Hi all, About a month ago I posted on here announcing Slater, a graph db project I've been working on. Lots of good feedback, suggestions and github stars (thank you!).
A suggestion I heard here and in a few other places was to add some practical use case examples such as an LLM agent memory example, so in honour of 100 github stars I figured it was a good idea. So I borrowed a few details from FalkorDB and added some missing features (including BM25 full text indexing) to enable Graphiti to use slater as a backend store.
I added a sister repo with instructions called graphiti-slater which has everything you'll need to run a local docker compose stack with agent memory served through MCP.
A suggestion I heard here and in a few other places was to add some practical use case examples such as an LLM agent memory example, so in honour of 100 github stars I figured it was a good idea. So I borrowed a few details from FalkorDB and added some missing features (including BM25 full text indexing) to enable Graphiti to use slater as a backend store.
I added a sister repo with instructions called graphiti-slater which has everything you'll need to run a local docker compose stack with agent memory served through MCP.
Available here:
- Slater: https://github.com/hikari-systems/slater
- Slater adapter for graphiti: https://github.com/hikari-systems/graphiti-slater
- Graphiti itself: https://github.com/getzep/graphiti