How to Use Claude Code: A Beginner's Guide

Build with AI · Beginner guide

Claude Code can read your project, write code, and run commands for you. Here is how to set it up and use it safely, even if you have never opened a terminal.

By the end of this guide, you will have Claude Code installed, you will know how to talk to it, and you will know how to undo its work if something goes wrong. That last part matters most.

I build a lot of apps, and I use more than one AI tool to do it. Claude Code is one of the strongest. It is also easy to misuse if nobody shows you the safety steps. I checked every step and price below on the official Claude Code docs and pricing page on October 19, 2026.

What Claude Code is, in plain words

Claude Code is a coding helper made by Anthropic, the company behind Claude. Its docs call it "an agentic coding tool that reads your codebase, edits files, runs commands, and integrates with your development tools."

In plain words, you type what you want. Claude Code opens the right files, makes the changes, and can run commands to test them. You stay in charge. By default, it asks for your OK the first time it uses each tool, like editing files or running commands.

You can use it in four main places: your terminal, a code editor like VS Code, a desktop app, and the web. The docs call the terminal version "the full-featured CLI." This guide uses the terminal, but the ideas are the same everywhere.

What you need before you start

• A paid Claude plan or an API account. Claude Code is on Pro, Max, Team, and Enterprise. It is not on the Free plan. You can also use a Claude Console account that bills by usage.
• A computer that meets the basics. The setup page lists macOS 13.0 or newer, Windows 10 1809 or newer, or Ubuntu 20.04 or newer, with 4 GB of RAM or more.
• A project folder. This can be an app you started in another tool, or an empty folder for a tiny test.
• Git, if you can. Git saves versions of your work. On Windows, the docs recommend Git for Windows.

Source: claude.com/pricing and the Claude Code quickstart, checked October 19, 2026. Prices do not include tax.

Step 1: Install Claude Code

Open your terminal. On a Mac, search for Terminal. On Windows, open PowerShell. If you have never used a terminal, Anthropic has a terminal guide that shows how to open one and paste a command.

Then paste the one command for your system from the official quickstart:

The install shows no progress while it downloads, so give it a minute, then open a new terminal window and type claude --version. If you see a version number, you are set.

Step 2: Start your first session

Move into your project folder, then start Claude Code. The quickstart shows it like this:

Swap the path for your real folder. The first time, it asks you to log in. Follow the prompts in your browser with your Claude account. After that, it remembers you.

When it opens, you will see the version, the model, and the folder it is working in. Type /help to see the commands. Type /exit when you are done.

Step 3: Ask questions before you change anything

This is the step most beginners skip. Do not skip it. Let Claude Code look around first. The quickstart suggests starting with simple questions like these:

You do not need to point it at files. The docs say it "reads your project files as needed." If it gets the basics wrong, fix that now, before it writes any code.

For bigger jobs, use plan mode. The best practices page says to explore first, then plan, then code. Press Shift+Tab until the status bar shows plan mode. In plan mode, Claude reads files and answers questions but does not edit your source files. When the plan looks right, approve it and let it build.

Step 4: Make your first small change

Start tiny. Ask for one small, clear change. Say what done looks like. For example, ask it to add a line of text to your home page, or to fix one typo you found.

Claude Code shows you the change. If it asks before editing, pick Yes to approve. How often it asks depends on the permission mode. Press Shift+Tab to switch modes at any time.

Source: Claude Code docs, Configure permissions, checked October 19, 2026.

Step 5: Save your work with Git and checkpoints

AI tools can change many files fast. You want a one-step way back. Claude Code gives you two.

Checkpoints. Claude Code saves the state of your code before each prompt you send. To go back, press Esc twice when the prompt box is empty, or type /rewind. You can restore the code, the chat, or both.

Git. Checkpoints are for quick undo inside one session. The docs say plainly that they are not a replacement for version control. Git keeps your history for good. The nice part is that you can ask Claude Code to handle Git for you in plain English:

Commit before every new task. If the task goes badly, go back to the last commit and try again with a clearer request.

Step 6: Teach it your project with CLAUDE.md

A file called CLAUDE.md gives Claude Code notes about your project. The docs say Claude reads these files "at the start of every session." Put in things like how to run your app, your rules, and your style.

You do not have to write it from scratch. Type /init. The docs say it "analyzes your codebase and creates a file with build commands, test instructions, and project conventions it discovers." Then add anything it could not guess, like who your users are or what you never want changed.

How to ask for better results

The quickstart's tips for beginners are short: be specific, use step by step instructions, and let Claude explore first. Here is what that looks like in practice.

• Name the goal and the finish line. Not just a contact form, but a contact form that saves name and email and shows a thank you message.
• Give one task at a time. Small tasks are easier to check and easier to undo.
• Give it a way to check its work. The best practices page puts this near the top. Ask it to run the tests, or to tell you exactly how to test the change yourself.
• Paste the exact error. If something breaks, paste the full error message into the chat and ask it to fix the root cause. The best practices page shows this same move.
• Clear the chat between tasks. The docs say to run /clear between unrelated tasks to reset the context.

What it costs and how to watch your usage

On Pro or Max, Claude Code is part of your plan. Claude's pricing page says your activity across Claude on web, desktop, mobile, and Claude Code "all draws from the same pool." Limits reset on a rolling five-hour window, and paid plans add weekly limits.

Type /usage inside Claude Code to see where you stand. If you use a Console account instead, you pay by usage. Anthropic's cost docs say the average across company deployments is "around $13 per developer per active day." That is a company average, not a beginner budget.

If you want to compare Claude Code with Cursor before you pay, read my Cursor vs Claude Code guide.

If the terminal feels like too much

That is normal. Claude Code also runs as a desktop app and on the web, so you can skip the terminal. Or start with a no-code AI app builder and come back later. I compare those paths in Lovable vs Claude Code and vibe coding vs no-code.

I use both kinds of tools. My Freedom Launch Accelerator guide app runs on Next.js, Supabase, and Claude. If you want a gentle first project with Claude, try my vibe coding with Claude guide.