both reasonable) answers. Do not chase identical output. Learn the workflow: break work into small steps, read every diff, and recover from mistakes.

Conventions

• code font is something you type or a file name.
• Prompts you can give Claude are shown in a quoted block, like this:

Add a GET /stats endpoint that returns counts by status. Follow the pattern in src/app.js. Add a test first.

About versions

Claude Code changes quickly. Menus, model names and commands evolve. Where a detail is likely to drift, the chapter says so and points you to the official documentation. The ideas — agent loop, context, memory, permissions, review — stay stable.

// share