getzep/graphiti

Build Real-Time Knowledge Graphs for AI Agents

What it solves

Graphiti addresses the limitations of static knowledge graphs and traditional RAG (Retrieval-Augmented Generation), which often struggle with frequently changing data. It allows AI agents to maintain a "context graph" that tracks how facts evolve over time, ensuring that agents can distinguish between what is currently true and what was true in the past without needing to recompute the entire graph when new data arrives.

How it works

Graphiti builds a temporal knowledge graph from structured and unstructured data. It uses "episodes" (raw source data) as the ground truth, from which it derives entities and relationships. Each fact is assigned a validity window (start and end time), meaning old facts are invalidated rather than deleted when new information contradicts them. The framework supports both prescribed ontologies (defined via Pydantic models) and learned ontologies (emerging from data). Retrieval is handled through a hybrid approach combining semantic embeddings, keyword search (BM25), and graph traversal.

Who it’s for

It is designed for developers building interactive, context-aware AI agents that operate on evolving, real-world enterprise data or user interactions where historical accuracy and provenance are critical.

Highlights

  • Temporal Fact Management: Tracks the validity of facts over time, preserving historical context.
  • Provenance: Every derived fact traces back to the specific "episode" of raw data that produced it.
  • Incremental Updates: Integrates new data in real-time without requiring batch re-processing.
  • Hybrid Retrieval: Combines semantic, keyword, and graph-based search for high-precision, low-latency queries.
  • Pluggable Backends: Supports multiple graph databases including Neo4j, FalkorDB, and Amazon Neptune.

Related

  • Project
    deeplethe/utopia

    An enterprise world model and bitemporal knowledge graph that integrates ontology and time-awareness to provide a trustworthy, auditable foundation for AI agents and decision-making.

  • Project
    xoai/sage-wiki

    A graph-based knowledge base that uses LLMs to compile raw documents into an interlinked wiki and knowledge graph for shared human-AI memory.

  • Project
    vercel-labs/lat.md

    lat.md is an npm‑distributed CLI that creates a markdown‑based knowledge graph for a codebase. It lets you write linked documentation in `lat.md/`, annotate source files with `@lat` comments, and validates the graph. Agents and developers can search it semantically (offline MiniLM or remote OpenAI/Vercel embeddings) and use commands like `lat check`, `lat locate`, and `lat expand` to keep design decisions, constraints, and test specs in sync with the code.

  • Project
    angelnicolasc/graymatter

    GrayMatter is a Go‑written, single‑binary memory layer for LLM agents. It stores facts persistently, retrieves the most relevant ones (top‑8 by default), and reduces context tokens by about 90 %. It also auto‑builds a typed knowledge graph, integrates with many MCP‑compatible clients (Claude Code, Cursor, Codex, etc.), and provides a terminal UI and health‑checking tools. Install via `go install`, Homebrew, Scoop, or a pre‑built binary; use the CLI (`graymatter recall`, `remember`, `kg render`, etc.) or embed the Go library directly in your agent code.

  • Project
    trailhq/Graft

    Graft is an open‑source TypeScript/Node.js tool that builds a local, markdown‑based knowledge graph of a codebase (summaries, key snippets, typed links) using tree‑sitter for structural parsing and optional LLM summarisation. The graph is cached in a `graft/` folder, refreshed only on file changes, and automatically wired into coding agents like Claude Code, Cursor, Codex, and Gemini. Benchmarks claim up to 46 % fewer tool calls, 42 % token savings, 60 % latency reduction, and a 12‑point correctness boost on SWE‑bench. Integration is via a single `graft init` command that writes agent‑specific instruction files; the CLI also offers search, visualization, and telemetry controls. The project is MIT‑licensed and targets faster, cheaper AI‑assisted development.