How to use this course

By the end of this lesson: you will know how each lesson is laid out, and you will have a project folder, a progress file and a working habit for using the course, so you can start the next lesson without guessing.

How every lesson is built

Every lesson has four parts, always in the same order.

• Video: the why. What we are doing, what it replaces and what goes wrong if you skip it. Watch it first, at normal speed. It is short on purpose.
• Written steps: the how. Commands and menu names are written out so you can copy them. If you want to follow along, keep the video and the steps open side by side.
• Prompts to copy: wherever the AI does the work, you get the exact text to paste into Claude Code. You can edit the prompts, but run the original first.
• Checklist: the last section of every lesson. If you can't tick every box, you aren't done. Move on only when you are.

The video gives you the reasoning, so you can make your own decisions later. The written steps give you the exact sequence, so you don't have to remember it. Don't rely on only one of them.

Follow the order, with one exception

The lessons build on each other. The server comes before the deploy. The database comes before the first feature. Skipping ahead usually costs you an evening of confusing errors.

The exception is the "Start here" chapter and the "From idea to plan" chapter. You can read those anywhere. If you already have a server and a domain, skim the setup lessons, tick their checklists and move on. If a checklist item surprises you, go back and do that lesson properly.

Make a working folder

Work on a laptop or on the server. Make one folder for your first project now. The name doesn't matter, so use whatever fits your idea.

mkdir ~/my-first-app
cd ~/my-first-app
git init

We cover Git properly in its own lesson. For now, git init is enough. It gives you the option to commit before every AI change, and that habit starts today.

Install Claude Code once

The next chapter walks through installing Claude Code, so I won't repeat the commands here. The install method changes between releases. When you get there, check the official Claude Code quickstart for the current one.

Two things from the docs are worth knowing now. You open your terminal in a project directory, type claude, and it prompts you to log in on first use. Claude Code always asks for permission before modifying files. Read those permission prompts. They are your first review step.

Keep a progress file

This is the one piece of homework for this lesson. You will do this course over weeks, not in one sitting. A plain text file in your project tells you, and the AI, where you stopped.

Open Claude Code in your project folder:

cd ~/my-first-app
claude

Prompt for Claude Code:

Create a file called PROGRESS.md in this folder. It is my course log.

Structure:
1. A "Current lesson" line at the top (leave it as "Start here: How to use this course").
2. A "Done" section with a checkbox list, one line per lesson. Chapters: Start here, Set up your workshop, From idea to plan, Build the app, Ship it, Make money, Get users, Run it.
3. A "Notes to self" section, empty for now.
4. A "Decisions" section where I will record choices such as my domain, my server provider and my app idea. Never put passwords, API keys or private IPs in this file.

Do not create any other files. Show me the file before you finish.

Read what it wrote. If it has invented extra sections or put in lessons you didn't ask for, say so and have it fix them. That is a small version of the loop we use for everything: ask, read, correct.

Then save it:

git add PROGRESS.md
git commit -m "Add course progress file"

Meet CLAUDE.md before you need it

Every project we build has a file called CLAUDE.md. It holds the rules the AI follows in that project: your stack, your conventions, and the things you never want it to do. You write your first one in the next chapter.

For now, just know it exists and where it lives. Claude Code loads CLAUDE.md from your current working directory and every directory above it. The /memory command lists your memory files across user and project scopes. The official memory docs cover the details if you want to read ahead.

How to run any prompt in this course

This routine applies to every prompt you will paste from now on:

1. Commit first. Run git status. If there are uncommitted changes, commit them. Then you can always undo what the AI does.
2. Paste the prompt. Swap in your own details where we use placeholders such as your-app.com or YOUR_SERVER_IP. Never paste a real password or key into a prompt.
3. For anything bigger than a small edit, ask for a plan first. Read it and approve it, or correct it. Only then say "implement step 1".
4. Review the diff. Run git diff and read it. You don't need to understand every line, but you should know which files changed and why.
5. Test it, then commit. If it works, commit. If it doesn't, go back to your last commit and try a smaller step.

Here is a prompt you can reuse whenever a task feels large:

Prompt for Claude Code:

I want to add: [describe the feature in two or three sentences].

Do not write any code yet. First give me a plan:
- which files you would create or change
- the steps, in order, small enough to review one at a time
- anything you are unsure about, as questions for me

Wait for my approval before implementing step 1.

If a session gets long and messy, you can leave it and come back. The claude -r command resumes a previous conversation. Update PROGRESS.md before you close the terminal, so that tomorrow's you knows where you left off.

When something doesn't match

Menus, prices and install commands change. If a screen in the video looks different from yours, check the official docs for that tool before you assume you did something wrong. Where we can't know the current state, a lesson tells you what to look for instead of giving a guess.

If an error appears, copy the exact message and the command that caused it. A good question has both. The debugging lesson in the "Build the app" chapter covers this in depth.

Common mistakes

• Watching the videos and skipping the steps. You will feel productive and have nothing running. Do the steps in the same sitting.
• Skipping the checklist. The box you skip is the one that breaks a later lesson, usually a few days later.
• Pasting a prompt without reading what comes back. The AI will confidently do the wrong thing. Review every diff.
• Changing prompts before running them as written. Run the original first, then adapt it. Then you know which change caused a problem.
• Pasting secrets into the chat. If a key ends up in a prompt, treat it as leaked and rotate it. We cover this properly in the secrets lesson.

Checklist

• I know the four parts of a lesson: video, written steps, prompts, checklist.
• I made a project folder and ran git init in it.
• I created PROGRESS.md and committed it.
• I know the routine: commit, plan first for big changes, review the diff, test, commit.
• I know CLAUDE.md exists and that I will write mine in the next chapter.
• I will never paste real passwords or API keys into a prompt.