Agents > Xedant Data Agent > Docs > Lakehouse

Lakehouse

The Lakehouse is the data root folder on your server where the scripts put everything they collect. The data is arranged on “project → source” shelves, and the Lakehouse browser shows it clearly. The name is borrowed from data engineering: a “data lake” (lake) + a “data warehouse” (house). The data is your own plain files on your own disk — copy and move them freely.

What lies in a source folder

README.md — the source’s description: where the data comes from, the structure, the fields, the change history;
*.parquet — the raw data exported by the extract.py script; Parquet is always the source of truth and is never rebuilt;
*.sqlite — the database for queries and visualization: built by the load.py script and rebuilt on every run;
*.log — logs of extraction, preparation and loading.

The projects and sources overview

The Lakehouse’s home screen is a table of all projects and sources: size, freshness of the data, whether a ready database is present (marked with an icon). Search and sorting work. If the store is empty, the interface suggests: run a load script — the data will appear here.

The folder browser

Inside a source it is a regular file browser: breadcrumbs at the top, folders before files, a folder-size summary. If the folder has a README, it renders above the listing — you immediately see the source’s description without opening files.

Viewing files by type

Parquet — the data schema and the first rows (Apache Arrow) — a quick check that the export worked;
SQLite — a table browser: tables, fields, rows;
Markdown — a finished document, like a page;
text and logs — a paging viewer;
any other type — refuse-free message: only text and Markdown files can be viewed.

The link to data sources

A SQLite database referenced by a data source carries a badge in the Lakehouse — you can see at a glance that the source is registered and ready for queries. Details on the Data Sources page.

Why you care

You always see what has been collected, from where, and how fresh it is — no need to take the run logs’ word for it. Deleting the extra or moving to another server are ordinary file operations. And the prepared databases from the Lakehouse automatically become sources for Analytics Agent — that has its own page.