Query builder

The builder produces typed, parameterized queries validated against the loaded schema and compiled with bound parameters. It supports inner joins, nested filters, sorting, column grouping, and a row limit you control, up to the cap the application controls (see Limits & risk model).

Schema first

Map is its own screen in the sidebar: tables, keys, indexes, and referenced external tables, with system schemas kept out of the way. The map is read-only and does not rewrite the query you are building. The builder also has a Schema rail for adding tables to the canvas. Looking at things is safe here, which is rather the point.

Joins and grouping

Joins are INNER JOIN on column equality (a.x = b.y). Each join column must be the leading key of an equality-capable index, or the join must be a complete foreign key. Outer joins are not on the menu.

Group by in the builder is SQL grouping of columns, not Explore aggregations. COUNT and friends belong in Explore, on the sample you already fetched.

Running a query

The button says Run. Before execute, the query is scored from the schema, then the database is asked for a plan. A clean assessment runs immediately. A missing plan or a missing optimizer cost requires an explicit confirmed retry, labeled Run with safeguards. A high but valid cost does not.

Working with the sample

Results arrive as a bounded sample. Open Explore for Pivot, Worksheet, and Visualization. Those modes work on the fetched rows; they do not query the database themselves. Recipes can optionally restore and re-run the builder query. Pivot all you want; you still cannot raise the row cap.