Claude CodeGetting Started

Getting Started with Claude Code

This guide will walk you through installing Claude Code, understanding its interface, and running your first AI-powered development sessions.

9 min read

Getting Started with Claude Code

This guide will walk you through installing Claude Code, understanding its interface, and running your first AI-powered development sessions.

Time: 30-60 minutes Prerequisites: Command line familiarity, Anthropic API key or Claude Pro subscription


Installation

Option 1: NPM (Recommended)

Option 2: Homebrew (macOS/Linux)

Option 3: Download Binary

Download from https://claude.ai/download for your platform:

  • macOS (Apple Silicon/Intel)
  • Linux (x86_64)
  • Windows (x64)

Authentication

With Anthropic API Key

With Claude Pro Subscription


First Launch

Start Claude Code

Initial Setup

On first launch:

  1. Accept terms of service
  2. Choose default model (Sonnet 4.5 recommended)
  3. Configure base permissions
  4. Claude creates ~/.claude/ directory

Interface Overview

Terminal Interface

Key Keyboard Shortcuts

Shortcut Action
Tab Toggle extended thinking
Shift+Tab (twice) Enter/exit Plan Mode
Escape (twice) Rewind to previous checkpoint
Ctrl+C Cancel current operation
Ctrl+D Exit Claude Code
Ctrl+R Reverse history search
↑/↓ Navigate command history

Your First Session

Step 1: Explore Your Codebase

Step 2: Ask Questions

Step 3: Make Changes

Step 4: Test Changes

Step 5: Commit


Essential Slash Commands

Get Help

Context Management

Workflow Commands


Configure Your First Settings

Create Project Settings

Add to .gitignore

Create Team Settings

Share settings with your team by committing .claude/settings.json:


Your First Custom Command

Create a Code Review Command

Use the Command


Your First Hook

Auto-Lint on File Edits

Now Claude automatically runs lint after any file edit!


Your First MCP Server

Add GitHub Integration

Use GitHub Integration


Common Workflows

Feature Development

Bug Fix Workflow

Code Review Workflow


Understanding Modes

Plan Mode (Shift+Tab twice)

When to use:

  • Unfamiliar codebase
  • Complex changes
  • Want to review before applying

What it does:

  • Read-only exploration
  • No file edits
  • No command execution
  • Creates detailed plan

Example:

Extended Thinking Mode (Tab)

When to use:

  • Complex algorithms
  • Architecture decisions
  • Security analysis
  • Performance optimization

Levels:

  • Tab or "think" - Basic thinking
  • "think hard" - More reasoning
  • "think harder" - Deep analysis
  • "ultrathink" - Maximum reasoning

Example:


Troubleshooting

Issue: "Permission Denied" for Bash Commands

Solution:

Issue: Can't Find Files

Solution:

Issue: Context Window Full

Solution:

Issue: MCP Server Not Working

Solution:


Next Steps

Immediate (Day 1)

  1. Complete Tutorial 1: Your First AI-Assisted Feature
  2. Create your first custom command
  3. Set up a simple hook
  4. Practice the explore → plan → code workflow

Short-term (Week 1)

  1. Configure permissions for your project
  2. Add an MCP server (GitHub or database)
  3. Create team-shared commands
  4. Read Best Practices

Medium-term (Month 1)

  1. Build custom hooks for your workflow
  2. Create project-specific skills
  3. Optimize context usage
  4. Explore advanced features

Quick Reference Card

Essential Commands

Common Prompts


Resources

Documentation

Community


Ready for hands-on practice?Start Tutorial 1

Have questions? → Check the Resources page for help


← Back to Claude Code Overview | Next: Use Cases →

Stay in the loop

Get weekly insights on data engineering, analytics, and AI—delivered straight to your inbox.

No spam. Unsubscribe anytime.