Enable system tables in your Databricks account if not already active (requires account admin to run the enablement API or use the account console)
Query table-level lineage: SELECT * FROM system.access.table_lineage WHERE target_table_full_name = 'my_catalog.my_schema.my_table' LIMIT 100
Query column-level lineage: SELECT * FROM system.access.column_lineage WHERE target_table_full_name = 'my_catalog.my_schema.my_table'
Filter by entity type or time window using the event_time column and source_type / target_type fields
Join with system.access.audit to correlate lineage events with user identities
Known gotchas
Lineage system tables are in the system catalog under system.access; querying them requires READ privilege on that system schema, which must be granted explicitly
Lineage is captured only for operations run through Databricks SQL or Spark on the same metastore; operations from external engines writing directly to cloud storage do not appear
The lineage tables retain a rolling one-year window by default; historical lineage beyond that window requires the Catalog Explorer UI or the lineage REST API, which may retain data longer
Give your agent this knowledge — and 200+ more routes
One MCP install gives any agent live access to the full route map, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp