Introduction to Claude Code

Claude Code is Anthropic's agentic AI coding assistant that runs in your terminal. Unlike IDE-based tools like Cursor or GitHub Copilot, Claude Code operates as an autonomous agent—it can explore your codebase, run commands, make changes across files, and iterate until tasks are complete.

What is Claude Code?

Claude Code brings Claude's capabilities directly into your development workflow:

• Terminal-native: Works in any terminal, IDE-agnostic
• Agentic: Autonomously completes complex tasks
• Full codebase access: Reads and understands your entire project
• Tool use: Runs commands, edits files, searches code
• Interactive: Confirms actions before executing

How Claude Code Differs

Traditional AI Assistants

Most AI coding tools work reactively:

1. You ask a question
2. AI gives an answer
3. You manually implement it
4. Repeat

Claude Code (Agentic)

Claude Code works autonomously:

1. You describe what you want
2. Claude Code explores your codebase
3. It plans the approach
4. It makes changes across multiple files
5. It runs tests to verify
6. It iterates until complete

Key Capabilities

Codebase Understanding

Claude Code automatically reads and indexes your project:

You: What does this project do?

Claude Code: [Reads files, analyzes structure]
This is a Next.js e-commerce application with:
- User authentication via NextAuth
- Product catalog with PostgreSQL backend
- Shopping cart using Zustand
- Stripe checkout integration
...

Multi-File Editing

Make coordinated changes across your project:

You: Add a "favorites" feature to the product system

Claude Code: I'll add a favorites feature. This will involve:
1. Adding a favorites table to the database schema
2. Creating API routes for add/remove favorites
3. Adding a favorites button to ProductCard
4. Creating a Favorites page to view saved items

Let me implement this...
[Makes changes to 6 files]

Command Execution

Claude Code can run terminal commands:

You: Run the tests and fix any failures

Claude Code: Running tests...
[Executes: npm test]

Found 3 failing tests. Let me analyze and fix them...
[Reads test output, identifies issues]
[Edits source files to fix bugs]
[Runs tests again]

All 47 tests passing.

Iterative Problem Solving

Claude Code doesn't just try once:

You: Build and deploy this to Vercel

Claude Code: Starting build...
[Runs: npm run build]

Build failed - TypeScript error in src/utils.ts line 45.
Let me fix that...
[Edits file]
[Runs build again]

Build successful. Deploying to Vercel...
[Runs: vercel deploy]

Deployed to: https://your-app.vercel.app

When to Use Claude Code

Great For

✅ Complex refactoring across many files ✅ Bug investigation and fixing ✅ Adding new features end-to-end ✅ Understanding unfamiliar codebases ✅ Writing comprehensive tests ✅ Code reviews and improvements ✅ Build and deployment automation

Consider Alternatives For

⚠️ Quick inline completions (IDE tools are faster) ⚠️ Real-time pair programming (IDE integration is smoother) ⚠️ Visual development (AI builders may be better)

The Agentic Difference

Autonomy

Traditional tools wait for you. Claude Code takes initiative:

You: Make the checkout faster

Claude Code: I'll analyze the checkout flow...

Found these performance issues:
1. Unnecessary re-renders in CartSummary
2. N+1 query in order creation
3. Large bundle size from moment.js

I'll fix each of these:
[Implements React.memo]
[Optimizes database query]
[Replaces moment with date-fns]

Checkout now loads 3x faster. Want me to add performance tests?

Context Retention

Claude Code remembers your conversation:

You: Let's add user profiles

Claude Code: [Implements user profiles]

You: Add avatar upload to that

Claude Code: [Adds to the profile system it just created]

You: Now add social links

Claude Code: [Extends the same feature]

Verification

Claude Code confirms its work:

Claude Code: I've made the changes. Let me verify:
- Running TypeScript check... ✓
- Running tests... ✓ (47/47 passing)
- Running linter... ✓
- Building project... ✓

All checks pass. The feature is ready.

How Sessions Work

Starting a Session

Launch Claude Code in your project directory:

cd your-project
claude

During a Session

Claude Code maintains context of:

• Files it has read
• Changes it has made
• Commands it has run
• Your conversation history

Ending a Session

Exit with /exit or Ctrl+C. Your conversation history is saved.

Claude Code Architecture

┌─────────────────────────────────────────────────────┐
│                    Your Terminal                     │
├─────────────────────────────────────────────────────┤
│                    Claude Code                       │
│  ┌─────────────┐ ┌─────────────┐ ┌─────────────┐   │
│  │ File System │ │  Terminal   │ │   Search    │   │
│  │    Tool     │ │    Tool     │ │    Tool     │   │
│  └─────────────┘ └─────────────┘ └─────────────┘   │
├─────────────────────────────────────────────────────┤
│                   Claude API                         │
│           (claude-3.5-sonnet / opus)                │
└─────────────────────────────────────────────────────┘

Pricing Model

Claude Code uses API tokens:

• Input tokens: What you send (prompts, file content)
• Output tokens: What Claude generates (responses, code)
• Cost varies: By model (Sonnet is cheaper than Opus)

Typical costs:

• Simple questions: $0.01-0.05
• Complex tasks: $0.10-0.50
• Major refactors: $1-5

What You'll Learn

This module covers:

1. Installation: Get Claude Code running
2. Basic Commands: Essential interactions
3. Project Context: How Claude Code understands your code
4. Agentic Workflows: Complete real tasks
5. MCP Integration: Extend with custom tools
6. Advanced Patterns: Expert techniques

Summary

• Claude Code is an agentic AI that runs in your terminal
• Autonomous: It explores, plans, and implements solutions
• Full access: Reads your codebase, runs commands, edits files
• Iterative: Verifies work and fixes issues
• IDE-agnostic: Works with any editor, any project

Next Steps

Let's get Claude Code installed and configured on your system.