Claude Code: A Complete Guide to Anthropic's Terminal AI Coding Agent

TL;DR Claude Code is Anthropic's agentic coding tool that runs in the terminal, IDEs, Slack, GitHub, and on the web. Its core extension system has six layers: CLAUDE.md (persistent context), Skills (on-demand workflows), Hooks (deterministic automation), Subagents (isolated delegation), MCP (external tool connections), and Agent Teams (multi-agent collaboration).

#claude-code #anthropic #ai-tools #cli #coding-agent

Table of Contents

Installation
Core Capabilities
The Six-Layer Extension System
Runtime Environments
Update Channels
Pricing
How Claude Code Compares to Other Coding Agents
In-Depth Guides on This Site
Resources
References

🌏 中文版

Claude Code is Anthropic’s agentic coding tool. It understands your codebase, executes everyday tasks through natural language instructions, explains complex code, and handles git workflows — all from your terminal, IDE (VS Code / JetBrains), Slack, GitHub, or the web.

Installation

# Recommended (macOS / Linux)
curl -fsSL https://claude.ai/install.sh | bash

# Windows
winget install Anthropic.ClaudeCode

The npm installation method is now deprecated — use the native installer instead. After installation, sign in with your Claude account and you’re ready to go.

System requirements: macOS 13.0+ / Windows 10+ / major Linux distributions.

Core Capabilities

Capability	Description
Code read/write	Read, edit, and create files; understands entire project structure
Shell execution	Run arbitrary commands in the terminal (build, test, lint, etc.)
Git workflows	commit, branch, PR, merge — all via natural language
Extended Thinking	Enabled by default; reasons through complex problems before acting
Auto Mode	A safe alternative to `--dangerously-skip-permissions` for long-running autonomous tasks
Headless Mode	`claude -p` for programmatic execution; embeds into CI/CD pipelines or scripts

The Six-Layer Extension System

Claude Code’s extension system is what sets it apart from other coding agents. Each layer solves a different problem:

1. CLAUDE.md — Persistent Context

# CLAUDE.md (place in your project root)
This is an Astro + React project
Package manager: pnpm
Testing: vitest
Commit messages in English

This file is loaded automatically at the start of every session, so you never have to repeat project conventions.

2. Skills — On-Demand Workflows

Place .md files in ~/.claude/skills/. No SDK required — just write Markdown.

Reference Skills: provide knowledge (e.g., an API style guide) that Claude references throughout the session
Action Skills: trigger actions (e.g., /deploy) that Claude executes step by step
Supports auto-detection — you don’t always need a slash command to activate them

3. Hooks — Deterministic Automation

Configured in settings.json, hooks bind to lifecycle events (before/after tool calls, session start/end, etc.).

{
  "hooks": {
    "PostToolUse": [{
      "matcher": "edit",
      "command": "npx biome check --fix $CLAUDE_FILE_PATH"
    }]
  }
}

Hooks are guaranteed to run regardless of model behavior. Perfect for lint, format, security checks, and any operation that must run every single time.

4. Subagents — Isolated Delegation

The main agent splits tasks and spawns subagents to run them in parallel. Each subagent has its own isolated context window and does not inherit conversation history. Only the final result returns to the main context, keeping token usage lean.

5. MCP — External Tool Connections

The Model Context Protocol lets Claude connect to your toolchain: databases, GitHub, Sentry, Figma, Slack — 3,000+ integrations. Each MCP server becomes an immediately available tool.

6. Agent Teams — Multi-Agent Collaboration

Released in February 2026 alongside Opus 4.6. Multiple independent Claude sessions can message each other, divide work, and operate in parallel. Unlike subagents (a hierarchical delegation relationship), Agent Teams are peers collaborating as equals.

Extension System Summary

Layer	Problem it solves	When it loads
CLAUDE.md	Project conventions, persistent instructions	Auto-loaded at session start
Skills	Reusable workflows	On-demand or auto-detected
Hooks	Must-run automation	Deterministically on event trigger
Subagents	Isolation and parallelism	Spawned by the main agent as needed
MCP	External system connections	Always available after configuration
Agent Teams	Multi-agent collaboration	Manually initiated
Plugins	Package and distribute all of the above	Loaded per their own rules after install

Runtime Environments

Environment	Description
Terminal	Native CLI, the core experience
VS Code	Native extension with visual diff
JetBrains	Native extension
Web	claude.ai/code
Slack	Launch dev tasks directly in team conversations
GitHub	Trigger with @claude in PRs or Issues

Update Channels

Channel	Description
`latest`	Default; receives new features immediately
`stable`	Approximately one week delayed; skips releases with major regressions

Pricing

Requires a Claude Pro or Max subscription (or API access). Using Claude Sonnet 4.6 as an example:

	Price
Input	$3.00 / 1M tokens
Output	$15.00 / 1M tokens

How Claude Code Compares to Other Coding Agents

	Claude Code	Codex CLI	Gemini CLI	OpenCode	Pi
Vendor	Anthropic	OpenAI	Google	SST (open source)	badlogic (open source)
Extension layers	6 layers	Skills + MCP	MCP	Dual Agent	Extension
Open source	❌	✅	✅ (Apache 2.0)	✅	✅
Free tier	❌ (subscription required)	Partially free	1,000 requests/day	Completely free	Completely free
Core strength	Most complete extension ecosystem	OpenAI model integration	Free + 1M context	75+ model flexibility	Minimal + low token usage
IDE integration	VS Code + JetBrains	VS Code	VS Code	TUI	TUI

Claude Code’s core advantage is the depth of customization enabled by its six-layer extension system — from individual developers to enterprise teams, you can layer in exactly the automation and integrations you need. The trade-offs: it’s not open source and requires a paid subscription.

In-Depth Guides on This Site

This site has 20+ articles dedicated to Claude Code, indexed by topic below:

Extension system:

Configuration & environment:

Integrations & automation:

Miscellaneous: