DocsWhat you can doShare data publicly

For agents

Share a dataset with the world.

Put your data in its own catalog and publish it. Anyone can query it, no account needed; the rest of your data stays private.

  • Agent-guided

Before you start: Claude Code (or any other agent)Your data (a CSV or Parquet file)An admin login to your account

Ask your agent to load and publish it

Say what the data is and that it should be public. A good agent keeps it apart from your private data.

Claude Code (or any other agent)Example conversation
You

I measure NO2 around my neighbourhood (~/Downloads/no2-barrio-2026-09.csv). Put it in LakehouseBox and publish it so anyone can query it. Follow lakehousebox.com/docs/

Agent · example reply

I've installed the LakehouseBox command line. First, approve this computer in your browser (tick write).

My plan: put your readings (3,600 rows, one month) in a separate catalog called no2_barrio, since publishing makes a whole catalog public; your other data stays private. Then load the CSV as one table and publish that catalog.

Before you agree: everything in it becomes readable by anyone, including the station names, their coordinates and the table's history.

Example replies show what to expect. Your agent adapts to your setup.

Approve and decide

You approve the sign-in and confirm the plan. Publishing is your decision.

Your turn · in the browser

Approve your agent's access

Open the link your agent gives you, check the code, tick Write and approve. Then tell your agent the plan is fine, or what to change (another catalog name, fewer columns: say if the coordinates should not be public).

Publish it yourself

Your agent loads the data, but only an admin of your account can make a catalog public.

Claude Code (or any other agent)Example conversation
Agent · example reply

Your NO2 data is in LakehouseBox, but it isn't public yet: the key this computer got can load data, not publish. To publish: log in at lakehousebox.com/app/, open the no2_barrio catalog, choose Publish and type its name to confirm. Only no2_barrio becomes public.

Example replies show what to expect. Your agent adapts to your setup.

Your turn · in the browser

Publish the catalog

On your account page, open the catalog, choose Publish and type the catalog's name to confirm. Your agent cannot do this for you, on purpose: making data public is a decision for a person with admin rights. Then tell your agent "published".

Your agent checks it from the outside

It queries the data with no account and no keys, the way a stranger would, and gives you what to share.

Claude Code (or any other agent)Example conversation
You

published

Agent · example reply

Your NO2 data is public. I queried it from a fresh DuckDB with no account and no keys, and all 3,600 rows came back.

Public page: lakehousebox.com/app/your-account/no2_barrio

For September, Escuelas Aguirre and Plaza de España had the highest averages, about 26 µg/m³; the peak was 53.1 at Plaza de España.

  • Only no2_barrio is public. Your other catalogs stay private.
  • You can make it private again at any time, but anyone who already copied the data keeps their copy.
Example replies show what to expect. Your agent adapts to your setup.

Example · verified connection

Readable without an accountRest of the account privateIn the recorded run: 3,600 rows read back by an anonymous DuckDB session.
What becomes public, exactly

Everything in the published catalog: every table's data files and its history (older versions, until they are cleaned up), the catalog's guide for agents if it has one, and the list of its files. Your account's handle appears in every public link. People can read and query; nobody can write or delete. Other LakehouseBox users can also attach your catalog read-only by name. Making it private again stops new reads at once.

  • Make no2_barrio private again.

  • Add October's readings to the public table.

  • Write a short description of the dataset for people who find it.