Skip to main content
Use this page to check whether an endpoint behaves the way the API reference describes, before you build on it. Every Works and Differs row was executed against staging.
Every Works and Differs row was executed against staging with a partner sandbox credential — every single-resource read, and the full patient journey including writes. The By arrangement row is the prescription webhook feed, which no partner call can exercise. The complete list of published operations (109) is at the end of this page.

Legend

The journey at a glance

One row per call in the order you build them. The verdicts are the same ones used in the stage tables below; this table exists so the whole journey fits one screen in either theme.

Stage 1–3 · Patient and intake

POST /v0.1/questionnaire-responses performs almost no validation. Validate on your side. Each of the following returns 201:
  • subject pointing at a patient id that does not exist — and an encounter was created for it
  • a questionnaire_id that does not exist
  • an encounter_id belonging to a different patient — the response was attached to it
  • omitting an answer the questionnaire marks required: true
  • an answer keyed to a link_id the questionnaire does not contain
Also: status: "in-progress", or omitting subject, returns 201 and silently creates no encounter. Since subject is not in the reference’s required list, a strictly conformant client can produce orphaned intakes without ever seeing an error.Treat a 201 here as “accepted”, not “correct”. Verify ids before you send them, and read the response back.

Stage 4–6 · Commerce and order status

Clinical data

Record counts differ per tenant and are not shown here — see Known gaps and staging for how to check your own.

Two cross-cutting behaviours to design around

1. Unknown query parameters are handled inconsistently. Some endpoints reject an unrecognised parameter with 400; others accept it, ignore it, and return the unfiltered collection with 200. The ones that ignore it include products, appointments, appointment-types, configs, coupons, promotion-codes and billing-portals. A 200 does not confirm your filter was understood — verify the result looks filtered, especially before concluding a collection is empty or complete. 2. Response envelopes differ by endpoint. Three shapes are in use: {items, after?, request_id} for clinical collections (after only when there is a next page); {items, pagination, request_id} for the product catalogue; and {data, request_id} for single reads and writes. Do not write one shared pagination helper for the whole API. The product catalogue carries a pagination object, but catalogue paging is not implemented yet, so do not build a cursor loop on it. The reference also declares a top-level after for products; do not page on that either. It is not yet confirmed that one response always holds the whole catalogue: if products you expect are missing, send your OLH implementation contact the request_id.

Not part of the partner surface

Routes that are not in the published reference are not for partners. A probe of one may draw a 403, a 404 or even a 200; none of those makes it yours. Build only against what the reference documents, and treat anything else you find on the host as internal and liable to change without notice. A 403 complaining about the Authorization header rather than about scopes almost always means the method is not mapped on that path — check the reference for the correct verb before concluding an endpoint is closed to you. There is no partner /prices collection; read prices with GET /v0.1/products/{product_id}/prices or GET /v0.1/prices/{price_id}. Partners do not delete patients, questionnaire responses, consents or document references. Test data you create on staging is cleaned up by OpenLoop; ask your implementation contact.
There is no partner call that lists programs. Program membership is visible on products (program_code) instead — and note there is currently no way to look up a program’s intake questionnaire through the API. See Programs and identifiers.

Published operations

All 109 operations in the published reference, under /v0.1, grouped by resource. Each is published; the tables above say which have been exercised on staging and how they behave. Anything not listed here is not available to partners.