Keystone

Single-table design

When several kinds of thing share one table, the grid on its own tells you very little. Keystone lets you tell it what each row is, and then uses that everywhere.

Entity types

The Entity Types sheet with detected types listed
Detect proposes types from the rows you have loaded; nothing is applied until you add it.

Open Entity Types… from the share menu. Press Detect and Keystone proposes rules from the loaded rows — a sort-key prefix such as ORDER#, or a discriminator attribute with a handful of values. Accept the ones that are right, or write rules by hand. Rows are then colour-coded and labelled in the leading strip. Rules are saved per table as JSON, so a team can share them.

Focusing on one type

The grid focused on one entity type, showing only its rows and columns
Focused on one type: only its rows, and only the columns those rows carry.

The strip's header becomes a picker. Choose a type and the grid shows only its rows, shaped to the attributes those rows actually have. Hide columns while focused to keep a custom set for that type.

Item collections

The item collection view listing every item sharing a partition key, grouped by type
Everything for one partition key, grouped by type and ordered by sort key.

Right-click any row → Show Item Collection. Keystone runs one Query for that partition key and lays the results out by type — the customer, then their addresses, then their orders — each expandable to every attribute, with a button to open it in the editor.

Aliases

In the same sheet, give attributes readable names: gsi1pk becomes “Customer”. The grid shows the alias with the real name a hover away, and every request still uses the real name.