neuro-shopify-graphql

Shopify GraphQL uses a calculated query cost bucket that refills at a plan-dependent restore rate. Check extensions.cost.throttleStatus in every response and implement exponential backoff when errors contain the THROTTLED code, retrying up to five times with capped delays.

Use cursor-based pagination with first and after arguments, reading pageInfo.hasNextPage and pageInfo.endCursor to fetch subsequent pages. The maximum first value is 250; for larger datasets, use bulk operations instead of looping queries.

Use the metafieldsSet mutation, which idempotently creates or updates metafields by ownerId, namespace, key, type, and value. It accepts a maximum of 25 metafields per call, so larger batches must be split into multiple requests.

Use bulkOperationRunQuery when exporting datasets larger than a few thousand records, since it avoids pagination and rate limit pressure. Bulk queries must have exactly one top-level field, omit first/last arguments, and use edges/node syntax, returning results as a JSONL file.

A single query cannot exceed 1,000 points regardless of plan tier, and deeply nested connections multiply costs quickly. Reduce first/last values, prefer nodes over edges, avoid nesting more than two connection levels, and split large queries into multiple requests.

Use the inventorySetQuantities mutation with the compareQuantity field to implement compare-and-set semantics. If the current quantity does not match compareQuantity, the mutation returns a userError instead of overwriting, preventing lost updates in multi-writer environments.