Memento
Provides LLMs with a scalable knowledge graph memory system, enabling semantic retrieval, contextual recall, and temporal awareness.
About
Memento is a knowledge graph memory system designed for Large Language Models (LLMs). It provides LLM clients that support the Model Context Protocol, such as Claude Desktop and Cursor, with resilient, adaptive, and persistent long-term ontological memory. Memento utilizes Neo4j as its storage backend, offering a unified solution for graph storage and vector search, and provides features like semantic search, temporal awareness, and confidence decay to enhance LLM performance and knowledge retention.
Key Features
- Semantic Search: Finds related entities based on meaning using vector embeddings.
- 33 GitHub stars
- Temporal Awareness: Tracks complete history of entities and relations with point-in-time graph retrieval.
- Advanced Metadata: Supports rich metadata for both entities and relations with custom fields.
- Confidence Decay: Relations automatically decay in confidence over time, configurable with a half-life.
- MCP API Tools: Provides tools for entity and relation management, graph operations, and semantic search.
Use Cases
- Providing long-term memory for LLMs in applications like Claude Desktop and Cursor.
- Tracking the evolution of knowledge over time with temporal awareness features.
- Enabling semantic retrieval of information based on meaning rather than keywords.
Sponsored