@srishakatux right, the tool will detect if user published a translation. In order to test the flow, I would suggest logging in with a production wikipedia (e.g. https://yo.wikipedia.org) and start CX there. You can publish your translation to the user namespace, so that you can test the entire flow.

Thank you @srishakatux for your comments! Addressing them below:

Thank you for your comments @srishakatux! I'll address them below.

First thing that comes to my mind would be to match infoboxes to Cirrus search topics (e.g. Infobox person --> Biography, Infobox settlement -> Geography, etc), and then find articles in these topics, filter them based on whether they have an infobox and suggest the ones without infobox. I'm not sure whether this will work well for small wikis, though, as I noticed that a lot of the articles do not belong to any topic according to the "Get articletopic outlink prediction" tool.

@matmarex yes, this can be considered done, thanks for resolving the task!

The implementation of the feature as currently described in this task is simple and straightforward, given that we already have a proof of concept regarding installing gadgets from external sources (e.g. "en" wiki). I'm assigning 4 story points for the time being.

This task specifically targets the HotCat gadget. Supporting additional gadgets does not appear to be a requirement, and extending the list would likely require case-by-case consideration and implementation, as the user journey is tailored to HotCat and may differ significantly for other gadgets.

Setting the design details aside, this task requires the ContentTranslation extension to be installed on the current wiki, which is not the default. Assuming CX is installed, the task is simple and straightforward.

Following the user journey from an administrator’s perspective, the process feels cumbersome and potentially confusing. The proposed workflow in the interactive prototype requires switching back and forth between the StarterKit dashboard and Content Translation. It also presents instructions for novice translators within the StarterKit app, but these cannot be enforced there. The same issue applies to publishing, as administrators can easily publish translations without following the provided instructions.

@GMikesell-WMF the continue_offset indicates the number of page collection articles that have been processed, until the requested (count) amount of recommendations is found. For this reason, it cannot be used to skip any number of results, like you try to do in your example. The continue_offset parameter is intended to be used in the following way:

The error still occurs (121 occurrences last week).