How to Install and Set Up Claude Code (Beginner's Guide)

A lot of people get stuck before they even start, because every tutorial makes the setup more complicated than it needs to be. It doesn't have to be. This guide shows you how to install Claude Code and set it up from scratch, the beginner-friendly way, then build your first project, even if you've never written a line of code.

Claude Code is Anthropic's agentic coding tool: an AI that works on real files, runs commands, and builds working software for you. The setup takes about ten minutes, and once it's done you'll have a capable AI working on your machine.

Watch: this guide follows our video How to Set Up Claude Code as a Beginner, so you can read along or watch it built live.

Before you install: what you'll need

A few things need to be in place first. Get these sorted and the rest is quick.

You'll need a paid Claude plan. Claude Code isn't on the free tier; it starts with the Pro plan at $20 a month, or $17 billed annually, with Max plans for heavier use. Team and Enterprise plans include it too.

You'll also want three free tools installed. First, Visual Studio Code, a free code editor from Microsoft; this is where you'll run Claude Code. On a Mac, drag the app into your Applications folder after it downloads, as it doesn't always move there on its own. Second, Node.js. Third, Python.

You don't need to know any JavaScript or Python to use Claude Code. The reason to install Node and Python first is that Claude Code will often generate code in one of those languages, and having them installed means it can run and test what it writes. Install all three the same way you would any other app, following the prompts.

How to install Claude Code (the official extension)

There are three ways to install Claude Code, which is part of why beginners get confused. The terminal version is the most powerful but the most technical. The desktop app is the easiest to start with but the most limited. The VS Code extension sits in the middle, with most of the power and far less friction, which is why it's the one to start with.

Open VS Code and click the Extensions icon in the left sidebar (the four squares). Search for "Claude Code" and look for the official extension from Anthropic, marked with a blue verified tick and millions of installs. Be careful here: there are imposters, so check the publisher says Anthropic before installing. Click install and confirm you trust the publisher.

Once it's installed, a Claude icon appears in the sidebar. Click it to open the Claude Code panel. One thing to know: VS Code has its own built-in AI chat, which is not Claude Code, so make sure you're in the Claude Code panel specifically. On Windows, you may be prompted to install Git Bash during setup; that's expected, so trust the link and install it.

A quick warning that trips people up: don't open the terminal and type claude expecting it to work. The terminal version is a separate install from the extension.

Log in and connect your plan

With the panel open, you need to connect your account. You'll see a few login options; the one to use is your Claude subscription. Click it, approve the connection in the browser window that opens, choose your account, and authorise access. You'll see a "you're all set" message, and you can close the browser.

There's also an option to log in with an Anthropic API key instead, which bills usage as you go from $5. For beginners, the subscription is the better choice: it's a fixed monthly cost, so you're not watching a meter, and agentic coding gets through tokens quickly.

Set up your first project the right way

This is the step most people skip, and it causes most of the early frustration. Before you prompt Claude Code, always open a dedicated folder to work in.

Create an empty folder somewhere sensible, for example one called "Claude Code projects," then in VS Code go to File, then Open Folder, and select it. If VS Code asks whether you trust the authors of the files, say yes (you created it). One quirk to watch: opening a folder sometimes flips VS Code back to its default chat, so click back into the Claude Code panel before you start.

Working inside a folder is what gives Claude Code somewhere to create and edit files safely. Skip it and you'll quickly make a mess; do it and every project stays self-contained.

Your first build: plan mode and clarifying questions

Now the good part. The reliable way to build is to describe the outcome you want and let Claude Code plan it before it writes anything.

Before you send your first prompt, switch from the default "ask before edits" to plan mode at the bottom of the panel (or press Shift and Tab). In plan mode, Claude Code asks clarifying questions and lays out a plan instead of immediately changing files. Write a clear prompt that lists what you want; for a web app, name the features one by one rather than saying "build me an app." It helps to add that it should use its clarifying-questions tool to check any details.

Claude Code will then ask you a short set of questions, often with a recommended answer for each, which is helpful when you're not technical. Once you've answered, it produces a plan: the approach, the files it will create, and how it will test the result. Read that plan rather than rushing past it; if something's wrong, you can tell it what to change before a single file is written.

When you're happy, accept the plan and Claude Code gets to work, ticking off its own to-do list as it creates files and writes code. It will ask permission for certain actions, with a plain-English explanation of each; for routine ones you can choose to allow them for the whole project so you're not clicking approve constantly. When it's done, test the result. If something breaks, the easiest fix is to screenshot the problem, paste it into the chat, and describe what went wrong; because Claude Code checks and corrects its own work, it will usually read the error, fix it, and try again.

If you'd rather build a practical tool than a toy app, our walkthrough of building a personal AI agent with Claude Code is a good next step.

Make it reusable and stay inside your limits

Two habits will save you a lot of time and usage once you're up and running.

The first is skills. When you've built something you'll do again, save it as a skill using Anthropic's Skill Creator, so next time it runs from a single command instead of a fresh back-and-forth. As you get more comfortable, you can even connect Claude Code to your own tools and APIs. You can also create a claude.md file by running /init; Claude Code reads it at the start of every session in that project, which is how it "remembers" your setup and rules between conversations.

The second is managing context. Claude Code works within a context window, and every message re-sends the whole conversation, so a long thread gets slower, pricier, and less accurate. Type /context to see how full yours is, and start a fresh conversation with /new when you switch to a new task. This matters because usage runs on a rolling five-hour limit plus a weekly cap; long, bloated conversations burn through tokens faster. You can also switch models with /model, with Opus the most capable choice.

If you're not a developer and this is starting to feel like more setup than you want, that's worth noticing. Claude Cowork does the same kind of autonomous, multi-step work for business tasks with no terminal and no install beyond the desktop app, and for most operators it's the better starting point.

Knowing where you sit on the AI skills ladder helps you choose the right tool to learn first.

Start building with Claude Code

That's the whole setup: a paid plan, VS Code with Node and Python, the official extension, your account connected, and a dedicated folder to work in. From here, the way to learn is to build, test, and save what works as a skill.

If you'd rather put AI to work on real business tasks without living in a code editor, our hands-on AI automation training takes you from your first task to automations that run across the tools you already use.