MCP Memory Service
by rbownes
This is a Rust implementation of the Model Context Protocol (MCP) Memory Service. It provides memory storage and retrieval functionality over stdio for easy integration with MCP clients.
Last updated: N/A
What is MCP Memory Service?
The MCP Memory Service is a Rust implementation of an MCP server that provides memory storage and retrieval functionality. It allows MCP clients to store, retrieve, search, and delete memories using various storage backends and embedding models.
How to use MCP Memory Service?
To use the server, build it using cargo build. Then, run it using cargo run, optionally configuring it with environment variables. Send JSON-RPC requests to its stdin to interact with the memory storage and retrieval tools. You can also register it with an MCP client by adding it to the client's MCP configuration.
Key features of MCP Memory Service
Implements an MCP server for memory storage and retrieval
Communicates over stdio for easy integration with MCP clients
Supports store, retrieve, search, and delete operations
Supports in-memory and ChromaDB storage backends
Supports dummy and ONNX embedding models
Use cases of MCP Memory Service
Storing and retrieving memories for AI models
Semantic search of memories based on content
Tag-based search of memories
Integration with MCP clients like Claude
Testing and development of MCP memory services
FAQ from MCP Memory Service
What are the prerequisites for running the server?
What are the prerequisites for running the server?
Rust and Cargo (1.75.0 or later), Node.js and npm (for testing), and optionally a ChromaDB server (for production use).
How do I configure the server?
How do I configure the server?
The server is configured using environment variables. See the README for a list of available variables and their descriptions.
How do I use the ONNX embedding model?
How do I use the ONNX embedding model?
Export a transformer model to ONNX format, save the tokenizer as a tokenizer.json file, and set the environment variables MCP_MEMORY_EMBEDDING_MODEL to onnx and MCP_MEMORY_EMBEDDING_MODEL_PATH to the path of the ONNX model.
How do I register this server with an MCP client?
How do I register this server with an MCP client?
Add the server's configuration to the MCP client's configuration file, specifying the command, arguments, and environment variables for the server.
What storage backends are supported?
What storage backends are supported?
The server supports in-memory storage (for testing) and ChromaDB storage (for production).