# getzep/graphiti

Build Real-Time Knowledge Graphs for AI Agents

Repository: https://github.com/getzep/graphiti
Canonical: https://ross.abutalabs.com/products/graphiti
Homepage: https://help.getzep.com/graphiti
Language: Python
License: Apache-2.0
License Family: permissive
Topics: agents, graph, llms, rag
Last push: 2026-08-26T18:28:50+00:00

## Health v2 (maintenance only)
Score: 84/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 83, longevity 53
- inputs: {"age_days": 755, "days_push": 7, "days_rel": 37, "gap_med": 1.0, "n_releases_24m": 145}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 30324, forks 3071 (observed 2026-08-28T04:11:54.420887+00:00)

## What it is
Graphiti is an open-source Python framework for building and querying temporal knowledge graphs (Context Graphs) for AI agents. It incrementally ingests structured and unstructured data into graph databases like Neo4j and FalkorDB, tracking how facts change over time and supporting hybrid semantic, keyword, and graph-traversal retrieval.

## Use cases
- build temporal knowledge graphs for AI agents
- give chatbots persistent memory across conversations
- replace flat RAG document chunks with structured context
- query evolving entity relationships with hybrid search
- connect a knowledge graph to Claude or Cursor via MCP
- personalize agent responses using learned user facts
- track historical facts and provenance in an agent memory layer

## When to choose
- you need agent memory that updates incrementally without batch recomputation
- you want temporal awareness of when facts were valid, not just current state
- you need hybrid retrieval combining semantic, keyword, and graph traversal
- you want an MCP server to give IDE assistants persistent context
- you can run Neo4j, FalkorDB, Neptune, or Kuzu as a backend

## When to avoid
- you need a fully managed memory service without operating a graph database (consider Zep Cloud)
- your retrieval needs are simple document search served well by a vector store
- you cannot use an LLM provider, since graph construction depends on LLM inference
- you need a non-Python SDK as the primary interface

## Facets
- artifact type: framework
- maturity: active
- function: rag, agent-framework, search-engine, llm-inference, mcp
- domain: large-language-models, databases, graph-processing, developer-tools
- platform: python, self-hosted, cross-platform
- tags: knowledge-graph, temporal-graph, agent-memory, neo4j, falkordb, context-graph, hybrid-search, mcp-server, graph, ai-agents, retrieval-augmented-generation, docker

## Member repositories
- getzep/graphiti (main) score 84

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:54.420887+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T16:52:46.752895+00:00, confidence not recorded.
  - readme: https://github.com/getzep/graphiti (fetched 2026-08-28T04:11:54.420887+00:00, sha b903e99a59b7)
  - homepage: https://help.getzep.com/graphiti (fetched 2026-08-29T07:49:46.903099+00:00, sha ca504cfb7b87)
  - site_page: https://help.getzep.com/graphiti/getting-started/overview (fetched 2026-08-29T07:49:46.946781+00:00, sha b8367375b2ca)
  - site_page: https://help.getzep.com/graphiti/getting-started/quick-start (fetched 2026-08-29T07:49:46.948875+00:00, sha 818bb58e12ef)
  - site_page: https://help.getzep.com/graphiti/getting-started/mcp-server (fetched 2026-08-29T07:49:46.951086+00:00, sha defc06a91691)
  - site_page: https://help.getzep.com/concepts (fetched 2026-08-29T07:49:46.941461+00:00, sha aea70949e2d2)
  - site_page: https://help.getzep.com/changelog (fetched 2026-08-29T07:49:46.944198+00:00, sha 928712432c98)
  - site_page: https://help.getzep.com/graphiti/integrations/lang-graph-agent (fetched 2026-08-29T07:49:46.952967+00:00, sha 1378d250acdd)
- Data as of 2026-08-30T08:39:29.467469+00:00.