Claude Code Concepts: CLI, Hooks, Skills, and More

Overview

Claude Code consists of several components working together:

┌─────────────────────────────────────────┐
│              Claude Code                │
├─────────────────────────────────────────┤
│  CLI ──→ Main interface                 │
│  Hooks ──→ Behavior customization       │
│  Skills ──→ Extended capabilities       │
│  MCP ──→ Communication protocol         │
└─────────────────────────────────────────┘

1. CLI (Command Line Interface)

What Is It?

CLI is the primary way to interact with Claude Code via Terminal.

Basic Commands

# Start a new session
claude
 
# Start with a specific task
claude "Add authentication system"
 
# Resume a previous session
claude --resume
 
# Non-interactive mode (for scripts)
claude -p "Fix lint errors" --no-input

In-Session Commands

Important Flags

# Specify model
claude --model claude-sonnet-4
 
# Enable deep thinking
claude --thinking
 
# Verbose mode
claude --verbose
 
# Specify working directory
claude --cwd /path/to/project

2. Hooks

What Are They?

Hooks let you execute custom code at specific points in Claude Code's workflow.

Hook Types

PreToolUse ──→ Before any tool use
PostToolUse ──→ After tool succeeds
Notification ──→ When notification arrives
Stop ──→ When task finishes

Example: Hook to Prevent File Deletion

// ~/.claude/settings.json
{
  "hooks": {
    "PreToolUse": [
      {
        "matcher": "bash",
        "command": "check-dangerous-commands.sh",
        "timeout": 5000
      }
    ]
  }
}

# check-dangerous-commands.sh
#!/bin/bash
if echo "$CLAUDE_TOOL_INPUT" | grep -q "rm -rf"; then
  echo '{"decision": "block", "reason": "Dangerous deletion blocked"}'
  exit 0
fi
echo '{"decision": "allow"}'

Hook Use Cases

3. Skills

What Are They?

Skills are additional capabilities you add to Claude Code via SKILL.md files.

Skill Structure

skills/
└── my-skill/
    ├── SKILL.md      # AI instructions
    ├── scripts/      # Helper scripts
    └── templates/    # Templates

Example: API Helper Skill

# SKILL.md
---
name: api-helper
description: Helper for working with REST APIs
---
 
## When to Use
When the user requests creating or modifying API endpoints.
 
## Instructions
1. Use Express.js for backend
2. Add validation with Zod
3. Add unified error handling
4. Document endpoints in Swagger
 
## Templates
Use `scripts/create-endpoint.sh` to create new endpoints

Common Skills

4. MCP (Model Context Protocol)

What Is It?

MCP is an open protocol for connecting AI to external data sources and tools.

How It Works

┌──────────────┐     ┌─────────────┐     ┌──────────────┐
│ Claude Code  │────▶│ MCP Server  │────▶│ External API │
└──────────────┘     └─────────────┘     └──────────────┘

Setting Up MCP Server

// ~/.claude/settings.json
{
  "mcpServers": {
    "github": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-github"],
      "env": {
        "GITHUB_TOKEN": "ghp_xxx"
      }
    },
    "postgres": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-postgres"],
      "env": {
        "DATABASE_URL": "postgresql://..."
      }
    }
  }
}

Available MCP Servers

• GitHub: Manage repos, issues, PRs
• Postgres/MySQL: Database queries
• Slack: Send messages
• Google Drive: Read/write files
• Notion: Manage pages
• And more...

5. CLAUDE.md

What Is It?

A file you place in your project root to help Claude Code understand project context.

Suggested Contents

# CLAUDE.md
 
## About the Project
E-commerce application with Next.js and Supabase.
 
## Technologies
- Next.js 14 (App Router)
- TypeScript
- Tailwind CSS
- Supabase (Auth + Database)
 
## Structure
- `app/` - Next.js pages
- `components/` - React components
- `lib/` - utilities
- `supabase/` - migrations
 
## Commands
- `npm run dev` - local development
- `npm run build` - production build
- `npm test` - run tests
 
## Code Rules
- Use TypeScript strict mode
- Each component in separate file
- CSS with Tailwind only

Putting It All Together

Claude Code
    │
    ├── CLI ────────────── User interface
    │
    ├── CLAUDE.md ──────── Project context
    │
    ├── Hooks ─────────── Behavior customization
    │   ├── PreToolUse
    │   ├── PostToolUse
    │   └── ...
    │
    ├── Skills ─────────── Extended capabilities
    │   ├── SKILL.md
    │   └── scripts/
    │
    └── MCP ────────────── External connections
        ├── GitHub
        ├── Database
        └── ...

What's Next?

In the next article, we'll cover:

• Practical benefits of Claude Code
• Real use cases
• When to use it and when not to

Read Next: Claude Code Benefits and Use Cases