Getting Started with Claude Code

Install Claude Code, configure your project, and run your first session.

Install Claude Code

Claude Code is a CLI tool that runs in your terminal. Install it with one command:

macOS / Linux (recommended):

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

Windows (PowerShell):

irm https://claude.ai/install.ps1 | iex

Via npm (requires Node.js 18+):

npm install -g @anthropic-ai/claude-code

You will need a Pro, Max, Team, Enterprise, or Console account to authenticate. The free Claude.ai plan does not include Claude Code. You can also bring your own provider via Amazon Bedrock↗, Google Vertex AI↗, or Microsoft Foundry↗.

First Run

Navigate to any project directory and type claude to start an interactive session:

cd my-project
claude

Claude will open a browser window for authentication on first run. After that, it analyzes your project structure and begins an interactive conversation. Try these first commands:

give me an overview of this codebase
what does @src/index.ts do?

Configure CLAUDE.md

Create a CLAUDE.md file in your project root to give Claude persistent context. You can auto-generate one:

/init

Or create it manually with the essentials:

# Project: MyApp

## Build & Test

- npm run dev # start dev server
- npm test # run tests
- npm run lint # lint check

## Architecture

- React 18 + TypeScript frontend
- Express API at /api/v1
- PostgreSQL via Prisma ORM

## Conventions

- Functional components with hooks
- Named exports, kebab-case files
- Tests colocated in __tests__/

Keep it concise. For every line, ask: if I remove this, will Claude make a mistake? If not, remove it. See the CLAUDE.md deep guide for advanced patterns.

Add Your First Skill

For reusable workflows, create a skill instead of growing CLAUDE.md:

---
name: review
description: Use when reviewing current changes before a commit or PR.
---

Review the current changes for:
- Potential bugs and edge cases
- Security vulnerabilities
- Performance issues
Suggest concrete fixes with file paths.

Skills are auto-discovered and load only when relevant. See Skills for the full pattern.

Understand Permissions

Claude Code uses a layered permission system. Press Shift + Tab to cycle the default three modes:

• default — Claude asks before every write or shell command.
• acceptEdits — File edits and common filesystem commands run without prompting.
• plan — Claude explores via reads and shell, but won't edit your source.

Three more modes (auto, dontAsk, bypassPermissions) are opt-in — see Permissions & Security.

Launch directly in a mode:

claude --permission-mode plan          # read-only exploration
claude --permission-mode acceptEdits   # auto-approve edits

See Permissions & Security for the full picture including per-command allow/deny rules.

Session Naming & Resume

Name sessions you'll want to return to:

claude --name "auth-refactor"    # set a name at launch (-n for short)

Resume later by name or pick from a list:

claude --resume "auth-refactor"  # resume by name (-r for short)
claude --continue                # resume most recent (-c for short)

Create an isolated parallel worktree when you want to experiment safely:

claude --worktree

Use /memory to inspect persistent context and /agents to manage subagents.

Related Reading

• Keyboard Shortcuts — Master the CLI shortcuts
• CLAUDE.md & Memory — Advanced context configuration
• Claude Code Architecture — Understand the full harness
• Plan Mode — Safe read-only exploration