The complete guide

What is Claude Code?

A plain-English guide to Anthropic's AI coding agent - what it is, how it works, what you can build with it, and how to start, even if you've never written a line of code.

What is Claude Code?

Claude Code is an AI coding agent built by Anthropic. You describe what you want in plain English, and it works directly inside your project - reading and editing your real files, running commands, checking the results, and fixing what breaks until the job is done. Think of it less like a chatbot and more like a capable developer you can hand a task to and trust to carry it out.

It runs in your terminal (and connects to your editor), so it isn't tied to any one app. That combination - natural language in, real changes to a real codebase out - is what makes it feel different from the AI tools most people have tried.

How Claude Code works

Under the hood, Claude Code is powered by Anthropic's Claude models. What makes it an agent rather than just a chatbot is that it can take actions: it reads the files in your project for context, makes edits across as many files as a task needs, runs your commands and tests, reads the output, and iterates. You stay in control - approving changes and steering - while it does the heavy lifting.

It also keeps context. A simple CLAUDE.md file lets you give it standing instructions about your project, and it can be extended with MCP servers, skills, subagents, and hooks to connect to your tools and data. If those terms are new, our glossary explains each one in plain English.

What you can build with it

Because it works on real projects in any stack, the range is wide. People use Claude Code to build websites, web and mobile apps, automations, AI agents, internal tools, data scripts, and full SaaS products. It's just as useful on an existing codebase as it is starting from scratch.

  • Marketing sites and landing pages
  • Web and mobile apps
  • Automations that save hours a week
  • Custom AI agents and assistants
  • MCP servers that connect your tools
  • Client work and internal dashboards

For a fuller list with real examples, see our use cases page.

Who Claude Code is for

The short answer: anyone who can describe what they want. Experienced developers use Claude Code to move faster and offload tedious work. But it has also opened the door for people who have never written a line of code - founders, creators, marketers, students - to build things they previously had to hire out.

We've built dedicated guides for each kind of builder - see Claude Code for beginners, for founders, and for creators.

How it differs from ChatGPT, Copilot & Cursor

It's easy to lump all AI coding tools together, but they work at different levels. ChatGPT is a general chat assistant you copy code out of. GitHub Copilot autocompletes lines as you type inside your editor. Cursor is an AI-first editor. Claude Code is an agent that takes whole tasks and completes them across your project - no copy-paste loop, no babysitting each line.

We break each one down fairly on our comparisons page - including Claude Code vs ChatGPT and Claude Code vs Cursor.

How to install and download Claude Code

Claude Code is distributed as an npm package, which means a Claude Code download is really just one terminal command once you have Node.js on your machine. You do not go to an app store or click through an installer - you run npm install -g @anthropic-ai/claude-code and it is ready. That is true on Mac, Windows (WSL), and Linux.

The Claude Code install process takes most people under ten minutes including the Node.js setup if you don't already have it. After installation you authenticate with your Anthropic account (more on that below), navigate to a project folder in your terminal, type claude, and you are in. No desktop app required, no IDE plugin required to get started.

The official installation guide lives in Anthropic's documentation. Claude Code Club's curriculum covers the same steps with extra context on which Node version to use, how to handle common install errors, and how to configure your first CLAUDE.md file so the agent understands your project from day one.

The Claude Code CLI - what it is and why it matters

CLI stands for Command-Line Interface. It means Claude Code lives in your terminal rather than a browser tab or a specific editor window. That is a deliberate design choice by Anthropic, not a limitation. Because the Claude Code CLI runs at the system level, it can read any file, execute any script, install packages, run tests, spin up local servers, and do everything a developer would do at the command line - all on your behalf, in response to plain-English instructions.

The CLI is also where the agentic loop lives. When you give Claude Code a task, it doesn't hand you back a block of code and wait. It opens files, makes changes, runs the code, reads the output, and if something breaks it tries to fix it - all inside the same session. You watch the work happen in real time and can intervene at any point.

For people who have never used a terminal, this sounds intimidating. In practice, the commands you need day to day are minimal: navigate to a folder, type claude, describe your task. Claude Code Club covers the terminal basics alongside the Claude Code workflow so you're never stuck on the setup.

Logging in - Anthropic account and API access

To use Claude Code you need an account with Anthropic - the company that built it. There are two ways to authenticate: through a Claude Pro or Max subscription on claude.ai, or by using an Anthropic API key directly. Both work; which one makes sense depends on how you plan to use it.

Claude Pro / Max: You log in with your claude.ai credentials and Claude Code draws from your subscription's included usage. This is the simpler path for most people starting out. Claude Pro is Anthropic's standard paid tier; Max is a higher-usage tier aimed at heavy users. A Claude Code login through this route means you're using the same account you might already have for the claude.ai web interface.

API key: Developers who want more control, or who are running Claude Code programmatically in scripts or CI pipelines, authenticate with an Anthropic API key and pay per token. This is more flexible and often cheaper at low-to-moderate usage volumes, but it requires setting up billing on the Anthropic platform.

