Identify if there's system that could work for the app to be able to find, and surface a batch of uploaded videos from the Comms team. They are able to upload 30-100 videos in batches to Commons. The first one could cover an experiment period, and then they plan to upload periodically after that. Comms should be able to self-service.
Note any requirements that Comms must follow as they upload them (adding certain categories, for example)

Yes, a specific Category would be a sensible way to bucket these videos, and would make them easily searchable. The chosen category could be a new category or an existing one, such as Category:Wikipedia_videos. If we decide to create a new category, this would of course need to be done with transparency and buy-in from the community. The agreed-upon category would need to be "non-controversial", i.e. guaranteed to keep existing, since it will need to be hardcoded in the apps for searching.

Commons supports CirrusSearch similarly to other wikis, and gives us a way to search for media within a certain category, and also search by creation date. The search term would be something like: incategory:"<category name>" creationdate:>2026-07-01

Surface 4-5 new videos within the Community feed each day

We can surface however many videos were uploaded on that day. If the Comms team themselves could observe a natural cadence of uploading 4-5 videos per day, that would make it simplest for us.
Ideally we should avoid a scenario where 100 videos are uploaded on the same day, and then we need to "keep track" ourselves which ones we have shown.

Surface article title, article image, & short description of article(s) or pages mentioned in the video

Very important: Anything that is "mentioned in the video" must be written in the description (text) or the structured data of the uploaded file. (there's no way to "parse the video" itself to determine what it's talking about.)
Therefore we would need to agree very precisely on the formatting of the text description that must be written for each of these uploaded videos. At a minimum, the description would need to contain wikilinks to any Wikipedia articles that should be surfaced, so that we can parse the description and pull the links out of it. Something like: Articles mentioned: [[:en:Article1|Article 1]], [[:en:Article2|Article 2]], etc.

Surface license information

This is trivial, but important to note that any videos uploaded to Commons must be under a license that is legally compatible with Commons.

A couple of other practical considerations:

• We will need to check with SREs regarding the scale of new traffic that this will load onto Commons, and whether that scale will be manageable.
• Are we considering languages other than English? If so, we would need to agree on how to specify which language the video is "intended" for. Potentially this could be done by making sure to write a description (when uploading the video) in the appropriate language.

If the Comms team themselves could observe a natural cadence of uploading 4-5 videos per day, that would make it simplest for us.
Ideally we should avoid a scenario where 100 videos are uploaded on the same day, and then we need to "keep track" ourselves which ones we have shown.

They are indeed hoping to batch upload 50-100 videos at a time and I don't think daily uploads are in scope. Are there any other ways we could keep track on the commons side like unique IDs. or anything?

Very important: Anything that is "mentioned in the video" must be written in the description (text) or the structured data of the uploaded file. (there's no way to "parse the video" itself to determine what it's talking about.)

Possible: Lucas has the links to each relevant article in the doc

This is trivial, but important to note that any videos uploaded to Commons must be under a license that is legally compatible with Commons.

Confirmed with Liv, we have permission to upload these to commons with a compatible license.

We will need to check with SREs regarding the scale of new traffic that this will load onto Commons, and whether that scale will be manageable.

@Dbrant could you take this as an action step?

Are we considering languages other than English? If so, we would need to agree on how to specify which language the video is "intended" for. Potentially this could be done by making sure to write a description (when uploading the video) in the appropriate language.

No, just English

Instructions for Comms

• The description needs to contain the wikilink to any Wikipedia articles that should be surfaced, so that we can parse the description and pull the links out of it. >Something like: Articles mentioned: [[:en:Article1|Article 1]], [[:en:Article2|Article 2]], etc.
  • Example: description=[[:en:British Airways Flight 5390|British Airways Flight 5390]]
• All videos should have the same category attached to them, a category unique to these uploads that is unlikely to be used by others. Recommended to create a >subcategory under https://commons.wikimedia.org/wiki/Category:Wikipedia_videos (How to)

Are there any other ways we could keep track on the commons side like unique IDs. or anything?

This just prevents us from relying on Commons to tell us which videos were uploaded on a certain day, and makes it much more complicated for us to determine which files to show on which day.
If we ask Commons for a list of videos, and it gives us all 100 of them, how should we decide which ones to show first?

Liv said that they are timeless, so it does not matter to me.