Guides

Build an inventory integration

Turn paginated inventory responses into a durable application model.

On this page

Build a snapshot only after every requested page succeeds. Keep the currently published snapshot separate from a refresh that is still in progress.

Prerequisites

one working request from Make your first request;
a server-side key and explicit request deadline;
a storage model that separates assets, descriptions, retrieval state, and timestamps;
the canonical Inventory schema.

Integration flow

Validate the Steam ID, application, and context.
Retrieve every page with a bounded loop.
Join assets to descriptions using documented identifiers.
Persist a retrieval timestamp and request ID.
Replace the prior snapshot atomically only after a complete successful read.

Keep private, unavailable, partial, and empty inventories as different states.

Treat each returned cursor as opaque and pass it unchanged. Stop when the response says no page remains, when your overall deadline expires, or when a defensive page limit is reached. Deduplicate assets by their stable identity while assembling the candidate snapshot.

Description records can be shared by several assets. Join them using the documented identifiers, not localized names. Preserve optional description fields as optional instead of manufacturing defaults.

Handle incomplete reads

Outcome	Recommended application state
Complete read with zero assets	Successful empty snapshot
Private inventory	Inaccessible; retain any last successful snapshot separately
Validation or authentication error	Configuration failure; correct before retrying
Rate limit or transient upstream failure	Refresh failed; use bounded retry or a declared stale policy
Deadline or page limit reached	Partial candidate; do not replace the published snapshot

Security and operations

Keep raw inventory data out of analytics and error reports unless a reviewed privacy policy requires it. Log the request ID, operation, status, duration, page count, and publish decision without logging credentials or complete response bodies.

Coordinate refreshes for the same inventory so workers do not duplicate a paginated read. Continue with Synchronize inventory data, Caching, and Inventory errors.

Integration flow

Validate the Steam ID, application, and context.

Retrieve every page with a bounded loop.

Join assets to descriptions using documented identifiers.

Persist a retrieval timestamp and request ID.

Replace the prior snapshot atomically only after a complete successful read.

Keep private, unavailable, partial, and empty inventories as different states.

Assemble pages safely

Handle incomplete reads

Outcome	Recommended application state
Complete read with zero assets	Successful empty snapshot
Private inventory	Inaccessible; retain any last successful snapshot separately
Validation or authentication error	Configuration failure; correct before retrying
Rate limit or transient upstream failure	Refresh failed; use bounded retry or a declared stale policy
Deadline or page limit reached	Partial candidate; do not replace the published snapshot

Security and operations

Coordinate refreshes for the same inventory so workers do not duplicate a paginated read. Continue with Synchronize inventory data, Caching, and Inventory errors.