What problem does it solve? Getting data into MotherDuck correctly requires choosing the right ingestion path for each source type, and picking the wrong one leads to slow loads, failed imports, or unsupported operations on the Postgres endpoint. ## Core Features & Use Cases - Source-Based Path Selection: Classifies sources as object storage/HTTPS, local files, in-memory dataframes, or external databases, then picks CTAS, INSERT...SELECT, COPY, or Arrow bulk loads accordingly. - Client Path Guidance: Distinguishes native DuckDB clients (Python duckdb, Node.js @duckdb/node-api) from Postgres-endpoint thin clients, clarifying which operations each supports. - Format and Cloud Coverage: Provides advanced options for CSV, Parquet, JSON, Delta Lake, and Iceberg, plus CREATE SECRET authentication patterns for S3, GCS, and Azure. - Use Case: You have partitioned Parquet files in S3 and need them in MotherDuck. The skill guides you to a remote-read CTAS with hive_partitioning, then validates row counts before promoting to modeled tables. ## Quick Start Load the Parquet files from my S3 bucket into a new MotherDuck staging table and validate the row counts.
