Claude Code for Beginners: Install, Sign In and Run Your First Task

Most people meet Claude Code the way they meet any new tool: they open a terminal, feel lost, and fall back to pasting code into a chat window. That works, but it throws away the one thing that makes Claude Code different. It does not need you to paste anything.

This guide takes you from nothing installed to a finished first task. You will install it, sign in, ask questions about a real project, make one small change, and learn the handful of commands that cover most of your first week.

What Claude Code actually is

Claude Code is an assistant that runs in your terminal, in the folder of your project. When you give it a task, it works through a loop with three phases: it gathers context, takes action, and verifies the result. It repeats that loop until the task is done, and you can interrupt it at any point.

The difference from a chat is the tools. A chat can only answer with text. Claude Code can read your files, edit them, search the project, run commands such as your test suite, and use git. Each result feeds the next decision.

Say you ask it to fix the failing tests. It will typically run the tests to see what fails, read the error output, search for the files involved, read them, edit them, and run the tests again. You wrote one sentence. It did six steps.

If you are still deciding whether Claude Code is the right tool for what you are doing, read Claude Code vs Projects vs Chat first. This guide assumes you have decided.

Before you start

You need three things, and none of them is hard to get.

Claude Code is not part of the free Claude plan, so one of those accounts is required. If you have never opened a terminal, the official terminal guide shows how to open one and paste a command. It takes about two minutes.

Step 1: Install Claude Code

There are several ways to install it. The native installer is the recommended one, because it updates itself in the background.

Pick the one that matches your system, paste it into the terminal and press Enter. The installer shows no progress while it downloads, so do not assume it froze.

On Linux you can also install with apt, dnf or apk on Debian, Fedora, RHEL and Alpine. On native Windows, Git for Windows is recommended so Claude Code can use the Bash tool. Without it, Claude Code uses PowerShell instead. If you use WSL, you do not need Git for Windows.

Check that it worked

When the installer finishes, open a new terminal window. This matters: the old window does not know about the new program yet. Then run the version command.

$ claude --version2.1.283 (Claude Code)

If your shell says claude is not found or not recognized, the install folder is not on your PATH yet. Closing and reopening the terminal fixes it most of the time. If it does not, the official installation troubleshooting page matches each error message to a fix.

Step 2: Start your first session

Go to your project folder and type the name of the program.

$ cd ~/projects/my-website$ claude

The first time, Claude Code asks you to sign in. For a Claude subscription or Console account it opens your browser, you approve the sign-in there, and you are back in the terminal. Your credentials are stored, so you will not do this again.

You can sign in with any of these:

• A Claude Pro, Max, Team or Enterprise account, which is the recommended route.
• A Claude Console account. The first time you sign in, a "Claude Code" workspace is created automatically so you can track what it costs.
• Amazon Bedrock, Google Cloud or Microsoft Foundry, for people whose company provides access that way.

If you have the ANTHROPIC_API_KEY environment variable set and approve the key when asked, Claude Code skips the sign-in step.

Once you are in, you see a prompt with the version, the current model and the folder you are working in. Type /help to list the commands, /resume to continue an older conversation, or /login later if you need to switch accounts.

Step 3: Ask questions before you change anything

Your first prompts should not change anything. Questions are the safest way to see how Claude Code reads your project, and they teach you what it already knows.

Claude reads the files it needs and answers. You do not add files by hand, it finds them. Follow up with narrower questions, one at a time:

> what does this project do?Reading package.json, README.md, src/ ...This is a static marketing site built with Next.js. The pages live in src/app, shared components in src/components, and content in the content/ folder.

You can also ask Claude Code about itself. Questions like "what can Claude Code do?" or "how do I create custom skills in Claude Code?" get real answers, because it knows its own documentation.

Inside Claude Tutorial

Asking before acting is a habit worth building

Getting the assistant to explain what it sees before you ask for changes is the same discipline the app teaches for every kind of work, not only code.

Coming Soon

Download on theApp Store

GET IT ONGoogle Play

Step 4: Make your first change

Now ask for something small and easy to undo.

