Development and integrations

Use metafields to extend a resource with additional values. Consider a metaobject for structured content with multiple fields or reusable references, then define how the theme and APIs read it.

• Specify types, units, language and owners.
• Test references and missing values with real records.

Admin data is not automatically public storefront content.

Assign an owner to the source record and reference the applicable information from product, collection, support and summaries. List affected surfaces and compare visible content, APIs and markup after a change.

• Connect models, markets and policy versions.
• Trace one change through all consumers.

Shared data must preserve model, market and version exceptions.

Specify data inputs, merchant controls, missing-value behavior, languages and mobile presentation. Test real products and dynamic sources before extending reuse.

• List fields, editing boundaries and fallbacks.
• Compare output on preview and public routes.

Template reuse does not ensure every product has the required data.

Agree which system owns each product, inventory, price and order field, then verify the caller, allowed scope and write direction. A successful response still needs a resource-level and business-result check.

• Record field ownership, update direction, exception owners and audit evidence.
• Test permitted operations, denied operations and expired access.

Browser identity, app identity and Admin API access are different concerns. Use an authorized test environment for validation.

Assign source systems, stable identifiers and status meaning per product, inventory, customer and order. Define directions, conflicts and reconciliation, retaining transport and business completion separately.

• Trace one order across the connected systems.
• Define handling for delay, duplication and manual correction.

Different resources can have different owners; a later write should not silently resolve an unexplained conflict.

Verify the source and identify repeated deliveries. Keep business operations idempotent beyond the receipt counter, check current resource state when ordering matters, and reconcile to find missed processing.

• Record resource IDs, delivery IDs, processing state and failures.
• Check whether the business action already happened before replaying it.

Receiving a delivery, completing a task and reaching a consistent business result are separate states. Define recovery with the other system.

Record actual framework, API, dependency and hosting versions. Verify page output, caching, product state and the buying journey, and keep the release, recovery target and order-handling plan together.

• Check initial HTML, canonical URLs and product state by market.
• Rehearse the switch in preview and record how data will be reconciled after recovery.

Commands, capabilities and migration paths differ across Hydrogen versions and hosting environments. Use documentation for the deployed version.

Hand over versions, configuration and access ownership, release recovery and known issues. Have the receiving team perform a controlled check, recording code delivery, deployment and business acceptance separately.

• Record environments, interface versions and change owners.
• Check that monitoring, backups and recovery routes are usable.

Documentation and demonstrations do not establish that a production recovery exercise was completed.