Graphiti

概要

Graphiti MCP Pro significantly extends the foundational Graphiti framework, a powerful tool for building and querying temporally-aware knowledge graphs essential for AI agents operating in dynamic environments. Going beyond traditional RAG methods, Graphiti continuously integrates diverse data into a coherent, queryable graph, supporting incremental updates and precise historical queries without recomputation. This enhanced version specifically improves Graphiti's Memory Repository Service (MCP) by introducing asynchronous parallel processing for adding memories, new task management tools, and a unified configuration system. It also boasts broader AI model compatibility, supporting various OpenAI API-compatible models including DeepSeek, Qwen, and local LLMs via Ollama or vLLM, through flexible, segmented model configurations for large language models, small language models, and embedders. A comprehensive management platform with a Web UI provides robust service control, real-time configuration updates, usage monitoring, and log viewing, offering superior user experience and observability.

主な機能

• Granular, Separated AI Model Configurations for flexible deployment of LLM, Small LLM, and Embedder models
• 5 GitHub stars
• Asynchronous Parallel Memory Processing for efficient data ingestion
• Broad AI Model Compatibility supporting various OpenAI API-compatible LLMs and embedding models
• Comprehensive Web-based Management UI for service control, configuration, monitoring, and logging
• Dedicated Task Management Tools for monitoring and controlling add_memory operations

ユースケース

• Orchestrating diverse AI models for entity extraction and data vectorization within knowledge graph applications.
• Developing and managing advanced knowledge graphs as memory for AI agents in dynamic environments.
• Monitoring and configuring Graphiti memory services through a centralized web interface for enhanced observability.