What problem does it solve? Getting data out of SaaS tools like HubSpot, Stripe, Salesforce, GitHub, or Slack and into a form you can query with SQL normally requires writing custom ETL code and hand-modeling schemas. This Skill automates the full path: extract data with dlt into a local DuckDB file, introspect the schema, and generate a verified Wren semantic project that answers real SQL queries. ## Core Features & Use Cases - Guided dlt pipeline setup: Pick a verified SaaS source, configure credentials via environment variables or secrets.toml, and run the pipeline into DuckDB. - Automatic project generation: The bundled introspect_dlt.py script discovers tables, filters dlt internal columns, detects parent-child relationships from _dlt_parent_id, normalizes column types via sqlglot, and writes a complete Wren v5 YAML project. - End-to-end verification: Builds the MDL and runs real SQL queries against every model before declaring success, catching catalog naming and type errors early. - Use Case: Connect your Stripe account, load customers, charges, and subscriptions into DuckDB, then immediately ask SQL questions through a governed Wren semantic layer. ## Quick Start Ask the assistant to connect your HubSpot data to Wren using dlt and generate a queryable project from the resulting DuckDB file.
