Installing and Setting Up Claude Code
Installing and Setting Up Claude Code
~7 Min. Lesezeit Zuletzt aktualisiert am August 8, 2026
From here on, this series is almost entirely about Claude Code. In this chapter you'll install it and set it up once, so you can dive straight into your first chat in the next chapter.
Requirements
Claude Code is a command-line tool and runs on Windows, Mac, and Linux. To install it you need Node.js (the JavaScript runtime, which also brings along the npm package manager) and a terminal - on Windows, for example, "Command Prompt", "PowerShell", or Windows Terminal; on Mac and Linux, the respective default terminal program.
If you're not sure whether Node.js is already installed, check with a quick command in the terminal:
node --version
npm --versionIf both commands show a version number, you're all set. If instead you see an error like "command not found", install Node.js first from nodejs.org (the installer for the recommended, current version is perfectly sufficient).
Installing via npm
Claude Code itself is installed as a global npm package - "global" means the claude command is then available in any project folder afterwards, not just in a single project:
npm install -g @anthropic-ai/claude-codeOnce the installation finishes, you can check that everything worked:
claude --versionTipp: If the installation fails with a permission error: on some systems, installing global npm packages requires extra permissions. In that case, look up the current installation instructions for your operating system rather than simply repeating the command with elevated privileges (e.g. sudo) - that can cause further permission problems later on.
First launch and login
Switch to a project folder in your terminal (more on that in the next chapter) and start Claude Code with the simple command:
claudeOn the very first launch, Claude Code walks you through a short login process: you're asked to authenticate with your Anthropic account, usually via a link that opens in your default browser. Once you've signed in successfully, you're automatically returned to the terminal, and Claude Code is ready to use.
You typically only need to log in once per machine - every subsequent time you run claude, you're already signed in and land straight in a new session.
Finding your way around
Once Claude Code has started, you'll see a simple input line, similar to Claude.ai - just in the terminal instead of the browser. You can immediately ask a question or describe a task in plain language. In addition, you have access to so-called slash commands: short commands that start with a slash, for example /help for an overview of the available commands. Block 4 of this series covers slash commands in detail.
Achtung: End a session properly instead of simply closing the terminal window - for example with the key combination shown in the terminal itself as "exit", or by leaving the program the normal way. That way, work in progress doesn't get lost mid-execution.
That completes the setup. In the next chapter you'll open your first real project folder and ask Claude Code your first questions.