motherduck-load-data

What problem does it solve?

This skill helps you load data into MotherDuck correctly and efficiently, avoiding common ingestion mistakes like row-by-row inserts, wrong client paths, and unsafe transformations.

Core Features & Use Cases

• Ingestion-path selection: Choose between object storage/HTTPS remote-load patterns and native DuckDB client ingestion for local files, local DuckDB, and in-memory data.
• Production-friendly loading posture: Prefer batch loading, Parquet-first movement, CTAS and INSERT...SELECT patterns, and minimal transformations at the raw landing layer.
• Validation and promotion guidance: Recommend row counts/type checks after landing, then promote into staging/modeling only once the load is correct.
• Workflow fit: Works across first loads, appends, and whole-database uploads while accounting for Postgres-endpoint thin-client versus DuckDB-client behaviors.

Quick Start

Use motherduck-load-data to plan and generate the correct CTAS/INSERT workflow for your source system and target tables, such as loading Parquet from object storage into a raw/staging table with the right remote-read settings.