Offers community support and discussion channels for users of the Memory Engineering MCP server
Integrates with Git diff to detect code changes for incremental embedding updates and sync operations
Hosts the source code repository and integrates with Git diff functionality for incremental code synchronization
Provides persistent memory storage and semantic code search capabilities using MongoDB Atlas Vector Search with 1024-dimensional embeddings and hybrid search functionality
Distributes the Memory Engineering MCP package for easy installation and updates
Hosts the interactive demo and landing page showcasing the Memory Engineering MCP capabilities
🧠 Memory Engineering MCP
📦 NPM Package | 📖 Documentation
Persistent memory and semantic code understanding for AI assistants. Built on MongoDB Atlas Vector Search and Voyage AI embeddings.
🚀 Powered by voyage-code-3: The Code Understanding Model
voyage-code-3 is Voyage AI's specialized model that understands code like a senior developer:
Syntax-Aware: Distinguishes between
UserService.create()
andUser.create()
- knows one is a service method, the other is a model methodCross-Language: Recognizes that Python's
async def
, JavaScript'sasync function
, and Go'sgo func()
all represent asynchronous patternsSemantic Relationships: Understands that
hash_password()
relates toverify_password()
,salt
,bcrypt
, and security patternsArchitecture Understanding: Knows that controllers → services → repositories → models represents a layered architecture
Real-World Impact
✨ See It In Action
🔥 The Game Changer: Code Embeddings
This is what makes Memory Engineering different from everything else:
Revolutionary Code Chunking
Smart Semantic Boundaries: Tracks braces, parentheses, and indentation to capture COMPLETE functions (up to 200 lines) and classes (up to 300 lines)
Context-Aware: Every chunk includes its imports, dependencies, and surrounding context
Pattern Detection: Automatically identifies 27 code patterns (error-handling, async, authentication, etc.)
Why This Matters
Semantic Code Search That Actually Works
🧠 The 7 Core Memories
Inspired by Cline, but enhanced with MongoDB persistence:
activeContext - What you're doing RIGHT NOW (update every 3-5 min!)
projectbrief - Core requirements and features
systemPatterns - Architecture decisions and patterns
techContext - Stack, dependencies, constraints
progress - What's done, in-progress, and next
productContext - Why this exists, user needs
codebaseMap - File structure with embedded statistics
💪 Technical Architecture
MongoDB Atlas Integration
Vector Search: 1024-dimensional embeddings with cosine similarity
Hybrid Search: Combines semantic + keyword search
Auto-indexing: Manages compound, text, and vector indexes automatically
Connection pooling: 5-100 connections with retry logic
Voyage AI Integration - Powered by voyage-code-3
Why voyage-code-3 Changes Everything
Purpose-Built for Code: Unlike general models, voyage-code-3 understands syntax, patterns, and programming concepts
1024 Dimensions: Optimal balance between accuracy and performance
Code-Aware Embeddings: Knows the difference between
class Auth
andauthenticate()
semanticallyLanguage Agnostic: Works across JavaScript, TypeScript, Python, Go, Rust, and more
Technical Capabilities
Advanced Features
Reranking with rerank-2.5-lite: Re-orders results by true relevance (8% accuracy boost)
32K Context Window: 8x larger than before for understanding long files
Semantic Expansion:
auth
automatically searches for authentication, JWT, tokens, sessionsPattern Recognition: Identifies 27 architectural patterns automatically
Smart Batching: Processes 100 chunks simultaneously for speed
Code Intelligence
⚡ Quick Start
Installation
Configure Cursor/.cursor/mcp.json
First Run
🔬 voyage-code-3 vs Other Embedding Models
Technical Comparison
Aspect | voyage-code-3 | General Models (text-embedding-3) |
Code Syntax | Understands AST-like structures | Treats code as text |
Variable Names | Knows
≈
≈
| Sees as different tokens |
Design Patterns | Recognizes Singleton, Factory, Repository | No pattern awareness |
Error Handling | Links try/catch ↔ .catch() ↔ error boundaries | Misses connections |
Import Relationships | Tracks dependency graphs | Ignores imports |
Context Window | 32K tokens (full files) | 8K tokens typical |
Benchmark Results
