Awesome MCP › Databases

memgraph/ai-toolkit

⭐ 112 Python repository created 2025-04-01

Memgraph AI Toolkit is the vendor-maintained collection for using Memgraph, an in-memory graph database, in AI applications. It ships as several packages. memgraph-toolbox is the core layer of graph operations. mcp-memgraph is the MCP server, which is what makes the database reachable from an assistant: it exposes running Cypher queries, inspecting the schema, and the toolbox operations as MCP tools, so a model can explore a graph's structure and then query relationships across it. langchain-memgraph provides the equivalent integration for LangChain applications, and unstructured2graph handles the ingestion direction, turning unstructured documents into graph structure. The stated purpose is graph-powered retrieval-augmented generation: rather than retrieving text chunks by vector similarity alone, a query can traverse explicit relationships between entities, which answers connective questions that flat retrieval handles badly. Getting started is a Docker run of memgraph-mage with schema information enabled — the flag matters, since it is what lets the MCP server report the graph's structure — followed by pip installing the packages needed. Everything is Python and published on PyPI, with the packages versioned independently. The audience is teams building knowledge-graph or RAG systems on Memgraph who want both the agent-facing MCP surface and the framework integration maintained by the database vendor.

https://github.com/memgraph/ai-toolkit

mcpmemgraphgraph-databasecypherpythonlangchainragknowledge-graph