- CLAUDE CODE SOURCE LEAKED :)

⚡ The Complete Anthropic Claude Code CLI Source Code ⚡

---

🚨 FULL SOURCE CODE OF ANTHROPIC'S CLAUDE CODE CLI 🚨

The AI coding assistant that powers millions of developers worldwide

Leaked via npm source map exposure on March 31, 2026

📂 Browse Source • 🔧 Architecture • 🛠️ Tools • ⚡ Commands • 📊 Tech Stack

---

🔥 What Is This?

This repository contains the COMPLETE leaked source code of Anthropic's Claude Code — the revolutionary AI-powered CLI that lets developers interact with Claude AI directly from their terminal.

On March 31, 2026, A security researcher discovered that Claude Code's full TypeScript source was accidentally exposed via a .map file in the npm package:

🐦 "Claude code source code has been leaked via a map file in their npm registry!"

The source map pointed to unobfuscated TypeScript files hosted on Anthropic's R2 storage — making the entire codebase publicly downloadable.

---

⭐ Why This Matters

Feature	Details
🤖 AI Agent Architecture	See exactly how Claude Code's agentic system works
🔧 40+ AI Tools	File editing, web search, code execution, and more
💬 50+ Commands	Full slash command implementation
🔌 MCP Protocol	Model Context Protocol integration
🌐 IDE Bridge	VS Code & JetBrains integration code
🧠 Multi-Agent System	Agent swarm orchestration
🎤 Voice Mode	Voice input implementation
📊 512,000+ Lines	Production-grade TypeScript

---

📂 Directory Structure

📦 source code/
├── 🚀 main.tsx                 # CLI entrypoint
├── 📋 commands.ts              # Command registry
├── 🔧 tools.ts                 # Tool registry  
├── 🤖 QueryEngine.ts           # LLM query engine (46K lines!)
├── 📊 cost-tracker.ts          # Token cost tracking
│
├── 📁 commands/                # 50+ slash commands
├── 📁 tools/                   # 40+ AI agent tools
├── 📁 components/              # 140+ React/Ink UI components
├── 📁 services/                # API, MCP, OAuth, LSP integrations
├── 📁 bridge/                  # IDE communication layer
├── 📁 coordinator/             # Multi-agent orchestration
├── 📁 plugins/                 # Plugin system
├── 📁 skills/                  # Skill system
├── 📁 voice/                   # Voice input
├── 📁 vim/                     # Vim mode
└── 📁 ... and 20+ more directories

---

🏗️ Architecture Overview

🛠️ Tool System

Every capability Claude Code has is implemented as a modular tool

Tool	Description	Category
🖥️ BashTool	Shell command execution	System
📖 FileReadTool	Read files, images, PDFs, notebooks	Files
✏️ FileWriteTool	Create and overwrite files	Files
🔧 FileEditTool	Partial file modification	Files
🔍 GlobTool	File pattern matching	Search
🔎 GrepTool	Content search (ripgrep)	Search
🌐 WebFetchTool	Fetch URL content	Web
🔍 WebSearchTool	Web search	Web
🤖 AgentTool	Spawn sub-agents	Agents
⚡ SkillTool	Execute skills	Skills
🔌 MCPTool	MCP server integration	Protocol
📡 LSPTool	Language Server Protocol	Protocol
📓 NotebookEditTool	Jupyter notebook editing	Files
📋 TaskCreateTool	Task management	Tasks
👥 TeamCreateTool	Team agent management	Agents
🗺️ EnterPlanModeTool	Planning mode	Modes
🌳 EnterWorktreeTool	Git worktree isolation	Git
⏰ CronCreateTool	Scheduled triggers	Automation
🎯 SyntheticOutputTool	Structured output	Output

---

⚡ Command System

50+ slash commands for power users

Command	Description
/commit	📝 Create git commits with AI-generated messages
/review	🔍 AI-powered code review
/compact	📦 Compress conversation context
/mcp	🔌 MCP server management
/config	⚙️ Settings management
/doctor	🏥 Environment diagnostics
/login / /logout	🔐 Authentication
/memory	🧠 Persistent memory management
/skills	⚡ Skill management
/tasks	📋 Task management
/vim	⌨️ Vim mode toggle
/diff	📊 View changes
/cost	💰 Check usage cost
/theme	🎨 Change theme
/pr_comments	💬 View PR comments
/resume	⏮️ Restore previous session
/share	🔗 Share session
/desktop	🖥️ Desktop app handoff

