The public data catalog
Bring your data. Add the world.
Weather, holidays, places and more. Public datasets for people and agents to discover and join with their own data using SQL, each with its source, licence and a data card written for an agent.
Live catalog · eighteen datasets · the list as SQL · Parquet · Markdown · JSON
The datasets
The catalog is read from /pubdata/registry.json, rebuilt from the public tables themselves. Browse it at lakehousebox.com/app/commons/pubdata; with any LakehouseBox account, attach it read-only in your DuckDB with lhbox duckdb --catalog commons/pubdata (MCP: open_public_catalog).
The list is a table too.
Every dataset is a row of meta.datasets in the catalog, every column a row of meta.columns, refreshed with the data. Ask it which tables join on a key, or which refresh daily, in the same SQL.
datasets.parquet · columns.parquet · public-data.md · registry.json
SELECT dataset_id, refresh, row_count, join_keys FROM commons_pubdata.meta.datasets WHERE list_contains(join_keys, 'country_code');
More useful together
Your sales.
The weather.
One query.
Explore whether rainy days and public holidays coincide with changes in your sales.
Query with DuckDB. Save the result to your catalog.
SELECT s.day, s.revenue, w.rain_mm, h.holiday_name FROM private.sales_daily AS s LEFT JOIN public.weather_daily AS w ON s.day = w.day AND s.location_id = w.location_id LEFT JOIN public.holidays AS h ON s.day = h.day AND s.country_code = h.country_code;
Examples show the join pattern. Each dataset page gives its coverage and join keys.
How the catalog is meant to be used
Discover
Find data with clear sources and coverage.
Join
Combine it with your own tables.
Keep
Save useful results for your next session.