Skip to main content
The Catalogue is the inventory of governed datasets in your workspace — the same datasets your AI tools can query over MCP. It’s a read-only browse: each dataset is a card showing its friendly title, the exact technical name (e.g. silver_sales), a one-line description of what it holds, its layer, and how many columns it documents.

Layers

Every dataset sits in one of four layers, describing how far it’s been refined from raw source data toward analysis-ready. The catalogue orders the most query-ready first:
LayerWhat it is
SilverCleaned and joined — conformed data you can query directly
GoldAggregated, analysis-ready tables built for reporting
CanonicalYour mastered records (see canonical entities)
BronzeRaw mirrors of each source, exactly as it arrived
The Catalogue is where you confirm what’s available to ask about. To see which systems feed these datasets and how fresh they are, head to Sources.