Either way, the Claude Code login step is a one-time setup. Once your credentials are stored locally you don't re-authenticate on every session. Inside Claude Code Club we walk through both paths and help you choose the one that fits your workflow and budget.

Desktop app, IDE, and web - where you can use it

People sometimes search for a Claude Code desktop app expecting to download a standalone GUI application the way you might download Slack or Figma. Anthropic's primary distribution is the CLI (covered above), but there are several ways to bring Claude Code into a more visual workflow.

Terminal inside an IDE: The most common setup is running Claude Code in the integrated terminal inside VS Code or Cursor. This gives you the agent in a split pane next to your file tree and editor, so you can watch it make changes in real time. It does not require a separate download - just open your IDE's built-in terminal and run claude there.

Claude Code on the web: Anthropic has been rolling out Claude Code access through the claude.ai web interface for qualifying subscription tiers. This lets you use the agent in a browser without a local install, though the full file-system access and long agentic sessions that make it most powerful still work best in the local CLI setup.

IDE extensions and integrations: Claude Code has first-party support in some editors and the ecosystem is growing. Anthropic has published guidance on connecting Claude Code to VS Code, JetBrains IDEs, and others via the MCP protocol. For most builders, the terminal-inside-VS-Code setup is the sweet spot: full power with a familiar visual environment around it.

Claude Code pricing - is it free?

Claude Code is not free. Anthropic charges for the underlying model usage, either through a paid Claude plan or through API token consumption. There is no free tier for the agentic Claude Code product itself. Claude Code pricing from Anthropic depends on which model you use and how much you run it: lighter tasks on smaller models cost less; long multi-file sessions on Sonnet or Opus cost more.

That said, the cost is often much lower than people expect. Simple tasks might cost a few cents. A full afternoon of building might cost a dollar or two at API rates, or be covered within a Claude Pro subscription. The key to keeping costs reasonable is learning how to write efficient prompts, scope tasks well, and use CLAUDE.md to give the agent the context it needs without burning tokens on back-and-forth.

What Claude Code Club adds: Our $9/month membership is separate from Anthropic's charges - it covers the curriculum, community, templates, and support that help you get better results from every session. You pay Anthropic for Claude Code AI access; you join Claude Code Club to learn to use it well enough that the Anthropic spend pays for itself.

For a full breakdown of what's included in our membership, see the pricing page.

How to learn Claude Code

Learning Claude Code well is different from learning a programming language. You're not memorising syntax - you're developing judgment: knowing how to break down a project into tasks the agent can handle, how to write prompts that produce useful output, when to let it run autonomously and when to steer it, and how to debug when things go sideways.

The fastest way to build that judgment is by doing: shipping a real project inside a structured environment where you can see how other people approach the same problems and get feedback when you get stuck. That's the design behind Claude Code Club. The curriculum runs from first install through advanced agentic patterns, but the community is what makes it stick - 5,000+ people actively building, posting their work, and answering each other's questions every day.

We've also published free guides for specific starting points: Claude Code for beginners, for founders, and for creators. If you prefer to understand the terminology first, the glossary covers every concept from agents and MCP to CLAUDE.md in plain English.

How to get started

Getting started is simpler than most people expect. You set up access through Anthropic, install Claude Code, point it at a project (or an empty folder), and describe your first build. The hardest part isn't installation - it's knowing how to direct an agent well so you get great results instead of frustrating ones.

That's exactly what Claude Code Club teaches. For $9/month you get a structured curriculum, proven templates, weekly lessons, and a community of 5,000+ builders who help you ship your first real project in week one. See the curriculum or pricing to dig in.

Claude Code FAQ

What is Claude Code in simple terms?

Claude Code is an AI coding agent from Anthropic that you talk to in plain English. It reads and edits the real files in your project, runs commands, and works through whole tasks for you - like a tireless developer you can delegate to.

Is Claude Code free?

Claude Code itself is not free - it runs against Anthropic's API or a paid Claude plan, so you pay Anthropic for the usage. There is no free tier for the agentic Claude Code product. What Claude Code Club offers is a $9/month membership that teaches you how to get maximum value from your Anthropic spend.

How do I install Claude Code?

Claude Code is a CLI tool. You install it with npm (Node.js required), authenticate with your Anthropic account, and run it from your terminal. The full setup takes most people under ten minutes. Claude Code Club's curriculum walks you through it step by step.

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

No. Because you describe what you want in plain English, people with no coding background use Claude Code to build real websites, apps, and tools. Knowing the fundamentals helps, but it isn't a prerequisite to start.

What can you build with Claude Code?

Websites, web and mobile apps, automations, AI agents, internal tools, data scripts, MCP servers, and full SaaS products - on your own machine, in your own stack.

How is Claude Code different from ChatGPT or Copilot?

ChatGPT is a chat window you copy code out of; Copilot autocompletes lines as you type. Claude Code is an agent that works inside your actual project - editing files and running commands to complete whole tasks end to end.

Now go build something with it.

Claude Code Club is where you learn to use Claude Code for real - and ship your first project in week one. Join for $9/month.

Keep learning: glossary, use cases, and comparisons