Catalog

The catalog lists every table Embrasure has indexed: tables synced into your warehouse and tables in sources you query directly. Use it to find the right table before you write SQL or ask Ember about it.

Find a table

Open Catalog. The home view has three tabs:

Search from Search catalog. The filter menu narrows results by Type, Source, Size, Last changed, and Usage, and sorts by Name, Most used, Most rows, or Recently changed. Turn on Match column names to also match tables by their column names.

Every table has a stable URL you can share with a teammate.

Inspect a table

Overview shows About this table: the source, database or location, type, estimated row count, column count, and when its statistics were last updated. Copy full name copies the database.schema.table name to use in SQL.

The Columns section lists each column's type and whether it is nullable. Select a column to see its key, references, and notes. Tables with more than eight columns have a Filter columns box.

Tables in sources you query directly also have:

• Sample rows: a few rows from the source, when the connector allows sampling.
• Notes: context for your team, such as caveats or how a table should be used. Select Edit notes to change them.

Row counts are estimates from table statistics. For an exact count, run SELECT count(*) in SQL.

See how a table is used

Usage on a table's overview covers the last 30 days:

• Queries: how many queries read the table, and the share an AI agent wrote through MCP, the CLI, or Ember.
• Agent lookups: how often an agent opened the table's catalog entry through MCP or the CLI.
• Last queried, People, By channel, and Top users, including warehouse users that are not linked to a member.
• A daily chart of queries. Hover a day for its count.

The Queries column in Columns shows how often queries read each column. A SELECT * does not count toward any column.

Usage counts queries run through Embrasure. For Snowflake, Databricks, and BigQuery sources where Embrasure reads query history, it also counts queries from that history, so it includes BI tools, dbt, and people querying directly. For other sources, queries made outside Embrasure are not visible, and the panel says so.

Use the Usage filter set to No observed reads to find tables with no observed reads in 90 days, which are candidates for cleanup. A table is only flagged once Embrasure has 90 days of history and the table is older than that.

When a workspace turns off person-level observation in governance settings, usage still shows counts but not who ran the queries.

Agents searching the catalog through MCP see frequently used tables ranked slightly higher among equally relevant matches.