How to Use Claude Code: A Simple Guide for Beginners

Claude Code is an AI assistant that works inside your project. You tell it what you want in plain words. It reads your files, suggests changes, runs commands, and asks before it does anything important.

This guide covers the basics: how to install it, how to give it good instructions, and how to stay in control. We use Claude Code every day to build this site, so the advice below comes from real work.

What you need

Step-by-step

Step 1: Install Claude Code

Open your terminal and run the command for your system. These come from the ^[1].

On Mac, Linux or WSL:

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

On Windows PowerShell:

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

When it finishes, open a new terminal window and check that it worked:

claude --version

You should see a version number. If your terminal says claude is not found, close the window, open a new one, and try again.

Step 2: Start your first session

Move into your project folder and start Claude Code:

cd path/to/your/project
claude

The first time, it opens your browser so you can sign in. After that, it remembers you.

Start by asking questions, not asking for changes. This is the fastest way to see how it thinks about your project.

Explain what this project does in plain words. Then tell me:
1. The main folders and what each one is for.
2. How I run it on my computer.
3. Anything that looks unfinished, broken or risky.
Do not change any files yet.

Step 3: Save your project rules in CLAUDE.md

Every time you start a session, Claude Code reads a file called CLAUDE.md in your project folder. Use it for the things you are tired of repeating: how to run tests, which folders to leave alone, and how you like code written.

Type /init inside Claude Code and it writes a first version for you. Then edit it yourself. Keep it short. A few clear rules work better than a long manual.

Read CLAUDE.md and the project. Rewrite CLAUDE.md so it is short and practical. Include:
- How to install, run and test the project (exact commands).
- Folders or files that must never be changed.
- How we write code here, based on what you see in the existing files.
- Anything that has caused mistakes before: [ADD YOUR OWN NOTES].
Keep it under 60 lines. Show me the new version before saving it.

Step 4: Plan before you change anything

For anything bigger than a small fix, ask for a plan first. Press Shift+Tab to switch between modes. One of them is plan mode, where Claude Code reads and thinks but does not edit files. Read the plan, correct it, and only then let it start.

I want to [DESCRIBE THE CHANGE, FOR EXAMPLE: add a contact form to the about page].
Before writing any code:
1. List the files you would change and why.
2. Tell me anything you are unsure about.
3. Tell me how we will check that it works.
Wait for my answer before you start.

Step 5: Approve changes one at a time

By default, Claude Code asks before it edits a file or runs a command. Read what it wants to do before you say yes. This matters most for commands that delete files, install things, or touch anything outside your project.

The ^[2] explains the other options, such as letting it edit files without asking. Start with the default until you trust how it works on your project.

Step 6: Check the work

Claude Code can be confidently wrong. Ask it to prove the change works, then look for yourself.

Show me exactly what you changed, file by file. Then run the tests and the app, and tell me what you saw. If something could not be checked, say so plainly. Do not say it works unless you ran it.

Step 7: Save a version with Git

When you are happy, save the change with Git, so you can always go back. You can simply ask: "commit my changes with a clear message." If a change goes wrong, Git lets you undo it in seconds.

Useful commands

Common mistakes

• Asking for too much at once. "Build my whole app" gives messy results. Ask for one feature at a time.
• Vague requests. "Fix the bug" is hard. "Fix the login page: wrong passwords show a blank screen instead of an error" is easy.
• Long conversations. After a big task, use /clear and start fresh. Old details can confuse new work.
• Skipping Git. Without saved versions, a bad change is hard to undo.
• Trusting "done." Always run the app and look yourself.

Frequently asked questions

Do I need to know how to code to use Claude Code?

It helps. Claude Code can explain and build things for beginners, but you still need to judge whether the result is right. If you are new, ask it to explain each change in simple words.

Is Claude Code free?

No. You need a paid Claude plan or a Claude Console account with API credits. Check the current options on the Claude pricing page.

Can Claude Code break my project?

It can make bad changes, which is why it asks first by default. Keep your project in Git, read each change before approving it, and you can always undo.

What is the difference between Claude Code and the Claude chat app?

The chat app talks with you. Claude Code works inside your project: it reads your real files, edits them, and runs commands on your computer, with your permission.

Tools and resources

• Claude Code quickstart The official install and first-session guide. code.claude.com
• Claude Code best practices Anthropic's advice on instructions and project setup. code.claude.com
• Permission modes What Claude Code can do with and without asking. code.claude.com