DS DevShelfHub Projects · AI tools
Tutorials / Claude Code / Installation
Claude Code Beginner · 8 min read Page 3 of 25

Installation & First Session

By DevShelfHub

Get Claude Code running with the install method that's right for your OS, authenticate with your subscription or API key, and complete your first real session — then learn the handful of commands you'll reach for every day.

Series progress3 / 25
Claude Code installation tutorial — OS setup, authentication, and first session commands

Before you start

You'll need three things in place:

  • macOS, Linux, WSL, or Windows with PowerShell or CMD
  • An active Claude account at claude.ai, or an API key from console.anthropic.com
  • A real code project directory to experiment in — not a throwaway folder

Choosing your install method

The native installer is recommended for most users — it auto-updates in the background, so you always have the latest version. Run the line that matches your platform:

bash 
# macOS / Linux / WSL
curl -fsSL https://claude.ai/install.sh | bash

# Windows PowerShell
irm https://claude.ai/install.ps1 | iex

# Windows CMD
curl -fsSL https://claude.ai/install.cmd -o install.cmd && install.cmd && del install.cmd

Prefer a package manager? Those work too — just remember they do not auto-update, so you'll need to upgrade manually:

bash 
# Homebrew (macOS) — upgrade with: brew upgrade claude-code
brew install --cask claude-code

# WinGet (Windows)
winget install Anthropic.ClaudeCode

# Linux package managers: apt, dnf, apk are supported.
# See /en/setup for repository configuration.

Authentication

On first run, Claude Code opens a browser for OAuth login with your claude.ai account. That's all most people need. If you'd rather authenticate with an API key, set it in your environment instead:

bash 
export ANTHROPIC_API_KEY="sk-ant-..."

Note: Third-party providers such as Amazon Bedrock and Google Vertex require additional environment variables — we cover those in the advanced production chapters.

Your first session

Move into a real project and launch Claude:

bash 
cd your-project-directory
claude

Claude loads CLAUDE.md (if present), your auto memory, and system context, then shows a prompt. Ask it for the lay of the land — this is the "gather context" phase in action:

text 
> What does this codebase do? Give me a high-level overview.

Claude reads your files and explains the project back to you. You've completed the first phase of the agentic loop.

Essential commands

A small set of slash commands and key presses cover almost everything you'll do day to day:

Command What it does
/help List all available slash commands
/doctor Diagnose installation issues — your first stop when something's off
/model Switch models mid-session
/init Walk through creating a CLAUDE.md for your project
/context See what's consuming your context window
/compact Manually compact conversation history to free up space
Esc Interrupt Claude mid-action
Esc Esc Rewind to a previous checkpoint
Shift+Tab Cycle through permission modes

If something goes wrong

  • /doctor catches most installation problems automatically.
  • Using API key auth? Confirm ANTHROPIC_API_KEY is actually set in your shell.
  • For platform-specific quirks, see /en/troubleshoot-install.

Best practice: Read the first 20 lines of every Claude response before interrupting — its visible reasoning often explains exactly what it's about to do.

Common mistakes & best practices

Avoid

  • Using Homebrew/WinGet then wondering why you're behind on features (no auto-update)
  • Experimenting in an empty throwaway folder — it limits what you can learn
  • Hitting Enter before reading Claude's output and missing important context

Do

  • Start with a real project, not a toy example
  • Use the native installer so you get automatic updates
  • Read the first ~20 lines of every response before interrupting

Hands-on exercise

Install Claude Code, navigate to a real project (not a throwaway folder), and complete Quickstart steps 1–8: get a high-level overview, explore a file, make your first edit, run the test suite, and create a git commit — all through natural language.

Claude Code Installation FAQ

How do I install Claude Code?

Use the native installer for your platform: curl -fsSL https://claude.ai/install.sh | bash on macOS, Linux, or WSL, or the PowerShell or CMD command on Windows. The native installer is recommended because it auto-updates in the background.

Can I install Claude Code with Homebrew or WinGet?

Yes. brew install --cask claude-code works on macOS and winget install Anthropic.ClaudeCode works on Windows, and apt, dnf, and apk are supported on Linux. Package-manager installs do not auto-update, so you have to upgrade manually.

Do I need an API key to use Claude Code?

No. Most people authenticate through browser OAuth with a claude.ai account on first run. If you prefer, you can set ANTHROPIC_API_KEY in your environment to authenticate with an Anthropic API key instead.

What commands should I learn first in Claude Code?

Start with /help, /doctor, /init, /context, and /compact, plus the key presses Esc to interrupt, Esc Esc to rewind to a checkpoint, and Shift+Tab to cycle permission modes.

Claude Code won't start — how do I fix it?

Run /doctor first; it catches most installation problems automatically. If you authenticate with an API key, confirm ANTHROPIC_API_KEY is actually set in your shell, and check the troubleshoot-install docs for platform-specific quirks.

Quick summary

  • Use the native installer for automatic background updates
  • Homebrew and WinGet work but require manual upgrades
  • Authenticate via browser OAuth, or set ANTHROPIC_API_KEY
  • Learn the core commands early: /help, /doctor, /init, /context, /compact, Esc, Shift+Tab
← Previous Next: The Agentic Loop in Practice →