DuckDB

Databases None Windmill docs

Run analytical SQL over files — Parquet, CSV, JSON, on S3 or in the workspace — from any Windmill script or flow.

Set up in 2 steps

Write a DuckDB script

Pick DuckDB as the script language and write plain SQL. There is no resource to create — DuckDB runs inside the job.

Point it at your data

Query a file directly by path. To read from object storage, attach an s3 resource and address the file through it.

Scripts

Building blocks, if none of the projects above fit. Fork one, run it, or call it as a step inside your own flow.

ETL / Data pipeline on tabular data

Script duckdb Verified

Questions

Why is there no resource?

DuckDB runs in the job itself, with no server to reach. Credentials only come in when the data lives elsewhere — an s3 resource for object storage, for instance.