Set up Cube
You probably need a Cube Cloud administrator on your deployment to generate the API key—you may not have access yourself.
Atlan uses Cube's REST API to extract metadata. The connector authenticates with a Cube API key and your Cube tenant URL. From those, it discovers your deployments and environments, mints a short-lived metadata-sync token, and crawls the semantic model (cubes, views, dimensions, measures, and segments).
Gather your tenant URL
Your Cube tenant URL is the host you use to reach Cube Cloud, without the https:// prefix or any trailing path—for example your-org.cubecloud.dev. You enter this host and the API key when you crawl Cube.
Generate API key
- Sign in to Cube Cloud as an administrator.
- Generate a Cube API key with access to the deployment and environment you want to catalog. See Cube's authentication documentation for the current location of API-key management in your Cube Cloud instance.
- Copy the generated key and store it somewhere secure. You enter it when you crawl Cube.
Scope the key to what you want cataloged. The API key governs which deployments and environments Atlan can read. Use a key with access to every deployment and environment you intend to catalog so the crawl returns a complete semantic model.
Confirm warehouse access (optional but recommended)
For end-to-end lineage from warehouse tables and columns through into your Cube model, Atlan also needs:
- An Atlan connection to the upstream warehouse that Cube is built on (Snowflake, BigQuery, Redshift, Databricks, or any other supported source).
- The warehouse connection crawled before or alongside the Cube crawl, so the upstream tables and columns exist in Atlan when Cube's cubes and members are processed.
- Cube's data sources exposing the database name for each connection—Atlan builds the upstream warehouse identity (connector type, database, schema, table, column) from Cube's data-source metadata to resolve lineage.
Without the upstream warehouse connection, Cube metadata still catalogs cleanly but the lineage Process edges from warehouse tables and columns into your cubes and members are missing.
What's next?
You're ready to crawl Cube assets.