memory-engineering
Version:
Advanced Memory-Aware Code Context System with claude-flow-inspired architecture, showcasing MongoDB + Voyage AI strategic positioning
158 lines (122 loc) • 4.44 kB
Markdown
# Memory Engineering 2.3.0
Advanced Memory-Aware Code Context System with claude-flow-inspired architecture, showcasing MongoDB + Voyage AI strategic positioning.
## 🎯 Strategic Vision
**Subtly demonstrate the power of MongoDB's Voyage acquisition** through intelligent code indexing and hybrid search capabilities:
- **🏆 Atlas Edition**: MongoDB Atlas + Voyage Code 3 (premium, enterprise-grade)
- **🦙 Community Edition**: MongoDB Community + Ollama (free, local, privacy-focused)
## 🏆 What Makes This Special
### Enterprise-Grade Patterns (Learned from Industry Leaders)
- **AST-Based Code Splitting**: Syntax-aware chunking using Tree-sitter parsers
- **Incremental Synchronization**: Merkle tree-based change detection (10x faster updates)
- **Multi-Provider Architecture**: Voyage Code 3 for Atlas, Ollama for Community
- **Professional Error Handling**: Comprehensive validation and graceful degradation
- **Hybrid Search Ready**: Optimized for MongoDB Atlas text + vector search
### Memory-Focused Innovation
- **Context-Aware Memory**: Not just code search, but intelligent memory management
- **Progressive Search**: Memory → Codebase → Smart hints
- **Project Intelligence**: Understands patterns, decisions, and evolution
## ⚡ Quick Start
### Prerequisites
- **Atlas Edition**: MongoDB Atlas + Voyage AI API key ([get free key](https://voyageai.com))
- **Community Edition**: MongoDB Community + Ollama (no API keys needed!)
### Installation
Add to your MCP client config:
**MongoDB Community** (`.cursor/config.json`):
```json
{
"mcpServers": {
"memory-engineering": {
"command": "npx",
"args": ["-y", "memory-engineering"],
"env": {
"MONGODB_URI": "mongodb://localhost:27017/memory_engineering"
// No API key needed for Community!
}
}
}
}
```
**MongoDB Atlas** (`.cursor/config.json`):
```json
{
"mcpServers": {
"memory-engineering": {
"command": "npx",
"args": ["-y", "memory-engineering"],
"env": {
"MONGODB_URI": "mongodb+srv://user:pass@cluster.mongodb.net/memory",
"VOYAGE_API_KEY": "pa-YOUR-API-KEY-HERE", // Required for Atlas
"MONGODB_PROVIDER": "atlas"
}
}
}
}
```
**Claude Desktop** (`claude_desktop_config.json`):
- macOS: `~/Library/Application Support/Claude/claude_desktop_config.json`
- Windows: `%APPDATA%\Claude\claude_desktop_config.json`
That's it! The server will auto-install via npx.
### MongoDB Setup
**Option 1: MongoDB Community (Local)**
```bash
# macOS
brew install mongodb-community
brew services start mongodb-community
# Ubuntu/Debian
sudo apt-get install mongodb
sudo systemctl start mongodb
```
**Option 2: MongoDB Atlas (Cloud)**
1. Create free cluster at [mongodb.com](https://mongodb.com/cloud/atlas)
2. Get connection string
3. Set `MONGODB_PROVIDER=atlas` and use Atlas URI
## 📋 The 9 Tools
1. **memory_init** - Start EVERY session with this
2. **memory_project_brief** - Project requirements and constraints
3. **memory_product_context** - Product vision and user goals
4. **memory_active_context** - Current work tracking
5. **memory_system_patterns** - Architecture decisions
6. **memory_tech_context** - Technology stack
7. **memory_progress** - Progress and todos
8. **memory_search** - Search all memories
9. **memory_codebase_embed** - Index code for search
## 🔧 MongoDB Providers
### Community Edition (Default)
- Free, runs locally
- **No API key required!**
- Keyword-based search (no AI embeddings)
- Good for small-medium projects
- Perfect for getting started
### Atlas Edition
- Cloud-hosted MongoDB
- Full $rankFusion hybrid search
- Vector search with embeddings
- Scales to millions of memories
## 💡 Usage in Claude
```
1. Always start with: memory_init("/path/to/project")
2. For new projects: Follow the tool guidance
3. Update frequently: memory_active_context
4. Search anytime: memory_search
```
## 🛠️ Development
```bash
# Run in development mode
npm run dev
# Type checking
npm run typecheck
# Run tests
npm test
```
## 📚 Documentation
See `/docs` for detailed documentation:
- `memory-engineering/` - Architecture and planning
- `mongodb/` - Database patterns
- `mcp/` - MCP protocol details
## 🤝 Contributing
1. Read `CLAUDE.md` for AI agent instructions
2. Follow the 9-tool architecture
3. Maintain backward compatibility
4. Test with both MongoDB providers
## 📄 License
MIT