Memorized
Provides a high-performance, local-first memory server in Rust, equipping AI agents with an advanced hybrid memory system combining knowledge graphs, vector embeddings, and full-text search.
概要
Memorized delivers a high-performance, local-first memory server built in Rust, specifically designed to equip AI agents with an advanced and versatile hybrid memory system. It seamlessly integrates a knowledge graph, vector embeddings, and full-text search capabilities with a novel documentary memory, allowing agents to accurately recall exact documents alongside rich semantic relationships. This system supports efficient document ingestion, memory storage and retrieval, and automatic memory consolidation, ensuring robust and adaptive agent behavior while providing monitoring tools for system health and performance.
主な機能
- Advanced hybrid memory system combining knowledge graph, vector embeddings, and full-text search
- 0 GitHub stars
- APIs for tool discovery, document ingestion, memory management (add, search, consolidate)
- Real-time health monitoring with performance (p95Ms) and resource (rssMb) thresholds
- Novel documentary memory for accurate recall of exact documents alongside semantic relationships
- High-performance, local-first architecture built in Rust
ユースケース
- Empowering AI agents with sophisticated long-term and short-term memory capabilities
- Storing and retrieving complex information, including documents and semantic memories, for agent decision-making
- Enabling agents to adapt behavior based on real-time system health and performance metrics
