Zebra is a design we experimented with but later dropped T259240. As @Michael points out, there's more recent design work to improve the notification bar, and given that the zebra design never made it to production, I think the task should be closed as invalid.

is the Spanish version an old version of the code? How is it that it's only showing recommendation features?

@SToyofuku-WMF thanks for raising that. @ovasileva In terms of what experiments we're showing, we've disabled the recommendations experiment for all wikis and enabled the summarization experiment only on enwiki, so eswiki is not getting any experiments right now.

Given Jennifer's metrics spec and measurement plan, I think this task can be resolved.

@mpopov thanks for mentioning that. We've split out the schema related work into T380926 and we're taken the experiments fragment into account.

After consulting with Metrics Platform regarding the schema, based on the following specification, it looks like we should be able to use the analytics/product_metrics/web/base schema for this experiment. The ui actions can be sent to the action property and session id has an equivalent session_id field.

Looks good to me. Thank you @Jdlrobson for the very thorough plan!
I think it's great that we're repurposing existing UI code for this experiment from RelatedArticles.

Although we haven't completed all enabling technical work in this task, we have at least identified and made a plan for it.
@KSarabia-WMF detailed a roadmap for this A/B test instrumentation work in https://phabricator.wikimedia.org/T378094#10336836
In regards to the technical enablers specifically,

I just submitted the build (v1.0.7) to the Chrome store Thursday morning, and it was approved shortly after!
Unfortunately, I noticed a typo in the scheduled activation date (set to November 26th, instead of November 21st) , so I uploaded the fix as v1.0.8.

I'm attaching the zip file of the latest build. If this build passes QA, then this zip file will be uploaded to the chrome store as version 1.0.7

The following is a build with the latest instrumentation and summary experiment enabled for QA purposes:

I'm attaching a build of the extension for QA purposes.
This build might show two experiments running after the terms of service is accepted (but that can be ignored for this task).

The filters have been applied to the summary experiment in the following patch:
https://gitlab.wikimedia.org/repos/web/web-experiments-extension/-/commit/8db793ba76e371d583c819a472c5fbb36605500b

Oh, looks like I just needed to update my version of Codex via npm update looks better in the following build.

@JScherer-WMF here's a build with the latest changes. Looks like the verbage and button icons have been fixed. I looked into the infoChip color and I'm not sure why it doesn't get the yellow background, we're using the same code that's described in the docs:

I'm playing around with this and it still seems like some articles that should have summaries based on our article list still aren't getting a simple summary. e.g. https://en.wikipedia.org/wiki/2003_invasion_of_Iraq

I'm playing around with this and it still seems like some articles that should have summaries based on our article list still aren't getting a simple summary. e.g. https://en.wikipedia.org/wiki/2003_invasion_of_Iraq

task pointed 0 since it's intended as a story/epic specification

Given this is affects MediaWiki core and ResourceLoader, we think some input from the core platform team would be valuable.

I'm posting some generated code below that might accomplish this behaviour.
It keeps track of the last visited domain, current domain, stores the impression in the browser extension session storage, so when someone visits Wikipedia for the first time during a browser session, they'll be shown the popup once during that browser session.

The hypothesis we want to prove with this test is: