Taruvi Refine frontend

Use @taruvi/refine-providers to connect Refine.dev admin UIs to Taruvi. This skill covers the six providers, how to wire resources, the meta vocabulary that unlocks Taruvi-specific features (populate, aggregate, graph, upsert), and the gotchas that aren't obvious from the package API alone.

This skill is the frontend layer. If you're provisioning backend resources, switch to taruvi-backend-provisioning. If you're writing Python for a Taruvi function body, switch to taruvi-functions.

Core principles

1. Resource name = datatable name by default. Override with meta.tableName when they must differ (e.g., resource: "active_users" queries tableName: "users" with a filter).
2. Meta is the control surface. Refine's meta object is how you reach Taruvi-specific features (populate, aggregate, graph, bucketName, idColumnName, upsert, deleteByFilter). Don't try to push these through filters or query params.
3. Multiple providers, one client. Register all six providers against the same Client instance. Select providers by dataProviderName in hooks.
4. Auth is redirect-based. There is no credentials login in the default Refine flow. authProvider.login() redirects to Taruvi's login endpoint.
5. AccessControl batches. Permission checks are debounced 50ms via DataLoader. Tests that don't await will flake.

Setup