---

🔌 Service Layer

External integrations and core services

Service	Description
🔗 api/	Anthropic API client & bootstrap
🔌 mcp/	Model Context Protocol integration
🔐 oauth/	OAuth 2.0 authentication
📡 lsp/	Language Server Protocol
📊 analytics/	Feature flags (GrowthBook)
🔌 plugins/	Plugin loader
📦 compact/	Context compression
🧠 extractMemories/	Auto memory extraction
👥 teamMemorySync/	Team memory sync

---

🌉 Bridge System

IDE integration layer for VS Code & JetBrains

┌─────────────────┐     ┌─────────────────┐
│   VS Code /     │◄───►│  Claude Code    │
│   JetBrains     │     │     CLI         │
└─────────────────┘     └─────────────────┘
         │                      │
         └──────────┬───────────┘
                    │
         ┌──────────▼──────────┐
         │   Bridge Protocol   │
         │  - JWT Auth         │
         │  - Message Protocol │
         │  - Permission CB    │
         └─────────────────────┘

---

🔒 Permission System

Security-first design with granular permissions

Checks permissions on every tool invocation:

• ✅ default - Ask for approval
• 📋 plan - Plan mode restrictions
• ⚡ auto - Automatic approval
• 🔓 bypassPermissions - Skip checks

---

🚀 Feature Flags

Dead code elimination via Bun's bun:bundle:

import { feature } from 'bun:bundle'

// Inactive code is completely stripped at build time
const voiceCommand = feature('VOICE_MODE')
  ? require('./commands/voice/index.js').default
  : null

Notable flags: PROACTIVE, KAIROS, BRIDGE_MODE, DAEMON, VOICE_MODE, AGENT_TRIGGERS, MONITOR_TOOL

---

📊 Key Files

File	Lines	Description
🧠 QueryEngine.ts	~46,000	Core LLM engine - streaming, tool loops, retry logic
🔧 Tool.ts	~29,000	Tool interfaces & permission models
📋 commands.ts	~25,000	Command registry & execution
🚀 main.tsx	-	CLI entrypoint with parallel prefetch

---

💻 Tech Stack

Bun

TypeScript

React

Zod

Category	Technology
⚡ Runtime	Bun - Ultra-fast JavaScript runtime
📝 Language	TypeScript (strict mode)
🖥️ Terminal UI	React + Ink
🔧 CLI Parsing	Commander.js
✅ Validation	Zod v4
🔍 Code Search	ripgrep
🔌 Protocols	MCP SDK, LSP
🤖 API	Anthropic SDK
📊 Telemetry	OpenTelemetry + gRPC
🚩 Feature Flags	GrowthBook
🔐 Auth	OAuth 2.0, JWT, macOS Keychain

---

🎯 Notable Design Patterns

⚡ Parallel Prefetch

// main.tsx — blazing fast startup
startMdmRawRead()      // MDM settings
startKeychainPrefetch() // Keychain reads
// All run in parallel before heavy imports!

📦 Lazy Loading

Heavy modules loaded only when needed:

• OpenTelemetry
• gRPC
• Analytics
• Feature-gated subsystems

🤖 Agent Swarms

Multi-agent orchestration via coordinator/ — spawn sub-agents with AgentTool, create teams with TeamCreateTool

⚡ Skill System

Reusable workflows in skills/ executed via SkillTool — extensible with custom skills

🔌 Plugin Architecture

Full plugin system in plugins/ for built-in and third-party extensions

---

🔗 Quick Links

Resource	Link
📂 Source Code	source code/
🔧 Tools	source code/tools/
⚡ Commands	source code/commands/
🎨 Components	source code/components/
🌉 Bridge	source code/bridge/

---

⚠️ Disclaimer

This repository is for educational and security research purposes only.

The original Claude Code source is the property of Anthropic. This repository is NOT affiliated with, endorsed by, or maintained by Anthropic.

---

⭐ Star This Repo!

If you found this useful, please give it a star!

---

🔍 Keywords / Tags

claude anthropic claude-code ai artificial-intelligence llm large-language-model cli command-line typescript bun react ink terminal ai-coding ai-assistant code-assistant developer-tools source-code leaked security-research mcp model-context-protocol agent ai-agent multi-agent

---