How to Install Claude Code: Step-by-Step Guide

You have probably heard by now. Claude Code burns through usage limits! But most of us live in the web app… distant from the terminal app, around which the buzz is about.

Maybe that was enough to make you curious. Maybe you already knew exactly what it was and just want it running on your machine. Either way, this guide is for you.

In this article, you’ll learn how to install Claude Code from downloading it to running your first command.

Installation

1. Run the install command

First, open a terminal and run the line for your platform. No administrator rights needed.

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

A successful install. If you got the yellow setup note, that means Step 2 applies to you.

2. (Optional) Add it to your PATH

Without this, claude returns “command not found”. On Zsh (the macOS default):

echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.zshrc && source ~/.zshrc

On Bash, swap both ~/.zshrc references for ~/.bashrc. No warning? Skip ahead.

3. Verify

claude --version

Prints a version such as 2.1.226 (Claude Code). For fuller diagnostics without starting a session, run claude doctor.

First Run

Move into the project you want to work in, then launch:

cd ~/your-project
claude

4. Pick a theme

Arrow keys, then Enter. Change it later with /theme.

The theme picker, with colourblind-friendly and ANSI-only options.

5. Choose your login method

Have a Claude subscription? Pick option 1 and press Enter. (The three options are compared in Part 3.)

Subscription, Console billing, or a third-party cloud platform.

6. Authorise in the browser

Your browser opens a consent screen. Click Authorize.

The consent screen. What the permissions actually mean is covered in Part 3.

7. No browser? Use the manual flow

On a server, container, or SSH session, Claude Code prints a URL instead. Press c to copy it, open it anywhere you can browse, and approve the same screen.

The fallback prompt, waiting for a code.

The authentication code (redacted here) is like a password.

Paste it at the terminal prompt and press Enter. Input is masked as you paste.

Code pasted and masked.

8. Check the account, press Enter

9. Accept the terminal settings

Choose “Yes, use recommended settings“. This enables Option+Enter for newlines, without it, multi-line prompts send half-written.

10. Trust the folder

Read the path it names, confirm it is yours, then choose “Yes, I trust this folder“.

The workspace trust prompt. Make sure you trust the folder you’re giving in all the rights to.

11. You are in

Accept the fullscreen renderer if offered and it removes flicker on long responses and adds mouse support. A good features to have for those used to the traditional GUI interface.

The welcome panel: model, account, working directory, recent changes.

Then type an instruction in plain English and press Enter. Press ? for shortcuts.

The prompt, ready for work.

Read more: How to Create Custom Skills in Claude: A Step-by-Step Guide

Conclusion

Installing Claude Code comes down to three moves:

1. one command
2. one browser approval
3. a wizard you can clear in two minutes

Only two screens deserve a careful read, the consent screen, which ties terminal sessions to your usual usage allowance, and the folder trust prompt, which grants access to everything in that directory.

Keep the scope narrow from here: launch inside a project, run /init where you’ll return, and check /usage early. Do that, and everything else the tool offers builds on the clean install you just finished.

Full documentation lives at docs.claude.com/en/docs/claude-code.

Studying, evaluating, and explaining AI systems for over 6 years.

“𝘖𝘯𝘤𝘦 𝘮𝘦𝘯 𝘵𝘶𝘳𝘯𝘦𝘥 𝘵𝘩𝘦𝘪𝘳 𝘵𝘩𝘪𝘯𝘬𝘪𝘯𝘨 𝘰𝘷𝘦𝘳 𝘵𝘰 𝘮𝘢𝘤𝘩𝘪𝘯𝘦𝘴 𝘪𝘯 𝘵𝘩𝘦 𝘩𝘰𝘱𝘦 𝘵𝘩𝘢𝘵 𝘵𝘩𝘪𝘴 𝘸𝘰𝘶𝘭𝘥 𝘴𝘦𝘵 𝘵𝘩𝘦𝘮 𝘧𝘳𝘦𝘦. 𝘉𝘶𝘵 𝘵𝘩𝘢𝘵 𝘰𝘯𝘭𝘺 𝘱𝘦𝘳𝘮𝘪𝘵𝘵𝘦𝘥 𝘰𝘵𝘩𝘦𝘳 𝘮𝘦𝘯 𝘸𝘪𝘵𝘩 𝘮𝘢𝘤𝘩𝘪𝘯𝘦𝘴 𝘵𝘰 𝘦𝘯𝘴𝘭𝘢𝘷𝘦 𝘵𝘩𝘦𝘮.” — 𝖥𝗋𝖺𝗇𝗄 𝖧𝖾𝗋𝖻𝖾𝗋𝗍, 𝖣𝗎𝗇𝖾