KGLite: a Rust property graph with MCP and Bolt

KGLite is a fast Rust property graph with OpenCypher, a built-in MCP server, Bolt compatibility, OKF markdown storage and a YAML ontology.

Illustration: KGLite: a Rust property graph with MCP and Bolt

It’s a fast (Rust) inline property graph (OpenCypher) with tons of delicious features:

• builtin MCP server, you can even tune the tools via a simple YAML config
• Neo4j compatible Bolt service, you can consume the graph via any Bolt connection
• OKF storage: give it an OKF or Obsidian repo and it will serve it as a knowledge graph. That is, a markdown backend compatible with Google open knowledge format
• CSV and Pandas support: this includes support for spatial data and timeseries
• graph algorithms: shortest path and all that
• semantic search and vector embeddings
• fluent API in case you don’t like to use Cypher
• YAML ontology (!)
• can act as a self-contained memory for agents (schema exposure and so on)

…and much more.

Yes, it comes with the usual one-dev-few-stars warning. In comparison with LadybugDB and similar Python/Rust inline graph databases, it has something delightful and well-designed. The extensive documentation reveals someone with a vision and I look forward to where this is going.

https://kglite.readthedocs.io/en/latest/index.html