neocarta vs OpenMetadata
Neo4j Labs' semantic layer for data agents: ingest warehouse schema, business glossary, metrics and query history into one graph, then serve it over MCP so agents route queries and write grounded SQL. — versus — Open metadata platform turned AI context layer: 130+ connectors feed a unified knowledge graph of lineage, quality, ownership, glossaries and contracts — served to agents via MCP and APIs.
OpenMetadata does the catalog-as-agent-context move at platform scale: 130+ connectors, lineage, ownership, quality and contracts in one graph served over MCP. Neocarta is a small focused library for the query-routing and Text2SQL slice — much less to run, much less covered.
| neocarta | OpenMetadata | |
|---|---|---|
| Stars | 98 | 15k |
| Forks | 26 | 2.3k |
| Language | Python | TypeScript |
| License | Apache-2.0 | Apache-2.0 |
| Last activity | 4 days ago | yesterday |
| Topics | knowledge-graphs, data | knowledge-graphs |
| Curated connections | 5 | 4 |
neocarta — the curator's take
The honest fix for Text2SQL: the model is not bad at SQL, it is blind to your data landscape. Neocarta pulls schema metadata, foreign keys, sample values, glossary terms, governed metric definitions and real query history into a Neo4j graph — only metadata crosses over, data stays in the source — then serves it to agents over MCP with full-text, vector and hybrid search that returns columns, types, example values and the FK references needed to build a join. Embeddings are optional; catalog search works from schema alone. It ships a runnable LangGraph + BigQuery agent so you can see the routing loop end to end. Read the label though: Neo4j Labs, explicitly experimental, not a supported product, 98 stars, and it assumes you are willing to stand up and maintain a Neo4j instance next to your warehouse.
OpenMetadata — the curator's take
The mature-platform play on this shelf: a decade-class data catalog (lineage, contracts, governance) that now speaks MCP, which makes it the most battle-tested 'what does this data mean' answer an agent can get. When NOT: this is a platform with platform weight — Elasticsearch, MySQL/Postgres, ingestion framework — absurd overkill if you just want a knowledge graph over one source; the AI-context framing is new even if the metadata engine isn't.