How to Use Claude Code: A Beginner Walkthrough

TL;DR

• Claude Code is Anthropic's AI coding assistant that runs in your terminal.
• You need Node.js 18+ and an Anthropic account to get started.
• The claude command opens an interactive session where you describe what you want built.
• A CLAUDE.md file tells Claude about your project so it gives better answers.
• This walkthrough gets you from zero to your first working session in under 30 minutes.

If you have heard about Claude Code and want to know how to use Claude Code without wading through docs, you are in the right place. This is the walkthrough I wish I had when I first picked it up.

I use Claude Code every single day at Devwiz, my AI app and SaaS studio. I have built client systems with it, run it across 14+ businesses through the Njin Method, and it has become the primary tool in how I deliver AI work. So this is not theory. It is what actually works.

Let's get you set up.

---

What is Claude Code?

Claude Code is a command-line tool from Anthropic. It runs in your terminal and gives you direct access to Claude, Anthropic's AI model.

The big difference between Claude Code and other AI tools is that it can read your files, write code, run commands, and take action in your project. It is not just a chat window. It is an agent that works inside your actual codebase.

You describe what you want. Claude Code figures out how to do it and does the work.

Claude Code is an AI agent that runs in your terminal and can read, write, and edit files in your project.

---

What you need before you start

Before you install anything, check you have these three things.

1. Node.js 18 or higher

Claude Code is installed via npm, so you need Node.js. Run node --version in your terminal to check. If it returns v18 or above, you are good. If not, download Node from nodejs.org and install it first.

2. An Anthropic account

Go to console.anthropic.com and create an account. You will authenticate against this when you first run Claude Code, and your usage is billed to it.

3. A terminal

Mac: use Terminal or iTerm2. Windows: use PowerShell or Windows Terminal. Linux: any terminal works.

That is it. No Docker, no complex setup.

---

How to install Claude Code

This is the full installation process. It takes about five minutes.

Step 1. Install the Claude Code package

Open your terminal and install the package globally with npm. The package is published by Anthropic, and installing it globally means you can use the claude command from any folder.

Step 2. Verify the install worked

Run claude --version. You should see a version number. If you get a "command not found" error, close your terminal, reopen it, and try again.

Step 3. Authenticate

Run the claude command. The first time, Claude Code will ask you to log in. Follow the prompts. You can log in via the browser or connect your Anthropic account directly.

Once authenticated, you will see the Claude Code prompt. You are in.

Install Claude Code globally with npm, then run the claude command to authenticate.

---

Your first session: what to expect

Type a message and press Enter. Claude Code reads it, thinks, and responds.

Here is what makes it different from a web chat tool. Claude Code can actually do things. It will ask permission before it reads or writes files. You will see prompts like "May I read package.json?" or "May I create src/components/Button.tsx?". You type yes to proceed or no to skip a step.

Start simple. Try this as your first message:

What files are in this directory and what do they do?

Claude Code will scan your folder and give you a clear breakdown. It is a good way to test that everything is working.

---

Setting up a real project

Let's say you have an existing project, or you want to start a new one. Here is the workflow I use.

Step 1. Move into your project folder using the cd command in your terminal.

Step 2. Start Claude Code by running the claude command.

Step 3. Give Claude context about your project. Do not jump straight to asking it to build something. Give it a quick brief first, for example: "This is a React app for a client booking system. It uses TypeScript and Tailwind CSS. The main pages are in src/pages and components are in src/components." Now Claude Code knows where it is and what it is working with.

Step 4. Make a specific request. Be direct. Weak: "help me with the app." Strong: "Create a new page at src/pages/Confirmation.tsx that shows a booking confirmation message. Use the existing Button component from src/components." The more specific you are, the better the output.

Step 5. Review what it builds. Claude Code will show you the files it plans to create or edit before it does anything. Read them. If something looks off, say so before you confirm.

Give Claude Code project context first, then make specific requests. Review before confirming changes.

---

The CLAUDE.md file: your most important setup step

This is the thing most beginners skip. Do not skip it.

