GET https://graph.microsoft.com/beta/copilot/admin/catalog/packages/{id} returns radnom 424 response

Question

When calling Microsoft Graph Copilot catalog APIs, we first call the list endpoint

(GET /beta/copilot/admin/catalog/packages) to get package IDs, and then call the detail endpoint for each package

(GET /beta/copilot/admin/catalog/packages/{packageId}?$select=id,shortDescription,longDescription, categories, sensitivity, allowedUsersAndGroups, acquireUsersAndGroups).

Why do some detail calls return 424 Failed Dependency for package? It's not always the same packages; it's completely random.

Answer 1

Hello Marko Saric,
A 424 Failed Dependency error on GET /beta/copilot/admin/catalog/packages/{id} indicates that the catalog service relies on downstream services (such as Entra for allowedUsersAndGroups / acquireUsersAndGroups, the sensitivity-label service, or Teams/AppSource metadata) to build the detailed response, and one of these dependencies failed or timed out for that specific request. The list endpoint uses a cached index and is not impacted, while the detail endpoint queries each package individually, which is why failures can seem random and may change between runs.

These issues are usually temporary and should be retried:

1. Retry the failed IDs using exponential backoff with jitter (3–5 tries), and respect any Retry-After headers.
2. Simplify $select by removing allowedUsersAndGroups, acquireUsersAndGroups, and sensitivity fields to reduce dependency on directory and label services, which lowers the chance of 424 errors. You can fetch these fields separately if needed.
3. Limit concurrency (about 4–8 requests at a time) or use small $batch requests to avoid overwhelming downstream services.
4. If a particular package consistently fails, collect the request-id or client-request-id headers, as this information is useful for support cases.

If you found the information above helpful, I would appreciate it if you could share your feedback.
Your feedback is important to us. Please rate us: 🤩 Excellent 🙂 Good 😐 Average 🙁 Needs Improvement 😠 Poor