Getting Started with Claude Code: Your AI Development Companion

# Introduction

Claude Code represents a paradigm shift in how developers interact with AI. Unlike traditional code assistants that operate within IDE plugins, Claude Code lives directly in your terminal—where real development happens. It understands your entire codebase, executes commands, manages git workflows, and writes production-ready code through natural language conversations.

At Botique AI Solutions, we build our clients' custom AI systems using Claude Code. This guide shares everything we've learned about maximizing its potential.

# Prerequisites

Before diving in, ensure you have:

• Node.js 18+ installed on your system
• A terminal you're comfortable with (iTerm2, Warp, or the built-in terminal)
• An Anthropic API key (get one at console.anthropic.com)
• Basic familiarity with command-line interfaces

# Installation

Installing Claude Code is straightforward:

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

Verify the installation:

claude --version

You should see the current version number printed to the terminal.

# Initial Configuration

## Setting Up Your API Key

Claude Code needs your Anthropic API key to function. You can provide it in several ways:

# Option 1: Environment variable (recommended)
export ANTHROPIC_API_KEY="your-key-here"

# Option 2: Pass directly when starting
claude --api-key "your-key-here"

For persistence, add the export to your shell configuration (.zshrc, .bashrc, etc.).

## Creating Your First CLAUDE.md

The CLAUDE.md file is Claude Code's secret weapon. It's a special markdown file that Claude automatically reads when starting a conversation, providing persistent context about your project.

Create one in your project root:

# Project: My AI Application

## Tech Stack
- React 19 with TypeScript
- Vite for bundling
- Tailwind CSS for styling
- React Router for navigation

## Code Style
- Use functional components with hooks
- Prefer named exports
- Keep components under 200 lines
- Use absolute imports with @ alias

## Testing
- Run tests with: npm test
- Use Vitest for unit tests

## Common Commands
- `npm run dev` - Start development server
- `npm run build` - Production build
- `npm run lint` - Run ESLint

This context helps Claude write code that matches your project's conventions from the first interaction.

# Essential Commands

## Starting a Conversation

Simply type claude in your terminal to begin:

cd your-project
claude

Claude will read your CLAUDE.md and any relevant files to understand your project.

## The Explore, Plan, Code, Commit Workflow

Anthropic recommends this workflow for best results:

1. Explore: Ask Claude to read and understand relevant files
2. Plan: Request a plan before implementation (use "think" or "think harder")
3. Code: Have Claude implement the solution
4. Commit: Let Claude create commits with meaningful messages

Example session:

## Useful Slash Commands

Claude Code supports several built-in commands:

• /clear - Reset conversation context (use between unrelated tasks)
• /permissions - Manage tool access permissions
• /help - Show available commands
• /compact - Toggle compact output mode

# Best Practices

## Be Specific in Your Requests

Instead of vague requests, provide detailed context:

## Use Context Effectively

• Paste screenshots directly (Cmd+Ctrl+Shift+4 on macOS, then Ctrl+V)
• Reference specific files using tab completion
• Paste error messages in full
• Share relevant URLs for Claude to fetch

## Manage Context Window

For long sessions, use /clear between unrelated tasks. This prevents context overflow and keeps responses focused.

## Permission Management

Claude Code conservatively requests permissions. Customize using:

# Allow specific tools for the session
claude --allowedTools "Bash(npm run test)"

Or configure in .claude/settings.json:

{
  "permissions": {
    "allow": ["Read", "Write", "Edit"],
    "deny": ["Bash(rm -rf *)"]
  }
}

# Advanced Tips

## Multi-Claude Workflows

For complex projects, run multiple Claude instances:

1. One instance for writing code
2. Another for reviewing and testing
3. Use git worktrees for parallel work on features

git worktree add ../feature-auth feature/auth
cd ../feature-auth
claude

## Headless Mode for Automation

Use Claude in CI/CD pipelines:

claude -p "Run the test suite and fix any failing tests" --output-format json

## Custom Slash Commands

Create reusable prompts in .claude/commands/:

<!-- .claude/commands/review.md -->
Review the staged changes for:
- Code quality issues
- Potential bugs
- Performance concerns
- Security vulnerabilities

Provide specific, actionable feedback.

Use with: /project:review

# Common Pitfalls to Avoid

1. Skipping the plan phase - Always let Claude think before coding
2. Context overload - Use /clear between unrelated tasks
3. Vague prompts - Be specific about what you want
4. Ignoring CLAUDE.md - Keep it updated as your project evolves
5. Fighting the workflow - Let Claude handle git operations naturally

# Conclusion

Claude Code transforms terminal-based development from command execution to conversation. By providing rich context through CLAUDE.md, following the explore-plan-code-commit workflow, and leveraging advanced features like custom commands and multi-instance workflows, you can dramatically accelerate your development velocity.

Start with small tasks to build familiarity, then gradually tackle larger features. The more context you provide, the better Claude performs.

Ready to build your first AI-powered feature? Open your terminal and type claude.