Browsing Tables
The Tables tab lists every user table in the connected database, with a per-table inspector for schema detail. It’s metadata-only by design: AccessVault shows you table structure and health statistics, not row data. (The one-click MCP tool below is how you sample actual rows — through the audited bridge.)
The table list
Type in the Filter tables… box to narrow the list. Each row shows:
| Column | What it is |
|---|---|
| Name / Schema | Table name and its schema. |
| Rows | The planner’s row estimate — fast and approximate, not a COUNT(*). |
| Size | Total on-disk size. |
| Columns | Column count. |
| Last Analyzed | When statistics were last gathered. |
A red dot next to a table flags bloat: dead tuples exceeding 20% of live rows — a hint that vacuum isn’t keeping up. The footer totals the listed tables’ size and estimated rows.
The table inspector
Select a table to open its inspector:
- Stats — heap, index, and TOAST sizes; live rows and dead tuples; bloat ratio; sequential vs index scan counts and index usage.
- Schema — every column with its type, nullability, default, and a key icon on primary-key columns.
- Indexes — each index with its size,
PRIMARY/UNIQUEbadges, and full definition. - Foreign Keys — each constraint with the local column, the referenced
schema.table.column, and the constraint name.
One-click MCP sampling tool
At the top of the inspector, Add SELECT * tool to MCP catalog creates a catalog tool that
runs SELECT * FROM schema.table LIMIT 100 — a quick way to give Claude a capped sample of a
specific table without writing SQL yourself. The tool lands in the connection’s
catalog, where you can rename, edit, or disable it like any other
tool. Calls to it are audited like everything else that crosses the bridge.