The Fivetran Looker connector today syncs operational metadata from the Looker API — dashboards, looks, folders, users, query history.

It does not sync the LookML semantic layer: the model/explore/view/field definitions that live in source-controlled .lkml files.

We've prototyped this extraction in our project, which reads LookML files directly using the lkml parser and produces:

- Models — the top-level LookML model file
- Explores — entry points for queries, with join graph
- Views — view definitions with sql_table_name or derived_table SQL
- Fields — dimensions, measures, and parameters with types and SQL expressions
- Column lineage edges — COLUMN_LINEAGE_FROM triples derived from LookML SQL references

The ask: Extend the engineering Looker connector to sync this LookML metadata into the warehouse alongside the existing operational tables.

Concretely: parse the LookML project (from git or filesystem) and land lookml_model, lookml_explore, lookml_view, lookml_field tables with the relevant attributes. This would let downstream tools read the semantic layer the same way they read dashboards today — from the warehouse, no separate integration needed.