close
Skip to content

Yusufkuscu/Claude-code-leaksource

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

🔓 Claude Code - Leaked Source Code Archive

Leaked Python Educational Archive

📦 Complete archive of the leaked Claude Code source code from Anthropic (March 31, 2026)


🤔 What is Claude Code?

Claude Code is Anthropic's revolutionary AI-powered coding assistant that runs directly in your terminal. It's designed to help developers with:

  • 🔧 Code Generation - Write, edit, and refactor code using natural language
  • 📁 File Operations - Create, read, update, and delete files autonomously
  • 🐛 Debugging - Find and fix bugs with AI-powered analysis
  • 🧪 Testing - Generate and run tests automatically
  • 🌐 Web Fetching - Retrieve information from the web
  • 📊 Project Understanding - Analyze and navigate complex codebases
  • 🔄 Git Integration - Manage version control seamlessly

Claude Code became one of the most popular AI coding tools, known for its ability to perform complex, multi-step tasks autonomously while maintaining context across long sessions.


💥 The Leak

On March 31, 2026, Claude Code's entire source code was accidentally exposed. This repository preserves that codebase for educational and research purposes, allowing developers to study:

  • How enterprise-grade AI agents are architected
  • The internal workings of tool systems and command processing
  • Real-world patterns for context management and session handling
  • Production-level plugin and hook architectures

🏗️ Repository Structure

📦 Claude-code-leaksource
├── 📂 src/                         # Main source code
│   ├── 📂 assistant/               # AI assistant core logic
│   ├── 📂 cli/                     # Command-line interface
│   ├── 📂 commands/                # Built-in command implementations
│   ├── 📂 tools/                   # Tool system (file ops, web fetch, etc.)
│   ├── 📂 hooks/                   # Pre/post execution hooks
│   ├── 📂 plugins/                 # Plugin architecture
│   ├── 📂 services/                # Service layer
│   ├── 📂 skills/                  # Skill modules
│   ├── 📂 context/                 # Context & session management
│   ├── 📄 runtime.py               # Core execution engine
│   ├── 📄 commands.py              # Command definitions
│   ├── 📄 tools.py                 # Tool implementations
│   ├── 📄 context.py               # Context handling
│   ├── 📄 main.py                  # Entry point
│   └── ...
├── 📂 tests/                       # Test suite
├── 📂 assets/                      # Images and media
└── 📄 README.md

🔍 Key Components Explained

🛠️ Tools System

The heart of Claude Code's capabilities. Each tool is a modular unit that can:

  • Read/write files
  • Execute shell commands
  • Search codebases with grep/glob
  • Fetch web content
  • Interact with Git

⚡ Commands

Built-in slash commands like:

  • /help - Display available commands
  • /clear - Clear conversation history
  • /compact - Compress context
  • /init - Initialize project settings

🪝 Hooks

Extensibility points for custom behavior:

  • Pre-execution hooks
  • Post-execution hooks
  • Error handlers
  • Output formatters

🔌 Plugins

Modular extensions that add new capabilities without modifying core code.

🧠 Context Management

Sophisticated system for maintaining:

  • Conversation history
  • File state tracking
  • Session persistence
  • Memory optimization

🚀 Quick Start

# Clone the repository
git clone https://github.com/Yusufkuscu/Claude-code-leaksource.git
cd Claude-code-leaksource

# Explore the manifest
python -m src.main manifest

# View subsystems
python -m src.main subsystems --limit 20

# Run the test suite
python -m unittest discover -s tests -v

📚 What You Can Learn From This

Topic What You'll Discover
Agent Architecture How to structure a production AI agent
Tool Calling Patterns for implementing AI tool systems
Context Windows Managing long conversations efficiently
Plugin Systems Building extensible architectures
CLI Design Creating powerful command-line interfaces
Error Handling Graceful failure and recovery patterns
Testing AI Agents Strategies for testing non-deterministic systems

🎨 Check Out My Project: BLYTH AI

Blyth.ai

🌟 The Next-Generation AI Creative Workspace Platform

What is BLYTH AI?

BLYTH AI is a modern AI Creative Workspace platform — a production-grade SaaS that lets you build visual, video, and text generation workflows on a node-based infinite canvas.

💡 Think of it as "Lego blocks for AI" — build complex AI pipelines without writing a single line of code!

Example Workflow:

📝 Text Input → 🤖 LLM (Prompt Enhancement) → 🎨 Image Generation → 🎨 Video Generation → 📤 Output

✨ Core Features

Feature Description
🎯 Infinite Canvas ReactFlow-powered workspace with unlimited space
🧊 Drag & Drop Blocks 13+ block types for every AI task
⚡ Auto-Save Never lose work — saves every 5 seconds
🔄 Undo/Redo Full history support
🎨 Theme Customization Background patterns, colors, animations
📷 PNG Export Download your canvas as an image
⌨️ Keyboard Shortcuts Ctrl+Z, Ctrl+Y, Delete, and more

🧊 Block Types (13 Different Types)

Generation Blocks:

  • 🖼️ Image Block — Text-to-image, image-to-image, inpaint, outpaint, upscale
  • 🎬 Video Block — Text-to-video, image-to-video, video-to-video
  • 📝 Text Block — LLM text generation, prompt splitter

Utility Blocks:

  • ✂️ SAM Mask — Florence-SAM3 automatic segmentation
  • 🎨 Inpaint — Masked area editing
  • 🖌️ Style Ref — IP-Adapter style reference
  • 🔀 Compositor — Multi-image compositing
  • 📚 Multi-Ref — Flux 2 Pro Edit with 10 reference image blending

Input/Control Blocks:

  • 📝 Text Input — Static prompt input
  • 📤 Upload — Image/video upload
  • 📥 Output — View & download results
  • 📌 Note — Add notes to your canvas

🤖 Supported AI Models (30+)

Image Generation (12 models):

Provider Models
Black Forest Labs FLUX Pro, Kontext Max, Schnell, Dev
Stability AI Stable Diffusion 3.5, SDXL Turbo
OpenAI DALL-E 3
Recraft Recraft V3
Ideogram Ideogram V2
Midjourney Midjourney V6
Playground Playground V3
Leonardo AI Leonardo Phoenix

🎯 Who Is It For?

  • 🎨 Designers — No coding required
  • 📱 Content Creators — Build creative pipelines
  • 📈 Marketing Teams — Batch content generation
  • 🏢 Agencies — Multiplayer workflows

Join Waitlist

Be among the first to experience the future of AI-powered creative workflows!


⚠️ Legal Disclaimer

IMPORTANT: This repository is provided strictly for educational and research purposes.

  • 📌 This is NOT affiliated with, endorsed by, or maintained by Anthropic
  • 📌 No ownership is claimed over the original source material
  • 📌 No warranty or support is provided
  • 📌 Users are solely responsible for their use of this material
  • 📌 This archive exists to advance understanding of AI agent systems

🤝 Contributing

This is an archive repository. Contributions for documentation improvements, analysis, and educational materials are welcome.


📜 License

This repository contains leaked proprietary code. Use at your own discretion and risk. The original code is property of Anthropic.


📚 Archived for educational and research purposes | March 2026

🎨 Build AI Creative Workflows → blyth.ai

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages