Claude Code Complete Guide 2025: Installation, Commands & Best Practices

October 12, 202512 min read

Claude Code is Anthropic's official CLI tool that brings AI-powered development directly into your terminal. This comprehensive guide covers everything from installation to advanced workflows, helping you become a power user in 2025.

What is Claude Code?

Claude Code is an agentic coding assistant that lives in your terminal. Unlike IDE-based tools like Cursor or Copilot, Claude Code operates as a command-line interface that can read, write, and execute code across your entire project. It understands your codebase context and can perform complex multi-file operations autonomously.

100K+

Context window tokens

CLI

Terminal-native interface

MCP

Model Context Protocol

What sets Claude Code apart is its agentic nature—it doesn't just suggest code, it actively implements changes, runs tests, commits code, and even creates pull requests. Combined with MCP (Model Context Protocol), it can connect to external services like GitHub, databases, and APIs.

Installation & Setup

Getting started with Claude Code takes less than a minute. Here's how to install and configure it:

Step 1: Install via npm

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

Requires Node.js 18+ and npm. You can also use pnpm or yarn.

Step 2: Authenticate

$claude

On first run, Claude Code opens a browser window for OAuth authentication with your Anthropic account.

Step 3: Start Coding

$cd your-project

$claude

Navigate to any project directory and run claude to start an interactive session.

Essential Commands

Claude Code comes with powerful slash commands for common workflows. Here are the ones you'll use daily:

/help

Shows all available commands and their descriptions. Essential for discovering features.

/init

Initializes Claude Code in your project, creating a CLAUDE.md file with project context and guidelines.

/compact

Compresses conversation history to save tokens while maintaining context. Use when context gets too long.

/clear

Clears the current conversation and starts fresh. Useful when switching between unrelated tasks.

/cost

Shows token usage and cost for the current session. Great for budget tracking.

/doctor

Runs diagnostics to check your Claude Code installation and configuration.

/memory

Manage persistent memory that persists across sessions. Store preferences, patterns, and context.

MCP Servers: Extending Claude Code

The Model Context Protocol (MCP) is what makes Claude Code truly powerful. MCP servers let Claude connect to external tools and services—think of them as plugins that expand what Claude can do.

Popular MCP Servers

GitHub MCP

Create issues, PRs, manage repos, and review code directly through Claude.

PostgreSQL MCP

Query databases, run migrations, and analyze data without leaving the terminal.

Filesystem MCP

Enhanced file operations with watch capabilities and batch processing.

Slack/Discord MCP

Send messages, read channels, and integrate with team communication.

Configure MCP in settings.json

{
  "mcpServers": {
    "github": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-github"],
      "env": {
        "GITHUB_TOKEN": "your-token-here"
      }
    }
  }
}

Add this to ~/.claude/settings.json or run /mcp to configure interactively.

Hooks: Customize Claude's Behavior

Hooks let you run custom scripts before or after Claude performs actions. They're powerful for enforcing coding standards, running tests automatically, or integrating with your workflow.

Pre-commit Hook

Run linting and formatting before Claude commits code. Catches issues before they hit your repo.

{
  "hooks": {
    "pre-commit": "npm run lint && npm run format"
  }
}

Post-edit Hook

Automatically run tests after Claude modifies files. Ensures changes don't break existing functionality.

{
  "hooks": {
    "post-edit": "npm test -- --related"
  }
}

CLAUDE.md: Project Context File

The CLAUDE.md file is your project's instruction manual for Claude. It contains context, conventions, and guidelines that help Claude understand your codebase and write better code.

Example CLAUDE.md

# Project: My SaaS App

## Tech Stack
- Next.js 14 with App Router
- TypeScript (strict mode)
- Tailwind CSS
- Prisma + PostgreSQL

## Conventions
- Use kebab-case for file names
- Components go in src/components/
- API routes in src/app/api/
- Always use async/await, never .then()

## Testing
- Jest for unit tests
- Playwright for E2E
- Run `npm test` before committing

## Important Notes
- Never commit .env files
- Always add loading states
- Use server components by default

Run /init to generate a CLAUDE.md file automatically, or create one manually with your specific guidelines.

Pro Tips & Best Practices

1. Use Headless Mode for Automation

Run claude -p "your prompt" for one-off commands in scripts and CI/CD pipelines. Great for automated code reviews and refactoring.

2. Leverage Context Files

Use @filename to include specific files in your prompt. Claude automatically reads them and uses them as context.

3. Track Your Usage

Claude Code saves usage data in ~/.claude/cc.json. Upload it to Viberank to track your stats and see how you compare to other developers.

4. Use /compact Regularly

Long conversations eat tokens fast. Use /compactperiodically to compress history while keeping important context.

5. Combine with Conductor

For parallel development, use Conductor to run multiple Claude Code instances in isolated git worktrees. Build features simultaneously without conflicts.

Track Your Claude Code Journey

Curious how your Claude Code usage compares to other developers? Viberank analyzes yourcc.jsonfile to show detailed analytics about your AI-assisted development workflow.

Upload your stats and join the leaderboard

$npx viberank

See your token usage, session patterns, and how you stack up against the community