Skip to main content
Claude Code is a terminal-native AI coding assistant built by Anthropic. It brings Claude directly into your development workflow, letting you edit files, execute shell commands, search codebases, manage git, and coordinate multi-agent tasks — all from your terminal.

Quick Start

Get Claude Code running in under 5 minutes

Installation

Install Claude Code on your system

Tools Reference

Complete catalog of all ~40 agent tools

Commands Reference

All ~85 slash commands with descriptions

What you can do with Claude Code

Edit & Search Code

Read, write, and search files using intelligent tools with permission controls

Run Shell Commands

Execute bash commands, run tests, and manage your development environment

Git Workflows

Commit, review PRs, create branches, and address review comments

Multi-Agent Tasks

Spawn sub-agents and coordinate parallel work with team orchestration

MCP Integration

Connect to Model Context Protocol servers for extended capabilities

IDE Integration

Bridge to VS Code and JetBrains for a seamless IDE experience

Explore the architecture

1

Understand the core pipeline

Learn how user input flows through the CLI parser, Query Engine, and tool execution loop in the Architecture Overview.
2

Learn the tool system

Every capability Claude can invoke is a self-contained tool with its own schema, permissions, and UI. See the Tool System.
3

Master slash commands

Invoke powerful workflows with /commit, /review, /mcp, /memory, and ~80 more. See the Commands Reference.
4

Explore with the MCP server

Use the claude-code-explorer-mcp package to interactively browse the source code from any MCP-compatible client. See MCP Server Setup.
This documentation covers the leaked Claude Code source (@anthropic-ai/claude-code v0.0.0-leaked). The source is read-only reference material. All original code is the property of Anthropic.

Build docs developers (and LLMs) love