# Claude Code Quickstart

Claude Code is a coding agent that lives in your terminal. You talk to it in
plain English, it reads and edits the files in your project, and it asks before
it changes anything. This guide gets you from nothing installed to your first
committed change.

The whole loop in about twenty seconds: ask, review, approve, commit. Illustrative session.

## How do I install Claude Code?

On macOS, Linux, or WSL, run the one-line installer:

```bash
curl -fsSL https://claude.ai/install.sh | bash
```

That is it. Claude Code ships as a single standalone CLI from Anthropic, so there is
nothing else to set up.

## How do I sign in to Claude Code?

Move into a project you already have with `cd`, then launch the agent:

```bash
cd path/to/your-project
```

Run `claude` to start Claude Code. The first run asks you to authenticate.Start `claude` from inside the project folder, not your home directory, or it has no code to work on. Make sure that folder is a git repo too, so every edit is something you can review and undo.

Sign in with a Claude subscription (Pro, Max, Team, or Enterprise), a Console
account with API credits, or a company cloud like Amazon Bedrock or Google
Vertex AI.

Your login is stored locally, so you only do this once.

## The loop: ask, review, approve

Working with Claude Code is a three-step rhythm.

1. **You ask** for something in plain language. No need to name files.

   > what does this project do?

2. **It reads** whatever it needs · your folder structure, the relevant files,
   recent git history · and answers or proposes an edit.

3. **It asks permission** before touching anything. You see the exact change
   before it lands, and you approve or reject it.

That permission step is the safety rail. If you want fewer prompts once you
trust it, press Shift+Tab to cycle through permission modes, from
ask-every-time to accept-edits automatically.It is tempting to jump straight to accept-edits to stop the prompts. On a codebase you do not know yet, stay on ask-every-time for the first few changes. Once you have seen the kind of edits Claude proposes, then slide along the scale.

Shift+Tab cycles the safe modes. The far end, run with --dangerously-skip-permissions, turns every check off: only ever inside a throwaway sandbox, never on your real machine.

## What is a CLAUDE.md file?

`CLAUDE.md` is a markdown file in your repo that Claude Code reads at the start of
every session. It is where you write the things you would otherwise repeat
every time: build commands, coding style, where things live, the rules that
matter.

Get a starter without writing it by hand · run /init and it drafts one from
your codebase. Commit the
file and your whole team gets the same conventions for free. Keep it tight ·
under about 200 lines · so the guidance stays sharp. There is a whole guide on
[rules and memory files](/guides/rules-and-memory-files) when you want to go
deeper.

## Which slash commands should I know first?

Type `/` to see everything Claude Code offers. You will reach for these first:

- **help** · list commands and docs
- **memory** · create or update your project's `CLAUDE.md`
- **clear** · wipe the conversation and start fresh
- **model** · switch which model is answering
- **plan** · a read-only mode that proposes a plan before editing (covered in
  the next guide)
- **login** · re-authenticate or switch accounts

You do not need to memorise these. Type `/` at any point and Claude Code lists every command with a short description, so you can find what you need as you go.

## Your first ten minutes

A realistic warm-up on a project you already have:

1. `cd` into the project and run `claude`.
2. Ask **"what does this project do?"** and let it map the codebase.
3. Ask follow-ups: **"where is the entry point?"**, **"what does the build
   command do?"**
4. Make a small change: **"add a hello world function to the main file."**
   Review the proposed edit and approve it.
5. Check your work: **"what files have I changed?"**
6. Commit: **"commit my changes with a clear message."**

Ten minutes in, you have made and committed a change without opening a single
file yourself. That is the whole idea: you stay in charge of *what* and *why*,
and hand off the mechanical *how*.

You stay in charge of what to build and why. Claude Code handles the mechanical how, one reviewed and approved change at a time. That review step is the whole point: speed without giving up control.

Once this feels natural, the next guide turns it into a real
[workflow](/guides/claude-code-workflows) · plan
mode, memory, subagents, and code review.