Memgraph

Active
GitHub C++ NOASSERTION

Description

High-performance open-source in-memory graph database built in C++, Cypher-compatible, designed for GraphRAG, AI memory, agentic AI, and real-time graph analytics with native streaming and Kafka integration.

Key Features

  • In-memory graph storage — high-performance C++ implementation with nanosecond-level traversal
  • Cypher-compatible — supports openCypher query language and reuses the Neo4j ecosystem
  • GraphRAG-native — built-in vector indexing and graph algorithms to simplify retrieval-augmented generation
  • AI memory layer — supports persistent memory, context graphs, and agent state management
  • Streaming integration — native Kafka/Pulsar integration for real-time data ingestion
  • Multi-language clients — Python, Java, Go, Node.js, and .NET drivers available

Use Cases

💡 Acting as a long-term memory layer for agents, storing entities, relations, and context state
💡 Building GraphRAG systems that perform hybrid retrieval and multi-hop reasoning over graphs
💡 Processing real-time event streams into graph databases to drive agent decisions
💡 Replacing Neo4j in high-write scenarios where memory footprint is a bottleneck
💡 Running real-time graph algorithms in finance, fraud detection, and recommendation systems

Tags

graph-database graphrag ai-memory cypher streaming c++

Categories

🧩 Agent Memory 📚 RAG Tools

Quick Start

docker run -p 7687:7687 -p 7444:7444 memgraph/memgraph:latest
pip install gqlalchemy
from gqlalchemy import Memgraph
mg = Memgraph("127.0.0.1", 7687)
mg.execute("CREATE (:Agent {name: 'alpha'})-[:REMEMBERS]->(:Fact {text: 'user prefers dark mode'})")
Visit GitHub Visit Website View Docs

Related Projects