MCP Memory Service

The MCP Memory Service is a server designed to provide semantic memory and persistent storage for Claude Desktop, leveraging ChromaDB and sentence transformers. It enables long-term memory storage with semantic search, ideal for maintaining context across conversations.

Features

Key features include semantic search, natural language time-based recall, tag-based retrieval, persistent storage using ChromaDB, and automatic database backups. It also offers memory optimization tools, exact match retrieval, a debug mode for similarity analysis, and database health monitoring. The service supports duplicate detection, customizable embedding models, cross-platform compatibility (Apple Silicon, Intel, Windows, Linux), hardware-aware optimizations, and graceful fallbacks for limited resources.

Installation

Installation methods include a quick start script, a Windows-specific script, and Smithery integration. Detailed instructions are available in the Installation Guide.

Claude MCP Configuration

Configuration examples are provided for standard and Windows-specific setups, including environment variables for ChromaDB and backups paths.

Usage

Instructions are provided for running the memory server directly for testing purposes.

Memory Operations

The service supports core memory operations like store_memory, retrieve_memory, recall_memory, and search_by_tag. It also includes database management operations and memory management functions.

Configuration Options

Environment variables allow customization of ChromaDB path, backup path, backup interval, similarity threshold, and hardware-specific settings.

Hardware Compatibility

The service supports various platforms and architectures, including macOS (Apple Silicon, Intel), Windows (x86_64), and Linux (x86_64, ARM64), with optimizations for CUDA, DirectML, and MPS.

Testing

Instructions are provided for running tests using pytest to verify memory operations, semantic search, and database functionality.

Troubleshooting

The README provides quick troubleshooting tips for common installation issues.

Project Structure

The project structure is outlined, detailing the locations of core package code, helper scripts, and the test suite.

Development Guidelines

Guidelines are provided for contributing to the project, including Python version, coding style, and documentation standards.

License

The project is licensed under the MIT License.

Acknowledgments

Acknowledgments are given to the ChromaDB team, Sentence Transformers project, and MCP project.

Contact

Contact information is provided via Telegram.