feat: support embedded databases as alternatives to Memgraph

[vitali87]

Task Description

Explore supporting embedded/lightweight databases as alternatives to Memgraph, removing the requirement for Docker and a running database server. This would significantly lower the barrier to entry for new users.

Suggested candidates:

Database	Type	Notes
Kuzu	Graph + Vector	Embedded, Cypher compatible, drop-in Neo4j alternative
LanceDB	Vector	Embedded, serverless
Chroma	Vector	Embedded, simple API
Turso	SQL + Vector	SQLite based, embedded

Kuzu is particularly interesting because it supports Cypher queries natively, which could minimize changes to the existing query pipeline.

Acceptance Criteria

• Evaluate at least one embedded graph DB (Kuzu) for compatibility with existing Cypher queries
• Evaluate at least one embedded vector DB for semantic search
• Prototype a backend that works without Docker
• Abstract the storage layer so users can choose between Memgraph and an embedded alternative

References

• Original discussion: Run this project as MCP Server #61 (comment by @qdrddr)

Priority

• High
• Medium
• Low

Estimated Effort

• Small (< 2 hours)
• Medium (2-8 hours)
• Large (> 8 hours)

[qdrddr]

Kuzu is dead. Long live https://github.com/LadybugDB/ladybug (a fork).

Turso is working on a graph extension (not released yet).

[lvca]

Please consider also https://arcadedb.com open source dbms, Apache 2 license, graphs and vector as multi-model.

[qdrddr]

A Fast LPG Graph DB with Cypher language.
Can run without a Docker container. Though it still requires a network connection and is technically not in-process / embeddable at the moment. The TurungDB team is working on releasing a fully embeddable version soon.
https://github.com/turing-db/turingdb

uv pip install turingdb
turingdb -demon

# Set working graph
client.set_graph("mygraph")

# Create a new change on the graph
change = client.new_change()

# Checkout into the change
client.checkout(change=change)

# Create a node Person (Jane) - Edge (knows) - node (John)
client.query("CREATE (n:Person {name: 'Jane'})-[e:KNOWS]->(m:Person {name: 'John'})")

# Commit the change
client.query("COMMIT")
client.query("CHANGE SUBMIT")

# Checkout back to main
client.checkout()

# Query graph
client.query("MATCH (n) RETURN n")