Add a saved_queries block in a YAML file within your dbt project, providing a name and label for the saved query
Under the metrics key, list one or more metric names that should be returned together
Under group_by, list the dimension names using the entity__dimension format (e.g., order__order_date) to specify the slicing grain
Optionally add a where clause filter using Jinja-style syntax to pre-scope the query (e.g., filter by a specific channel)
Test execution with dbt sl query --saved-query YOUR_SAVED_QUERY_NAME and verify the output matches expected values before committing
Known gotchas
Saved queries do not support all metric types equally; conversion metrics and derived metrics with offsets may have constraints on which dimensions can be combined in a saved query
The saved query name must be unique across the entire dbt project; duplicates will cause a parse error during dbt parse
Saved queries are consumed by some native integrations (such as the dbt Semantic Layer Tableau connector) to pre-populate recommended analyses; poorly named saved queries will surface confusingly in those UIs
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