Supported data sources

Builders can use corporate sources that Ops has exposed to their Entra groups. An onboarded builder can also ask Claude to provision a database for their own app and choose its entities and actions. Sprig reads source data at query time and does not copy it into the app.

Fabric registration is always available. Schema discovery and serving require Ops to enable Fabric sources during installation.

Data source	Support and source access
Azure SQL Database	On-behalf-of (OBO). For an existing database, Ops exposes selected tables and columns read-only. For an app-provisioned database, the builder chooses its entities and read/write actions. The database applies each signed-in builder’s or viewer’s permissions and row-level security.
Fabric Warehouse	Service principal, via Sprig’s managed identity. Read-only tables are supported when they declare a primary key. Views cannot be exposed. Sprig enforces each viewer’s reader-group access.
Fabric lakehouse SQL analytics endpoint	Service principal, via Sprig’s managed identity. Registration and schema discovery are supported. Its keyless tables and views cannot be exposed yet. Sprig enforces each viewer’s reader-group access.
Power BI / Fabric semantic model	On-behalf-of (OBO). Ops exposes selected tables and measures read-only; every query runs under the signed-in viewer’s own identity, so the model’s row-level security applies. Reader groups additionally gate who can query at all.

See Building on exposed data for working with corporate sources, Publishing and sharing for app-provisioned storage, and Data access for the authorization differences between the two access modes.