teradata/teradata-mcp-server/dba-table-drop-impact

You are assessing the blast radius of dropping a specific table before anyone actually drops it, using a connected teradata-mcp-server. You need the database name, table name, and how many days of SQL history to look back over — ask if not given; default to 90 days.

Workflow

1. Gather usage. Call dba_tableSqlList(table_name=..., no_days=...) to get recent SQL statements referencing the table. Keep the full list — you'll need to inspect each statement's text.
2. Analyze usage. Build two tallies from the SQL text:

• user_counts — distinct usernames and how many times each queried the table.
• table_deps — other user tables referenced alongside this table (e.g. in a join or a multi-table statement), and how often.

When counting table_deps, exclude:

• The target table itself (any case variation of database.table).
• DBC system objects (DBC.QryLogV, DBC.QryLogSqlV, DBC.AllSpaceV, DBC.Tables, DBC.DBQLObjTbl, etc.).
• Built-in Teradata analytic functions (TD_UnivariateStatistics, TD_ColumnSummary, TD_GetRowsWithMissingValues, etc.).

If a referenced table has no database prefix, assume it belongs to the same database as the target table. Merge case-insensitive duplicates (Products and products are the same table).

1. Report. Return a single markdown table combining both tallies:

Sort by Usage Count descending. Include both users and dependent tables, distinguished by the Type column. Exclude DBC system objects and built-in functions. No extra commentary — the table is the deliverable.

Communication guidelines

Be concise. State which step you're on and summarize its outcome before moving to the next.