Files
In the browser
Section titled “In the browser”Open the builder and use Explore data: pick CSV, Parquet or JSON files from your disk, or paste URLs. They are registered with the in-browser DuckDB under their file names, and Explore writes a first dashboard over them. Nothing leaves your machine.
CREATE TABLE sales AS FROM read_csv('sales.csv'); -- a file you openedSELECT region ::XAXIS, sum(revenue) ::BARCHART FROM sales GROUP BY ALL ORDER BY 2 DESC;With the CLI and the servers
Section titled “With the CLI and the servers”Paths are relative to where the process runs. A dashboard that should work in several places reads its folder from a variable instead of a hard-coded path:
CREATE TABLE sales AS FROM read_parquet(getvariable('anofox_data') || '/sales/*.parquet');dashboard --data <dir>, dashboard --site … --data <dir|url> and the locked server’s data folder set anofox_data. A published static site can list its files so the browser downloads each one once.