A domain enters the engine as one or more data sources — a connector plus a field mapping (ADR-006). Each source is ONE system's view. The intake form validates, previews the canonical rows, and runs the structural checks on them; cross-system checks activate when ≥2 sources run together. Editing is the approval-gated write path (ADR-005); this view is read-only.
▚
orders
2 source(s)
fields=amount, customer, order_ref, status
·● structural: clean (sample)
csv
path:/opt/control_desk/config/demo_orders_billing.csv
demo seed — orders_demo_billing (live only, overwritten on redeploy)
fields=amount, customer, order_ref, status
·● structural: clean (sample)
csv
path:/opt/control_desk/config/demo_orders_crm.csv
demo seed — orders_demo_crm (live only, overwritten on redeploy)
▚
products
1 source(s)
fields=name, price, sku, stock
·~50 rows
·● structural: clean (sample)
sample
n:50 · seed:42
seeded sample source — products as seen by system a (synthetic generator)