Claude Code tutorial for beginners, step by step

Updated on 7 October 2026

This tutorial takes you from zero to your first project built with Claude Code: you install it, sign in, start your first conversation, learn how to ask for things and build a simple web page. You don't need to know how to code; you need a computer, a paid Claude plan and about an hour.

It brings together, in one place, the ideas from the first two modules of the "Claude Code for Beginners" course. The commands come from the official Claude Code documentation, checked on 7 October 2026.

What is Claude Code, in one sentence?

Claude Code is an AI agent that works on your computer: it reads your project files, changes them, runs commands and checks the result, only with your permission.

The difference from a chat: you don't just get an answer that you copy yourself. You get a helper that does the work in your folder. Think of a skilled handyman you call to your home: you say what you want, he looks around, makes a plan and gets to work, but he asks before drilling into the wall. More in the guide what is Claude Code.

What do you need before you start?

  • A computer with macOS 13 or later, Windows 10 (version 1809) or later, or Linux.
  • A paid Claude plan: Pro, Max, Team or Enterprise. The cheapest, Pro, costs $20 a month plus tax, or $17 a month billed yearly (claude.com/pricing, checked on 7 October 2026). The free plan does not include Claude Code. Details in how much does Claude Code cost.
  • An internet connection and about an hour for the whole tutorial.

You don't need any other software installed first.

Step 1. How do you open the terminal?

The terminal is a window where you type commands as text instead of clicking buttons. You can't break anything just by opening it.

  • On a Mac: press Cmd + Space, type "Terminal" and press Enter.
  • On Windows: press the Windows key, type "PowerShell" and click "Windows PowerShell".

You'll see a short line ending in %, $ or > and a blinking cursor. That's the prompt line: where the terminal waits for you to type. A command is one line of text plus Enter. When the prompt line comes back, it means "done, waiting for the next one".

Three tips to start with:

  • copy commands exactly; one missing space changes everything;
  • if lots of text scrolls past, that's normal: wait;
  • lines that start with # in the examples are just comments.

More in the lesson what is the terminal and how do you open it.

Step 2. How do you install Claude Code?

You type one command, the one for your computer, and press Enter.

On a Mac or Linux:

curl -fsSL https://claude.ai/install.sh | bash

On Windows (PowerShell):

irm https://claude.ai/install.ps1 | iex

The command downloads the installer from the official site and runs it. When it finishes, close the terminal, open a new one and check:

claude --version
claude doctor

If you see a version number, it's installed. claude doctor runs a full check and tells you if anything is wrong. If you get "command not found", close the terminal and open a new one. Common errors and their fixes are in the guide how to install Claude Code on Mac and Windows.

Step 3. How do you sign in to your account?

Type claude and press Enter. The first time, your browser opens and you log in with your claude.ai account. After that you stay signed in; you don't log in every time you start it.

To switch accounts, type /login inside Claude. Besides a subscription, you can also sign in with a Claude Console account, where you pay for what you use. To start with, a subscription is simpler.

Step 4. How do you start your first conversation?

A project is just a folder with files. Make a new folder and start Claude in it:

mkdir first-project
cd first-project
claude

mkdir makes the folder, cd moves into it, and claude starts Claude Code there. You'll see a welcome screen and a place to type. In the examples below, the > sign shows where you type; don't copy it.

Your first request can be very simple:

> Hi! Tell me briefly what you can do for me in this folder.

Claude answers in plain language. In a folder with files, it reads them on its own and explains what they are. It doesn't change anything without asking you.

Step 5. Which keys and commands do you need from day one?

Only a few. These are enough for your first weeks:

Key or command What it does
Enter Sends your message
Esc Stops Claude in the middle of what it's doing
Shift+Tab Switches the working mode (for example plan mode, where it only proposes)
@file Points Claude at a specific file
/clear Clears the conversation so you start fresh
/help Lists the commands
/init Creates the CLAUDE.md file, the project's memory

Commands that start with / are called slash commands, and you type them inside Claude, not in the regular terminal. The full list, explained, is in the lesson what are the main slash commands.