Claude Code finds the right file and shows you the change it wants to make. Depending on your settings, it may ask for your approval first. If it does, read what it proposes and choose Yes.

That prompt is deliberately tiny. The point of this step is not the function. It is seeing the full cycle once: you ask, Claude finds the file, shows the edit, you approve, and the file changes.

What Claude Code can do without asking is controlled by the permission mode of the session. Press Shift+Tab at any time to cycle between modes. Newer versions start interactive terminal sessions in auto mode, where a background check blocks risky actions instead of asking you about every one, and older versions or other setups ask more often. Look at the status line to see which mode you are in. The next guide, Claude Code Permission Modes and Plan Mode, Explained, covers every mode and when to use each.

Step 5: Use git by talking to it

If your project uses git, you can run it through conversation. These all work as written.

Even when you trust it, read the commit message and the diff before you accept. You are the reviewer of everything Claude Code produces.

Step 6: Fix a bug or add a feature

Real tasks go better with real detail. Compare these two ways of asking for the same fix.

The better versions name the symptom, the place and what done looks like. Claude Code can find files itself, so you do not have to list them, but it cannot read your mind about which problem you mean.

For bigger jobs, break the work into steps in a numbered list.

And when you do not know the codebase yet, let Claude explore first: ask it to analyze the database schema, or how the checkout works, before you ask it to change either. A few minutes of reading usually prevents a wrong change.

The commands you will use every day

You run some commands in your shell, before Claude Code starts, and others inside a session.

A few keys save a lot of time. Type / to see every command and skill you have. Press Tab to complete a command, ↑ to bring back earlier prompts, and Shift+Tab to cycle the permission mode. Press Esc to stop Claude immediately if it heads in the wrong direction.

When the first day goes wrong

These are the problems people hit most, and what each one means.

When none of that fits, ask Claude Code itself. Type something like "how do I fix this error?" with the message pasted in. The /doctor command also runs a setup checkup that diagnoses installation and configuration problems.

You do not have to use the terminal

The terminal is the original home of Claude Code, and everything above applies there. The same assistant is also available as a desktop app, inside VS Code and JetBrains editors, on the web at claude.ai/code, in Slack, and in automated pipelines with GitHub Actions and GitLab. The interface changes how you see and talk to it. The loop of gathering context, acting and verifying is the same everywhere.

Frequently asked questions

Is Claude Code free?

No. It needs a Claude Pro, Max, Team or Enterprise subscription, a Claude Console account with API credits, or access through a supported cloud provider. The free Claude plan does not include it.

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

You need a project to point it at, and the more you understand the project, the better you can judge its work. You do not need to write the code yourself. Beginners get the most from the question-first habit in Step 3: ask Claude to explain before you ask it to change.

Is it safe to let it edit my files?

You control what it can do without asking through the permission mode, and file edits can be rewound with Esc twice. Treat its output like a colleague's pull request: read the diff before you commit. Keep your project in git so every change is easy to review or revert.

What is the difference between Claude Code and Claude in a chat?

A chat answers with text about what you paste in. Claude Code works inside your project: it reads the files itself, edits them, runs commands and checks its own work. Use the chat to think, use Claude Code to do.

Does Claude Code remember my project between sessions?

Each session starts with a fresh context. To carry knowledge across sessions you write it in a file called CLAUDE.md, and Claude Code also keeps notes of its own about your preferences. The next-but-one guide, CLAUDE.md Explained, shows how.

Your first-day checklist

Continue reading

• Claude Code Permission Modes and Plan Mode, Explained for deciding how much Claude Code may do without asking.
• CLAUDE.md Explained: Teach Claude Code Your Project Once so you stop repeating the same instructions every session.
• Claude Code Best Practices: Habits That Get Better Results for prompts, checks and context management.
• How to Make Multi-File Changes With Claude Code for your first change that touches many files.
• Connecting Claude to Your Own Data and Tools With MCP for reaching beyond your own folder.