What Is Claude Code? A Beginner's Guide

Suraj Kumar

Suraj Kumar

Last updated November 4, 2026·8 min read

Claude Code course · lesson 1 of 30
  1. 1What Is Claude Code? A Beginner's Guide
  2. 2How to Install Claude Code and Sign In
  3. 3Your First Claude Code Session, Step by Step
  4. 4Claude Code Terminal Interface and Keyboard Shortcuts
  5. 5Claude Code Models, Effort Levels and Fast Mode Explained
  6. 6How Claude Code Works: The Agent Loop, Tools and Context
  7. 7Claude Code Plan Mode: Think Before You Build
  8. 8Claude Code Permission Modes and Permission Rules Explained
  9. 9Managing Context in Claude Code: /clear, /compact, /rewind and /resume
  10. 10Claude Code Prompting Patterns: How to Write Prompts That Work
  11. 11Debugging, Refactoring and Writing Tests with Claude Code
  12. 12CLAUDE.md Explained: Give Claude Code Lasting Instructions
  13. 13Claude Code Rules Files and Auto Memory Explained
  14. 14Claude Code settings.json Explained: Files, Scopes and Precedence
  15. 15Claude Code Output Styles and the Status Line
  16. 16Claude Code Skills and Custom Slash Commands: Build Your Own
  17. 17Claude Code Subagents Explained: Delegate Work Safely
  18. 18Claude Code Hooks Explained: Automate and Enforce Rules
  19. 19Claude Code MCP Servers: Connect Claude to Your Tools
  20. 20Claude Code Plugins and Marketplaces: Install, Manage and Stay Safe
  21. 21Claude Code Git Workflows and Worktrees for Parallel Sessions
  22. 22Claude Code Headless Mode (claude -p) and the Agent SDK
  23. 23Claude Code GitHub Actions: Run Claude in Your CI Pipeline
  24. 24Code and Security Review with Claude Code: /code-review and Safe Habits
  25. 25Claude Code Scheduled Tasks and /loop: Run Prompts on a Timer
  26. 26Claude Code Costs and Token Efficiency: Track and Reduce Usage
  27. 27Claude Code Sandboxing, Secrets and Dev Containers Explained
  28. 28Rolling Out Claude Code to a Team: Managed Settings and Policy
  29. 29Troubleshooting Claude Code: Fix Common Problems with /doctor and Safe Mode
  30. 30Claude Code in Real Projects: Build a Feature, Fix a Bug, Run a Migration
View full syllabus

By the end of this lesson you will know what Claude Code is, what it can do for you, where you can run it, and whether it fits the way you work. You do not need to install anything yet. The Installing and signing in lesson covers installation.

What Claude Code is

Claude Code is an AI coding assistant made by Anthropic. You describe a job in plain English, and it does the work inside your project: it reads your files, changes code, runs commands, and checks whether the result works.

The word to notice is agentic. An agentic tool does not just answer a question and stop. It takes steps toward a goal, looks at what happened, and decides what to do next. If you ask Claude Code to fix a failing test, it can open the test, find the code it covers, change that code, run the test again, and report back. You watch each step and approve the ones that matter.

Diagram of the Claude Code loop: you describe a goal, then it reads files, plans, edits code, runs tests and checks the result, going around again if a test fails, and finally reports back to you

A few terms used throughout this course:

  • Codebase is all the files that make up your project.
  • Terminal is the text window where you type commands instead of clicking buttons. On a Mac it is the Terminal app, on Windows it is PowerShell or CMD.
  • Command is one instruction you type into the terminal, such as ls to list files.
  • Prompt is the message you write to the AI.
  • Session is one continuous conversation with Claude Code.

A short example

Say you have a small website and a button that does nothing when clicked. You open your terminal in the project folder, start Claude Code, and type:

text
1The "Subscribe" button on the homepage does nothing when I click it. Find out why and fix it.

