CodeMemory - AI-Powered Code Understanding with Persistent Memory

CodeMemory is a VSCode extension that gives your code editor a long-term memory. It understands your entire codebase, remembers past interactions, and provides intelligent assistance that gets better over time.

🚀 Features

1. Persistent Code Memory

• Indexes your entire codebase into a local vector database
• Maintains context across sessions
• Learns from your interactions

2. Intelligent Code Understanding

• Semantic search across your codebase
• Understands relationships between functions, classes, and modules
• Tracks dependencies and usage patterns

3. Context-Aware Assistance

• Answer questions about your codebase
• Explain complex code sections
• Find related code automatically
• Suggest improvements based on your patterns

4. Privacy-First Design

• All processing happens locally by default
• Your code never leaves your machine without explicit permission
• Optional cloud features with full control

📦 Installation

1. Install from VSCode Marketplace: Search for "CodeMemory"
2. Or install manually:

   code --install-extension codememory-0.0.1.vsix
   

🛠️ Setup

Quick Start

1. Open a project in VSCode
2. CodeMemory will automatically start indexing (if auto-index is enabled)
3. Configure your preferred LLM provider in settings

Configuration

Open VSCode settings and search for "CodeMemory":

{
  // LLM Provider settings
  "codememory.llmProvider": "openai", // or "anthropic"
  "codememory.apiKey": "your-api-key-here",
  
  // Indexing settings
  "codememory.autoIndex": true,
  "codememory.indexFileTypes": [
    ".ts", ".js", ".py", ".java", ".cpp", ".go", ".rs"
  ],
  
  // Storage settings
  "codememory.localStoragePath": "", // defaults to workspace/.codememory
}

💡 Usage

Commands

Access these commands via Command Palette (Cmd/Ctrl + Shift + P):

• CodeMemory: Index Current Workspace - Manually trigger indexing
• CodeMemory: Explain Selected Code - Get AI explanation of selected code
• CodeMemory: Find Related Code - Find similar or related code sections
• CodeMemory: Ask Question About Codebase - Ask any question about your code
• CodeMemory: Show Memory Status - View indexing statistics

Example Queries

• "How does the authentication system work?"
• "Show me all API endpoints"
• "What does the UserService class do?"
• "Find all database queries"
• "How is error handling implemented?"

🏗️ Architecture

Core Components

1. Code Parser: Language-specific parsers extract semantic information
2. Vector Store: ChromaDB stores embeddings locally
3. Context Manager: Manages retrieval and memory
4. LLM Interface: Integrates with OpenAI/Anthropic

How It Works

1. Indexing Phase:

   • Parses code files into semantic chunks
   • Generates embeddings for each chunk
   • Stores in local vector database

2. Query Phase:

   • Converts queries to embeddings
   • Retrieves relevant code chunks
   • Provides context to LLM for response

3. Learning Phase:

   • Tracks successful queries
   • Updates relevance scores
   • Improves over time

🔧 Development

Building from Source

# Clone repository
git clone https://github.com/vikasj2in/codememory-vscode
cd codememory-vscode

# Install dependencies
npm install

# Compile
npm run compile

# Run tests
npm test

# Package extension
vsce package

Project Structure

codememory-vscode/
├── src/
│   ├── core/           # Core extension logic
│   ├── indexer/        # Code parsing and indexing
│   ├── vectorstore/    # Vector database integration
│   ├── context/        # Context management
│   ├── llm/            # LLM integrations
│   ├── ui/             # UI components
│   └── extension.ts    # Main entry point
├── test/               # Test suite
├── resources/          # Icons and assets
└── package.json        # Extension manifest

🤝 Contributing

We welcome contributions! Please see CONTRIBUTING.md for guidelines.

Areas for Contribution

• Language parser improvements
• Additional LLM provider integrations
• Performance optimizations
• UI/UX enhancements
• Documentation improvements

📝 Roadmap

Phase 1 (Current)

• ✅ Basic indexing and retrieval
• ✅ LLM integration
• ✅ Simple UI

Phase 2

• 🔄 Incremental indexing
• 🔄 Multi-repository support
• 🔄 Team knowledge sharing

Phase 3

• 📋 Advanced architectural understanding
• 📋 Automated refactoring suggestions
• 📋 Integration with CI/CD

🐛 Known Issues

• Large repositories (>10k files) may take time to index initially
• Some exotic language constructs might not be parsed correctly
• Memory usage scales with codebase size

📄 License

MIT License - see LICENSE file for details.

🙏 Acknowledgments

• Built with LangChain
• Vector storage by ChromaDB
• Embeddings by Transformers.js

📞 Support

• Issues: GitHub Issues
• Discussions: GitHub Discussions
• Email: codememory-support@cloudivian.com

Made with ❤️ by developers, for developers.