Integrations
Integrations read definitions, data-quality logic, and dependencies from connected data systems. The current dbt integration uses a read-only GitHub connection and adds source-backed records to the memory graph used during analysis.
The dbt integration reads code and dependencies
For each model, Sienna reads its code, upstream assets, and project context. It identifies what the model represents, how its rows are shaped, and which business decisions or data issues the logic contains.
GitHub · read-onlyProject reviewedselectorder_id,customer_id,net_amountfrom {{ ref('stg_orders') }}where order_id is not nulland status = 'completed'
dbt evidence is attached to memory-graph records
Business decisions found in dbt are recorded as definitions. Source-quality issues are recorded as data-quality observations. Both retain evidence and dependency links for use during analysis.
- What it means
- An order whose status is completed.
- Source evidence
fct_orders.sql · line 50- Depends on
stg_orders.status
- Affected data
FCT_ORDERS.ORDER_ID- Source evidence
fct_orders.sql · line 49- Affects
- Completed order
New evidence is reconciled with existing records
Sienna compares new evidence with existing records, adds source and dependency links, and surfaces conflicting meanings for review. The reconciled records are then available anywhere Sienna uses the memory graph.
Source evidence
fct_orders.sqlCompleted-order filterorders.ymlOrder ID constraintref('stg_orders')Upstream dependency
Reconcile with memory graph
- Match existing records
- Attach evidence and dependencies
- Keep conflicting meanings separate
Connected records
STG_ORDERS.STATUS