Provides a high-performance knowledge graph memory server with streamable HTTP and legacy SSE transports for AI agents.
Sponsored
The Memory server is a high-performance, Rust-based implementation of the Model Context Protocol (MCP) Knowledge Graph Memory Server. It's engineered for speed, security, and seamless integration with various AI agents, offering blazing-fast operations and hierarchical memory scopes (Global, Project, Private Agent) to manage diverse contextual data. Supporting modern MCP v2025-11-25 and legacy protocols, it includes production-ready features like API key authentication, background session cleanup, and atomic JSON storage, making it ideal for robust AI applications.
Key Features
01Hierarchical memory scopes: Global Shared, Project Shared, Private Agent
02Full support for MCP v2025-11-25 (Streamable SSE) and legacy protocols
030 GitHub stars
04Blazing fast performance with Rust, Axum, and dashmap (~5000+ RPS)
05Easy deployment via automated setup script for systemd --user service
06Production-ready with Bearer Token authentication and background session cleanup
Use Cases
01Storing and managing hierarchical memory for AI agents across different scopes
02Providing a secure and reliable knowledge base for AI agent context and shared resources
03Integrating a high-performance memory backend into AI agent architectures via HTTP/SSE