🎯 Real Power Examples
Finding Code You Forgot Exists
Understanding Patterns Across Codebase
Tracking Decisions
📊 Performance & Technical Metrics
Speed & Scale
Code sync: 100 files/batch with voyage-code-3 embeddings
Search latency: <500ms for 100k chunks with reranking
Memory operations: <100ms read/write
Reranking: +50ms for 23% better accuracy
voyage-code-3 Specifications
Embedding dimensions: 1024 (optimal for code)
Context window: 32K tokens (8x improvement)
Languages supported: 50+ programming languages
Pattern detection: 27 architectural patterns
Accuracy boost: 15% over general models
Code Understanding Capabilities
🎯 How voyage-code-3 Helps Different Tasks
Code Review & Refactoring
Debugging
Learning a New Codebase
Security Audit
🔧 Advanced Features
Smart Pattern Aliasing (Enhanced by voyage-code-3)
The system understands natural language variations:
"auth" → searches: authentication, authorization, login, JWT, token, session, OAuth
"db" → searches: database, MongoDB, schema, model, collection, repository, ORM
"error handling" → searches: try-catch, exception, error-handler, .catch(), Promise.reject
Incremental Sync
Only changed files are re-embedded:
Context Preservation
Every code chunk maintains context:
🛠️ Tools Reference
Tool | Purpose | Key Features |
| Initialize project | Scans code, creates memories, generates embeddings |
| Read/Update memories | Unified interface for all 7 memories |
| Semantic search | Memory + code search with patterns |
| Sync code embeddings | Smart chunking, incremental updates |
| Health & diagnostics | Status, environment, doctor mode |
🚀 Why This Works
Complete Code Understanding: Unlike other systems that break functions arbitrarily, we preserve semantic units
Rich Embeddings: Each chunk has context, patterns, and relationships
Behavioral Prompting: Dramatic prompts ensure AI assistants take memory seriously
MongoDB Scale: Handles millions of chunks with millisecond queries
Voyage AI Quality: State-of-the-art embeddings optimized for code
📦 Latest Updates
v13.4.0 (January 2025)
Enhanced memory quality with structured templates
Improved pattern detection in code embeddings (now 27 patterns)
Better validation for consistent memory creation
All improvements are backwards compatible
v13.3.2
Consolidated tools for simpler interface
Performance optimizations
📄 License
MIT - See LICENSE file
🔗 Links
Built with Model Context Protocol (MCP) by Anthropic
This server cannot be installed
remote-capable server
The server can be hosted and run remotely because it primarily relies on remote services or has no dependency on the local environment.
Provides persistent memory and semantic code understanding for AI assistants using MongoDB Atlas Vector Search. Enables intelligent code search, memory management, and pattern detection across codebases with complete semantic context preservation.
- 🚀 Powered by voyage-code-3: The Code Understanding Model
- ✨ See It In Action
- 🔥 The Game Changer: Code Embeddings
- 🧠 The 7 Core Memories
- 💪 Technical Architecture
- ⚡ Quick Start
- 🔬 voyage-code-3 vs Other Embedding Models
- 🎯 Real Power Examples
- 📊 Performance & Technical Metrics
- 🎯 How voyage-code-3 Helps Different Tasks
- 🔧 Advanced Features
- 🛠️ Tools Reference
- 🚀 Why This Works
- 📦 Latest Updates
- 📄 License
- 🔗 Links
Related MCP Servers
- AsecurityAlicenseAqualityProvides a structured documentation system for context preservation in AI assistant environments, helping users create and manage memory banks for their projects.Last updated -370MIT License
MongoDB MCP Serverofficial
AsecurityAlicenseAqualityA Model Context Protocol server that enables AI assistants to interact with MongoDB Atlas resources through natural language, supporting database operations and Atlas management functions.Last updated -2111,872638Apache 2.0- -securityAlicense-qualityA lightweight server that provides persistent memory and context management for AI assistants using local vector storage and database, enabling efficient storage and retrieval of contextual information through semantic search and indexed retrieval.Last updated -1MIT License
- -securityFlicense-qualityProvides a standardized interface for AI assistants to interact with MongoDB databases, enabling CRUD operations, schema analysis, and database management through the Model Context Protocol.Last updated -1