CLAUDE.md is a file you put at the root of your project. Claude Code reads it automatically at the start of every session. It is how you give Claude persistent instructions without repeating yourself every time.

Create a file named CLAUDE.md in your project root. Inside, write the things Claude needs to know every time: the project description, the tech stack, your key folders and what they hold, your conventions (functional components only, styles via Tailwind, PascalCase for component files), and anything it must never touch without asking.

Now every session starts with Claude already knowing your rules. You do not have to re-explain the stack or conventions each time.

For a deeper look at how to structure this file for maximum impact, read the guide on the CLAUDE.md strategy. It is worth it.

CLAUDE.md is a project instructions file that Claude Code reads automatically. Set it up once, benefit every session.

---

Common things to ask Claude Code

Here are real examples of what I ask Claude Code to do. These work well for beginners.

Understand existing code: "Explain what the function in src/lib/utils.ts called formatDate does and where it is used."

Write new features: "Add a loading spinner to the submit button in src/components/ContactForm.tsx. It should show while the form is submitting and hide once done."

Fix bugs: "The page at /dashboard is throwing a TypeError on line 42. Here is the error. Find the cause and fix it."

Write tests: "Write unit tests for the calculateTotal function in src/lib/pricing.ts. Use Vitest."

Refactor code: "The file src/pages/Index.tsx is 600 lines. Break it into smaller components. Keep the same functionality."

The pattern is always the same: be specific, give file paths, explain the expected result.

---

How Claude Code fits into a bigger workflow

Once you are comfortable with the basics, Claude Code becomes a proper workflow tool, not just a helper.

Here is how I use it on client projects at Devwiz. I set up a CLAUDE.md that covers the whole project. I write short task descriptions in plain English. Claude Code handles the code. I review, test, and ship.

For bigger projects, you can chain instructions. Finish one task, start the next in the same session. Claude Code keeps context from the conversation, so it remembers what it already built.

When a project gets complex, I bring in Model Context Protocol (MCP) to give Claude Code access to external tools and data sources. But that is for later. For now, just get comfortable with the basics.

If you want to see how Claude Code compares to other tools in the space, the Claude Code vs Cursor breakdown is a good read. And for a full picture of what Claude Code can do, the complete guide covers the whole system.

---

Tips that make a real difference

A few things I have learned from using this daily.

Keep sessions focused. Do not try to build five features in one session. Pick one. Do it well. Start a fresh session for the next thing.

Ask it to explain before it acts. If you are unsure about a change, ask: "Before you edit anything, explain what you plan to do." This saves you from surprises.

Use it to read code, not just write it. Claude Code is excellent at explaining unfamiliar code. Drop into a legacy codebase and ask it to walk you through what a function does. Huge time saver.

Do not approve things you do not understand. If Claude Code asks permission to run a command and you do not know what it does, ask first. It will explain.

Check the Claude Code tutorial for a deeper dive once you have the basics down.

---

FAQ

Do I need to know how to code to use Claude Code?
You do not need to be a developer, but some basic familiarity with terminals and file structures helps. If you can open a terminal and run a command, you can get started. Claude Code handles the code. You handle the direction.

Is Claude Code free to use?
No. You pay for Anthropic usage. The cost depends on how much you use it. For most beginners running a few sessions a week, the cost is low. Check Anthropic's pricing page for current rates.

What is the difference between Claude Code and the Claude website?
The Claude website is the web chat interface. Claude Code is the terminal tool that can take action in your files. They use the same underlying model, but Claude Code is built for doing work in real projects, not just answering questions.

Can I use Claude Code on Windows?
Yes. Use PowerShell or Windows Terminal. The installation steps are the same. Some file path conventions differ slightly on Windows, but Claude Code handles that automatically.

---

Ready to go further?

If you want to skip the learning curve entirely and get a custom AI system built for your business, that is what I do at Devwiz. We build AI tools, automations, and systems for real businesses.

Or if you want to talk through how AI could work in your specific operation, book a free AI transformation session. We will look at where AI fits, what is worth building first, and what to ignore.