A word you'll hear a lot: context. It's everything Claude "keeps in mind" in the current conversation: the messages, the files it has read, the output of commands. Think of a workbench: when it's too cluttered, the work gets worse. That's why you type /clear when you start a new task. See also context in the glossary.

Step 6. What does a good message for Claude look like?

The message you send is called a prompt. Claude is very capable, but it sees your project for the first time and doesn't remember yesterday's conversations. A good prompt has four parts:

[What to do] + [Where] + [What not to touch] + [How to check]

Here's the difference, using a cafe website as the example:

Bad (Claude has to guess almost everything):

> fix the button

Good:

> On a phone, the "Book" button does nothing when you tap it.
  It should open WhatsApp. Look in index.html, in the booking section.
  Don't change the colors or the text. Check: tell me what link the button
  has now and what link it has after the fix.

The second one says what happens, where to look, what to leave alone and how Claude will show you it worked. More in the lesson what does a good message look like.

Step 7. Which tricks help you work better?

  • Ask it to ask you. When you don't know what to say, write: "before you start, ask me the questions you need, one at a time". That way it learns what it needs from you instead of making things up.
  • Point at files with @. For example: "make @menu.html match the style of @index.html".
  • Give examples and pictures. You can drag a screenshot into the terminal; Claude reads it.
  • Ask for the plan first. "Stop, show me the plan first" beats fixing ten files afterwards.
  • One conversation, one task. New task, /clear.
  • Stop it early. If it heads the wrong way, press Esc and rephrase.

All of these, with examples you can try in a practice terminal, are in the lesson what tricks will help you.

Step 8. How do you build your first project: a page for a cafe?

Now you put it all together. In the first-project folder, with Claude running, type:

> I want to build a one-page website for a cafe.
  Before you start, ask me the questions you need, one at a time.
  Don't write any files until I say "start".

Answer the questions briefly (name, opening hours, menu, phone). Where you don't know, write "you choose". Then:

> Start. A single file, index.html, no libraries.
  It must look good on a phone. The booking button opens WhatsApp.

Claude shows you the file and asks for permission. You approve. Then ask "open index.html in my browser", look at it like a customer would and ask for fixes one at a time. At the end, ask it to check its own work: "is the WhatsApp link in the right format? Is it easy to read on a phone?".

That's your first project. To put it online and get a link you can send to anyone, follow the steps in the guide how to build a website with AI or in lab 0 of the course.

What mistakes do most beginners make?

The mistake What to do instead
"Make it all good" Say clearly when it's done and how to check
One three-hour conversation for five tasks /clear between tasks
Correcting the same mistake for the fifth time /clear and a new, better message with what you've learned
Accepting everything without looking Check especially what customers will see: prices, hours, phone
Working straight in your important documents Start in a test folder

What comes after this tutorial?

After the first two modules, the natural next step is CLAUDE.md, the file where you write the project's rules so you don't repeat them. Then permissions (what Claude may do on its own), a step-by-step working method and more advanced tools. They're all in the full course, and you'll find new words in the glossary.

Frequently asked questions

How long does it take to learn Claude Code?

The first steps, from installing to your first page, take about an hour. To feel confident, practise for a few evenings on small projects.

Can I learn Claude Code without a subscription?

You can read this tutorial and the free lessons. To actually work with it, you need a paid Claude plan or a Claude Console account with credit.

Do I need to know how to code?

No. You need to describe clearly what you want and check what it did. The course starts from zero.

Does it work on a phone?

You can take the course lessons on your phone. You use Claude Code on a phone through the Claude app or on the web, at claude.ai/code; the install in this tutorial is for a computer.

Take the tutorial as a course, with exercises

The "Claude Code for Beginners" course covers the same ideas in 1–2 minute lessons, with quizzes, review cards and a practice terminal where you try commands with no risk. The first two modules and lab 0 are free, no card needed.

Start the first lesson, free