Claude Code for beginners: your first hour

Claude Code for beginners comes down to four steps in the first hour: install the claude command, sign in with a paid Claude plan, give it one small task in a project folder, and learn five slash commands. From then on, Claude Code reads your files, runs commands and makes changes from plain-English requests, as far as its permission mode allows.

We assume you have a code project and have opened a terminal before. If the terminal is new to you, Anthropic's quickstart links a terminal guide first.

What beginners need before starting#

• A paid Claude plan or a Console account. As of 22 October 2026, Anthropic's pricing page includes Claude Code in Pro ($20 a month, or $17 a month billed yearly) and Max (from $100 a month). The free plan doesn't include it. A Claude Console account works too, billed at API rates.
• A project folder: any code you're working on, ideally one that's already in git (more on that below).
• About ten minutes for setup. The rest of the hour is using it.

Step 1: Install Claude Code#

Open a terminal and run the command for your system, from Anthropic's quickstart:

Then open a new terminal window and run claude --version. What you might see:

• A version number followed by (Claude Code): it worked.
• claude not found: the install folder isn't on your PATH yet. claude: command not found has the fix for each system.
• 'irm' is not recognized as an internal or external command: you pasted the PowerShell line into CMD.

The curl and irm installs update themselves in the background. Homebrew and WinGet installs don't, so run brew upgrade claude-code or winget upgrade Anthropic.ClaudeCode now and then. On native Windows, Anthropic also recommends Git for Windows, so Claude Code can use its Bash tool. On Linux you can install from Anthropic's apt, dnf or apk repository instead (install Claude Code on Linux), and a Chromebook runs the same command once ChromeOS's Linux environment is on (Claude Code on a Chromebook).

Step 2: Sign in#

1. Go to your project folder and start Claude Code:

   bash

   cd /path/to/your/project
   claude
   

2. On first use it opens your browser. Pick your Claude account and approve; that computer stays signed in.

3. The prompt appears, with the version, the current model and the working directory above it.

4. To switch accounts later, type /login inside Claude Code.

If ANTHROPIC_API_KEY is set in your shell, Claude Code asks whether to use the key, and if you approve, it skips the browser login and bills the key. Anthropic's help page says this results in "API usage charges rather than using your subscription's included usage". If you meant to use your plan, decline and run unset ANTHROPIC_API_KEY. Sign-in loops are covered in Claude Code keeps asking me to log in.

Step 3: Your first task#

Ask a question before you ask for a change:

what does this project do?

Claude reads the files it needs and explains. Then try a small, safe change:

add a short "How to run it" section to the README, based on package.json

The status bar shows the permission mode, and the mode decides what happens next (permission modes):

• Auto mode, where a new session usually starts on recent versions: Claude makes the change without asking, and a second model, a classifier, reviews risky actions in the background.
• Manual mode: Claude shows the change and asks first. Read it, then choose Yes. claude --permission-mode manual starts a session in Manual.
• Shift+Tab cycles through the modes at any time.

Anthropic's quickstart gives two tips for the prompt itself. Name the problem: not "fix the bug" but "fix the login bug where users see a blank screen after entering wrong credentials". And break a big ask into steps, because a numbered list of three steps works better than one paragraph that asks for everything.

The five commands that matter#

Claude Code has dozens of slash commands, and a beginner needs five of them most days. The quoted descriptions are Anthropic's own, from the commands reference:

Four more keys and flags:

• Esc stops Claude mid-action, and Esc twice opens the same menu as /rewind.
• /help lists every other command.
• claude -c picks up the last conversation in this folder.
• claude -r lets you choose an earlier one from a list.

How the limits work#

Claude Code shares your plan's limits with the Claude app. Anthropic's pricing page says so in one line, "Your activity across Claude on web, desktop, mobile, and Claude Code all draws from the same pool", and sets out the clock in another: "Every plan has usage limits that reset on a rolling five-hour session window, and paid plans add weekly limits on top."

• A long coding session can use up your five-hour window, and chatting in the Claude app in the same window draws on the same pool.
• At the limit you wait for the reset, move to a higher plan, or, on a paid plan, turn on usage credits to keep working at API rates (the pricing page's own list).
• Claude Code uses more than chat, because it reads files and runs commands, and each file it reads is context.

The 5-hour limit explained covers the window in detail, and how to check Claude Code usage shows where to read what's left.

How not to break things#

Claude Code changes real files on your computer. Three habits keep that safe.

1. Commit to git before you start. Then any change is one git diff to read and one git restore to undo. You can even ask Claude to "commit my changes with a descriptive message". If the project isn't in git yet, run git init and make a first commit before anything else.
2. Don't treat rewind as a backup. Checkpoints are handy, but Anthropic warns they "only track changes made through Claude's file editing tools. Changes made through Bash commands or external processes are not captured. This isn't a replacement for git."
3. Read before you approve. New sessions often start in auto mode, which doesn't ask. Early on, switch to Manual mode (the status bar reads ⏸ manual mode on), which stops and asks before most file edits and shell commands. Approve actions one by one until you've seen how Claude works on your project, and allow more once you trust it.

Later, add a check Claude can run: "Run the tests and paste the output" turns "looks done" into "is done". We do this with a failing test first, written up in test-driven development with Claude Code.

If you'd rather have windows than a terminal#

Not everyone takes to the terminal. Anthropic itself offers a desktop app and a VS Code extension, and Claude Code desktop app alternatives compares the desktop app with the others.

Shuriken is one of those alternatives. It runs Claude Code on your own computer, with your own Claude account, in an app with windows. Each task is a chat, its changes sit in a Changes panel, and each new chat works on its own git branch, so trying something doesn't touch your main copy. On a new computer, Shuriken installs the tools that are missing. It also runs Codex, Gemini CLI and other agents the same way, should you want them later.

The ideas you learn here, from prompts to limits, carry over either way. If windows suit you better, download Shuriken.