Attribute definitions for products, customers, and images now use attributeDefinitions(entityType:), attributesAddDefinition, and attributesUpdateDefinition. The mutations take an entityType, change one definition per call, and return the definition directly. They do not return result, status, message, or processId. Existing permissions still apply.

Use showInList instead of showInProductList, and remove brand from definition selections. Omitted settings keep their current values. Definition names must be unique within a company and entity type, and addProductAttributes rejects duplicate definitions in one request.

The warehouseExternal query and Warehouse.externalData are also deprecated. Read buildingIntegrationConnections on Warehouse or Integration instead. Create takes buildingId and integrationId; update and remove take connectionId. A building can have only one active connection to an integration.

onInvoice now returns a number, including zero when the order has no accounting documents. It no longer uses null to mean unknown. Fulfillment item quantities use the item’s product units; sales item quantities count complete bundles after pooling components. Partial bundles are not rounded up.

Removing a sales line with shipment records now preserves the line at quantity zero. Clients must not assume that a removed line disappears from the response.

fulfillWithComponents takes AUTO, YES, or NO. It replaces decompositionStrategy on products and the short-lived fulfillWith. AUTO keeps the product line when its stock covers the entire line and otherwise uses components for the whole line. YES always creates component lines. NO keeps the product line. New aliases and bundles default to AUTO, and other product types default to NO. Manual decomposition works with every value. Existing products were reset by product type when this shipped.

inventoryIncludesComponents replaces includeComponentsInAvailableToSell. It is a Boolean with the same meaning, and it defaults to true for aliases and bundles and false for other product types.

Fulfillment rules no longer have decompositionStrategy, so omit it. Endless applies the product setting when it creates a fulfillment, not as a delayed step or when a rule is assigned. It skips the setting for fulfillments whose processInEndless is false.

Queries and exports using these filters can return fewer orders for the same date range. Recheck reports that relied on those shipments being included. shipmentCreatedAt filters are unchanged.

Advancing a shipment to In Transit, Delivered, or Received records the corresponding date. Pass statusDate when recording a historical transition so it uses the intended date.

SupplySupplier.address now returns the ship-from address. Callers that need the remit-to address must select it explicitly. Dropship cost tiers now require a minimum order quantity of 1.

The JSON argument still works and stores a location without restricting allocation. Migrating to typed items changes that behavior: the location becomes a binding restriction, and insufficient stock there returns an error. Do not send both arguments in one request.

An empty string clears text such as contact fields, address line 2, the gift message, and the requested shipping method. An explicit null also clears the expected and shipping dates, the gift message, and the requested shipping method. An empty orderItems list or orderMetadata object clears its contents.

shipToSameAsBillTo on saveDraftOrder no longer defaults to true. Omit it to keep the current setting. Editing the shipping address no longer removes the billing address, and the reverse.

Deleting by IDs finishes in the request and returns deletedCount. Deleting by filters runs in the background across all pages and returns a processId. Filtered deletion skips records changed after the request. The mutation requires the new echo:delete_data permission, which company operators and operations integrations do not hold. Records created automatically are never deleted.

Fulfillments already allocated or in progress keep the earlier structure until they close, and their components do not appear in either representation. Integrations must account for that transition. Warehouse integrations now receive component SKUs with allocated unit prices that preserve the bundle’s total value.

The value remains the customer’s original shipping choice. It is distinct from requestedCarrierServiceName, which identifies the carrier service selected to fulfill the order.

The integration capability is renamed from forward shipments to send_shipments. Calling the mutation again no longer clears earlier delivery receipts. It retries outstanding work and skips shipments the integration already accepted.

Export process polling and subscriptions now require the requesting identity to own the process, including API clients. Use the same identity that requested the export to retrieve its status and download metadata.

Other filtering and sorting behavior is unchanged.

Automatic delivery still skips unchanged documents. The optional fulfillmentId limits delivery to one fulfillment; omitting it keeps order-level scope.

Authentication can return PASSWORD_EXPIRED or ACCOUNT_LOCKED instead of a generic login failure. A password reset unlocks a locked account. Password-reset token failures now return HTTP 401. Update clients that branch on the previous error codes or status codes.

When migrating, pass exactly one of demand, shipments, inventory, or inventoryChanges under input.export, with filters nested inside that selection. Missing, null, or multiple selections are rejected. An empty object for the selected dataset exports without filters.

An update that omits the shipping date or sends null preserves the existing date. Existing shipment dates are not changed automatically.

ECHO shipment reporting uses shippedAt when set, and the recording date when it is null, applied consistently to aggregation windows, daily buckets, and backfills. Clearing or adding a shipping date rebuilds both affected days without double-counting.