Here is the kind of thing that happens next. This transcript is simplified to show the idea:

text
1> I'll start by finding where the Subscribe button is defined.
2  Searched for "Subscribe" across 14 files
3  Read components/SubscribeButton.tsx
4
5> The button has no onClick handler, so clicking it does nothing.
6  I'd like to edit components/SubscribeButton.tsx. Allow? (yes / no)

You read the proposed change and answer. If you say yes, Claude Code makes the edit and can run your project's tests to confirm nothing else broke. You stay in control the whole time, because Claude Code asks before it changes files or runs commands that need your approval.

What you can use it for

People use Claude Code for tasks like these:

  • Building a new feature from a plain-language description
  • Finding the cause of a bug from an error message
  • Writing tests for code that has none
  • Explaining a codebase you have just joined ("how does login work here?")
  • Renaming something across dozens of files
  • Writing commit messages and opening pull requests (a pull request is a proposal to merge your changes into the main project)
  • Updating dependencies, fixing lint errors, and other chores

It works best when you give it a clear goal and let it look around, instead of pasting snippets and hoping for the best.

How it differs from other AI tools

You may already use a chat tool or an editor with AI suggestions. Claude Code overlaps with both, but the working style is different.

A chat tool such as the Claude or ChatGPT website only sees what you paste in. You copy code over, read the answer, and copy it back into your editor yourself. It cannot run your tests, so it cannot know whether its suggestion works.

An editor assistant, like the autocomplete in many code editors, suggests the next few lines while you type. That helps with small, local edits. It generally does not plan a change across ten files or run your build to check the outcome.

Claude Code sits closer to a teammate who has access to your project. It can read the files it needs, make coordinated changes in several places, run commands, and look at the results before it tells you it is done.

Comparison chart. Chat tools see your project and handle multi-step tasks to a limited degree and cannot edit files or run commands. Editor autocomplete sees your project and edits files to a limited degree and cannot run commands or handle multi-step tasks. Claude Code does all four

Many developers use more than one of these. They are not exclusive.

Where you can run it

Claude Code runs on several surfaces, which are the different places you can open it. They share the same engine, so your project instructions and settings follow you between them.

  • The terminal, which is the full-featured command-line version
  • VS Code and JetBrains editors, through an extension or plugin
  • The Claude desktop app, in its Code tab, which shows changes visually
  • The web at claude.ai/code, which needs no local setup

If you have never used a terminal, start with the desktop app or the web version. The terminal version is what most of this course shows, because it has the most features, and every lesson explains what to type.

What it can't do for you

Claude Code can make mistakes. It may misunderstand your request, write code that looks right and is not, or miss a detail of your project it never read. Treat its work the way you would treat a pull request from a new teammate: read it, run it, and test it before you rely on it. Later lessons cover permission modes, plan mode and review habits that make this easy.

It also needs access. Most surfaces need a Claude subscription or an Anthropic Console account, and the first time you start it you will be asked to sign in. The Installing and signing in lesson walks through this.

Mistakes beginners make

  1. Giving a vague goal. "Make the app better" leaves Claude Code guessing. "Add a loading spinner to the checkout button while the payment request is running" gives it something to act on.
  2. Approving everything without reading. The approval prompts exist so you can catch a wrong turn early. Read the change before you say yes.
  3. Expecting it to remember everything. Each session starts fresh unless you give it lasting instructions. The CLAUDE.md file lesson shows how a CLAUDE.md file solves this.
  4. Starting in the wrong folder. Claude Code works within the folder you open it in. Start it inside your project, not in your home directory.

Quick recap

  • Claude Code is an agentic coding tool. It reads your project, edits files, and runs commands to reach a goal you describe.
  • It asks for your approval before it makes changes that need it.
  • Chat tools work from pasted snippets, and editor autocomplete suggests lines. Claude Code handles multi-step tasks across a whole project.
  • It runs in the terminal, in VS Code and JetBrains, in the desktop app, and on the web.
  • Review everything it produces, as you would for any teammate's work.

Try it yourself

Before the next lesson, pick a small project you know well, or create an empty folder. Write down three tasks you would like help with, such as "explain what this project does" or "add a test for the login function". You will use them in Your first session, your first session.

Next lesson: How to Install Claude Code and Sign In. It covers installing Claude Code and signing in.