On this page
What is Claude Code?
What you'll learn
- Understand what Claude Code is and how it differs from other AI tools
- Know when to reach for Claude Code vs Copilot, Cursor, or other assistants
- Grasp the concept of an agent loop and why it matters
Claude Code is a command-line tool from Anthropic that lets Claude read, write, and run code on your machine. If you've used Copilot or Cursor, this is different — it's not autocomplete and it's not a chat sidebar. It's an agent.
What that actually means
When you run Claude Code, you're starting a session where Claude can:
- Read files in your project
- Edit files (with your permission by default)
- Run shell commands
- Use tools like
grep,git, or whatever you've installed - Keep going, step by step, until the task is done
This is called an agent loop. You describe what you want. Claude thinks, picks a tool, runs it, reads the result, and decides what to do next. It can do this dozens of times in a single task — refactoring a codebase, debugging an issue, building a feature — without you intervening.
How it differs from Copilot and Cursor
Copilot completes the line you're writing. Cursor gives you an AI-enhanced editor with a chat panel and inline edits. Claude Code gives you an autonomous terminal-native agent that works across your whole project.
They're not replacements for each other. You'll probably use all three:
- Copilot for autocomplete while you type
- Cursor for focused inline edits on a file you have open
- Claude Code for multi-file tasks, refactors, and things where you'd normally sit down for an hour
Who it's for
Claude Code is built for developers who already know how to code. It's not a beginner tool — it assumes you understand your project, your stack, and what you're trying to do. If you're comfortable in a terminal, you'll feel at home.
What you need
- macOS, Linux, or Windows (Windows via WSL recommended)
- Node.js 18+ installed
- An Anthropic API key or a Claude Pro / Max subscription
- A terminal you like using
That's it. No IDE required. No extensions to install. Just a CLI.
A real example
Here's what a typical Claude Code session looks like in practice:
$ cd my-project
$ claude
> Add a dark mode toggle to the header. Use the existing theme system.
[Claude reads src/components/header.tsx]
[Claude reads src/lib/theme.ts]
[Claude edits src/components/header.tsx — adds toggle button]
[Claude runs npm run typecheck]
> Done. Dark mode toggle added to the header, wired to the existing theme context. Typecheck passes.
That's one command. Claude figured out the architecture, made the right changes, and verified the work. This is what agent-based coding feels like.
What's next
In the next lesson, you'll install Claude Code on your machine and run your first command. Ready?
Reflect
Think about a recent coding task that took you more than an hour. Could an agent have helped? What steps would you have described to it?