Skip to Content
Data VaultTablesOverview

Tables overview

Tables are the structured, relational data at the heart of a project — the database your agents query and your apps read and write. You model tables, browse and edit their rows, view how they relate, and run SQL against them, all from the Tables area of the Data Vault.

Open it from Data Vault → Tables in the project sidebar. The header reads Data Tables.

Data Tables screen with the table list on the left and a table's rows on the right

The three views

The Tables area has three tabs across the top, plus an AI assistant:

  • Browse Tables — the table list, the row grid, and everything you do with a table’s data and access.
  • ER Diagram — an auto-generated map of your tables and their relationships.
  • Query Explorer — a SQL workspace to validate, run, and save queries.
  • Data Assistant — describe what you want in plain language and let AI create tables, change schema, and generate sample data.

Schemas: Shared vs Dedicated

Every table lives in a database schema. A schema selector sits at the top-right of the Tables area.

  • Shared — the default schema every project has. Tables are isolated from other projects by a subscription_id column and a Row-Level Security (RLS) policy that the platform adds automatically.
  • Dedicated — a separate schema with its own PostgreSQL credentials, available on higher plans. When your plan includes only the shared schema, the selector is disabled with an “Upgrade to Pro+ for dedicated schemas” hint.
ℹ️

You don’t create a schema by hand. The shared schema is created automatically the first time you add a table. Dedicated schemas come with the plans that include them.

How every table is built

When a table is created — by you or by the Data Assistant — it always gets:

  • A primary key (id), in one of three styles: UUID, Auto Increment (1, 2, 3…), or Prefixed (for example INC000001).
  • Optional audit columns — created_at, updated_at, created_by, updated_by — on by default.
  • In the shared schema, a subscription_id column plus the RLS policy for tenant isolation.

How agents and apps use tables

  • Agents read and write tables through the built-in database tools — db_query and db_get_schema for reading, and the write tools for changes.
  • Vibe-coded apps read and write the same tables through the app database tier.
  • All access runs under Row-Level Security, honoring the row filters and column masks set for each role.

Next

OverviewCreate tables