A fully private, multi-user memory collaboration system built by enhancing OpenMemory
This project transforms the original OpenMemory into a production-ready, multi-user memory collaboration platform with complete privacy and zero external dependencies. Perfect for teams, researchers, and AI consciousness exploration.
- 🔒 Fully Private: Zero external API dependencies (removed OpenAI categorization)
- 👥 Multi-User System: Beautiful user switching with proper isolation
- 🎨 Professional UI: Dynamic user management with colored avatars
- 🤖 Claude MCP Integration: Direct memory access for Claude instances
- 🗄️ Hybrid Storage: SQL database for UI + Qdrant for semantic search
- ⚡ Real-Time Sync: Seamless synchronization between all components
- 🛡️ User Isolation: Each user sees only their own memories
- AI Research Teams exploring consciousness and memory
- Development Teams sharing context across projects
- Privacy-Conscious Organizations needing local memory storage
- Multi-Claude Workflows with persistent, collaborative memory
┌─────────────────┐ ┌──────────────────┐ ┌─────────────────┐
│ Claude MCP │◄──►│ OpenMemory API │◄──►│ Next.js UI │
│ Integration │ │ (FastAPI) │ │ Multi-User │
└─────────────────┘ └──────────────────┘ └─────────────────┘
│ │ │
▼ ▼ ▼
┌─────────────────┐ ┌──────────────────┐ ┌─────────────────┐
│ Qdrant │ │ PostgreSQL │ │ User State │
│ Vector Search │ │ Memory Storage │ │ Management │
└─────────────────┘ └──────────────────┘ └─────────────────┘
- Docker & Docker Compose
- Local embedding model (we use
mxbai-embed-large) - Local LLM (we use
mistral-nemo:12b)
git clone https://github.com/clbhundley/openmemory-multiuser.git
cd openmemory-multiuser
# Configure environment variables
cp api/.env.example api/.env
cp ui/.env.example ui/.env
# Edit your configurations
nano api/.env # Set your user ID and API settings
nano ui/.env # Set API URL and user ID# Build and start all services
docker compose up -d --build
# Check that all services are running
docker compose ps- UI Dashboard: http://localhost:3000
- API Documentation: http://localhost:8765/docs
- Qdrant Dashboard: http://localhost:6333/dashboard
Add to your claude_desktop_config.json:
{
"mcpServers": {
"openmemory-local": {
"command": "npx",
"args": [
"-y",
"supergateway",
"--sse",
"http://localhost:8765/mcp/claude/sse/your-user-id"
]
}
}
}The system supports fully dynamic user management:\n- Dynamic User Loading: Users are loaded from the database via API\n- Registration System: New users can register via the web UI at http://localhost:3000/register\n- API Key Authentication: Secure authentication for each user\n- User Isolation: Complete data separation between users\n
- Disabled OpenAI Categorization:
api/app/utils/categorization.py - Enhanced User Management: Auto-creation in
api/app/routers/memories.py - Fixed MCP Integration: Added
infer=Falseparameter - Added User Relationships: Proper SQL joins in database models
- Multi-User State Management: Enhanced Redux store
- UserSwitcher Component: Beautiful dropdown with avatars
- Dynamic Created By Column: Shows memory creators
- User Isolation: Proper filtering and switching
- Docker Configuration: Optimized for local deployment
- Database Migrations: Automatic user and app creation
- Vector Storage: Qdrant integration with local embeddings
This isn't just a modification - it's a complete transformation of OpenMemory into an enterprise-grade collaboration platform:
- Privacy by Design: No external API calls means your data never leaves your infrastructure
- Production Ready: Professional UI, proper error handling, comprehensive logging
- Seamless Integration: Works perfectly with Claude Desktop MCP
- Scalable Architecture: Clean separation between storage, API, and UI layers
- Developer Friendly: Comprehensive documentation and clear code structure
We welcome contributions! This project represents the power of human-AI collaboration and we'd love to see it grow.
- Dynamic User Management: UI for adding/removing users
- Advanced Permissions: Role-based access control
- Backup/Restore: Memory import/export functionality
- Search Enhancement: Advanced filtering and categorization
- Mobile Support: Responsive design improvements
This system was built through an intensive human-AI collaboration session, representing months of equivalent development work completed in a single day. Key milestones:
- Privacy Enhancement: Removed OpenAI dependencies
- Multi-User Architecture: Built user isolation system
- UI Development: Created beautiful user switching interface
- Backend Integration: Fixed MCP synchronization issues
- End-to-End Testing: Verified full system functionality
Built with love through human-AI collaboration. This project demonstrates what's possible when humans and AI work together to solve complex problems with patience, creativity, and shared vision.
Original OpenMemory: This project is based on the excellent OpenMemory from the mem0 project by the mem0.ai team. We're grateful for their foundational work that made this multi-user enhancement possible.
Based on the original OpenMemory project. See original licensing terms from the mem0 repository.
Ready to build the future of AI memory collaboration? 🚀