Graphiti builds and queries temporal knowledge graphs that track how facts change over time, preserving provenance to source data. It is a developer framework requiring external provisioning of a graph database and LLM API services, and relies on coding via a Python framework to operate.
Project overview
It offers explicit bi-temporal tracking with automatic fact invalidation, allowing incremental updates without complete graph recomputation—a capability not found in traditional batch-processed RAG approaches.
Project type
MCP · AI Agent · RAG
Use cases
Knowledge Q&A · Automation
Deployment
Refer to project documentation
License
Apache-2.0
Best for
Developers and AI engineers building agent automation or enterprise RAG systems who require temporal knowledge tracking and are prepared to provision external databases and LLM services.
Key capabilities
Autonomously builds context graphs from unstructured and structured data, handling changing relationships while preserving full temporal history.
Provides a Model Context Protocol (MCP) server implementation allowing AI assistants to interact with Graphiti's context graph capabilities through the MCP protocol.
Provides a FastAPI-based API service for interacting with the Graphiti API.
Supports developer-defined entity and edge types upfront via Pydantic models (prescribed), or lets structure emerge from data (learned).
Limitations and risks
Using other LLM services without Structured Output support may result in incorrect output schemas and ingestion failures.
High concurrency ingestion pipelines may cause LLM Provider 429 Rate Limit Errors.
Getting started
Setup is rated medium difficulty because it requires coding via a Python framework to build and query graphs, in addition to provisioning an LLM API and a graph database. The path to first success involves installing the graphiti-core package, setting up Neo4j or FalkorDB via Docker, setting the OpenAI API Key environment variable, and initializing Graphiti to add episodes.
Evidence and sources
GitHub project description: Build Real-Time Knowledge Graphs for AI Agents
README: Traditional RAG approaches often rely on batch processing and static data summarization, making them inefficient for frequently changing data.
README: Graphiti continuously integrates user interactions, structured and unstructured enterprise data, and external information into a coherent, queryable graph.
README: Graphiti is the open-source temporal context graph engine at the core of Zep's context infrastructure for AI agents.