Requests & pagination

Base URL and headers

https://{store primary domain}{path}
Authorization: Bearer ciqra_at_…
Content-Type: application/json        (on requests with a body)

Paths have no version prefix and no store id. The store is the domain.

JSON conventions

• Property names are camelCase — productVariantId, updatedAt.
• Enums are strings — "status": "Active", never a number. Each enum’s allowed values are listed on the schema in the reference.
• Ids are UUIDs; timestamps are ISO 8601 with offset (2026-09-22T10:15:00+00:00).
• Unknown properties in your request body are ignored — they do not cause an error, so a typo silently does nothing. Check the schema.

Pagination

List routes use offset paging with skip and take:

GET /apps/catalog/products?skip=100&take=100
→ { "total": 1284, "skip": 100, "take": 100, "items": [ … ] }

Route	Default take	Maximum take
GET /apps/catalog/products	50	100
GET /apps/catalog/collections/{id}/products	50	250
GET /apps/inventory/levels	50	250
GET /apps/catalog/collections	—	not paginated

Values outside the range are clamped, not rejected: take=1000 returns 100 rows and says "take": 100. Read take from the response rather than assuming yours was used. Stop when skip + take >= total.

Offset paging is not a snapshot: rows added or removed while you page can shift. For a full sync, page by the stable order the route documents and reconcile by id.

Omitted versus null

Write routes distinguish three states for an optional field:

You send	Meaning
field absent	leave it unchanged
"field": null	clear it (where clearing is allowed)
"field": value	set it

This matters for partial updates: sending a full object with nulls for fields you did not intend to touch clears them. Send only the fields you are changing. Each field’s exact behaviour is on the route’s page — for example bulk product updates.