cube

Why we included this project

Cube Core is a self-hosted semantic layer: you define metrics, dimensions, joins, and access rules once in code, and it serves that model through SQL, REST, and GraphQL to whatever needs it downstream. That makes it attractive for product teams embedding analytics into their own apps, and for data engineers who want one governed metric layer feeding several BI tools instead of each tool defining its own. It is also a reasonable foundation for AI agents that need consistent, trustworthy numbers. Because it ships no UI, you are not stuck with a dashboard you have to work around; you bring your own front end or point an existing tool at it. It connects to Snowflake, Databricks, BigQuery, Postgres, and other SQL databases, so it usually slots into an existing stack without forcing a migration, and the built-in caching engine is worth knowing about if you need sub-second responses under concurrent load.