Getting Started with
Claude Code

What is Claude Code?

Claude Code is Anthropic's official CLI tool that brings Claude directly into your development workflow. It runs in your terminal, has full access to your filesystem and shell, and can read, write, and execute code in your projects.

Unlike chat-based AI tools, Claude Code operates as an agentic coding assistant — it can explore your codebase, make multi-file changes, run tests, commit code, and execute complex multi-step tasks autonomously.

Installation

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

Copycd your-project
claude

claude "What files are in this project? Give me a brief overview."

Creating Your CLAUDE.md

A CLAUDE.md file in your project root is the single most impactful thing you can do to improve Claude Code's output. It tells Claude about your architecture, conventions, and hard constraints.

Quick Start Template

# ProjectName — CLAUDE.md

## Project Identity
Brief description of what this project is and its tech stack.

## Architecture Rules
- Key patterns to follow (service layer, component structure, etc.)
- Directory structure conventions
- Import ordering rules

## Hard Constraints
```
NEVER: Describe things Claude must never do
WHY: Explain the consequence
INSTEAD: Provide the correct approach
```

## Testing
- Test framework and how to run tests
- Minimum test requirements for new code

## Git
- Branch naming conventions
- Commit message format

Where CLAUDE.md Files Live

• Project root — ./CLAUDE.md — Read when Claude Code runs in this project
• Home directory — ~/.claude/CLAUDE.md — Read for ALL projects (global preferences)
• Subdirectories — ./src/CLAUDE.md — Read when Claude navigates into that directory (useful for monorepos)

Essential Commands

Interactive Mode

# Start an interactive session
claude

# Start with a specific task
claude "Refactor the auth module to use JWT tokens"

# Resume the last conversation
claude --continue

One-Shot Mode

# Run a single command and exit (great for scripts)
claude -p "Generate a migration to add email_verified column to users table"

# Pipe input
cat error.log | claude -p "What's causing these errors?"

Useful Flags

# Use a specific model
claude --model claude-sonnet-4-20250514

# Set output format (for scripting)
claude -p "List all TODO comments" --output-format json

# Limit tool access
claude --allowedTools "Read,Grep,Glob"

Slash Commands (in interactive mode)

/help              — Show available commands
/clear             — Clear conversation history
/compact           — Summarize conversation to save context
/cost              — Show token usage and cost
/doctor            — Diagnose configuration issues
/memory            — View and edit CLAUDE.md files

Configuring MCP Servers

MCP (Model Context Protocol) servers extend Claude Code's capabilities by connecting it to external tools and services — databases, browsers, APIs, and more.

Adding an MCP Server

# Add a server (project-level)
claude mcp add server-name -- command arg1 arg2

# Example: Add a PostgreSQL MCP server
claude mcp add postgres -- npx @anthropic-ai/mcp-postgres "postgresql://..."

# Example: Add Playwright for browser automation
claude mcp add playwright -- npx @anthropic-ai/mcp-playwright

# List configured servers
claude mcp list

Popular MCP Servers

Productivity Tips

1. Be Specific in Your Requests

2. Use /compact Regularly

Long conversations consume context window. Use /compact to summarize the conversation and free up space for more code analysis.

3. Let Claude Run Tests

After making changes, tell Claude to run your test suite. It can read the output, identify failures, and fix them in the same session.

4. Use CLAUDE.md Hard Constraints

If Claude keeps making the same mistake (wrong import path, wrong pattern, etc.), add it as a NEVER constraint in CLAUDE.md. This eliminates the issue permanently.

5. Commit Regularly

Ask Claude to commit after each logical change. This gives you clean rollback points if something goes wrong.