@Tgr: Hey, Are you familiar with the ORES extension? What do you think if we enable this functionality via the extension?

This looks a lot like deferred changes, recently approved on enwiki, see T118696 for the technical implementation for bots and AbuseFilter, T150593 for ORES, and T150594 for the implementation request.

Related: T58526: Google uses unsighted version of the page

The Scoring Platform team is not likely to work on this task.

As a general note, we tend to see ORES as a tool to aid human decision-making, rather than something that replaces human judgment. A workflow you might consider is that you could look for articles with an unusually high rate of damaging edits, and then use this data to recommend to a human administrator that Flagged Revisions be applied. This way you get the rapid analysis that ORES provides but ultimately a human is in charge.

(This is just a perspective; you can pursue whichever implementation works for you.)

Just linking related ticket T218278 . There are in comments links to stabilizerbot which is bot implementation for this and to Cirdans kokolores repo for the same idea.

In T132901#5080667, @Harej wrote:

As a general note, we tend to see ORES as a tool to aid human decision-making, rather than something that replaces human judgment.

This is not about replacing human judgement (are you thinking of T165848: Decrease FlaggedRevs backlog by using ORES predictions models maybe?) but about using ORES to delay potentially harmful actions until human judgement becomes available, since patrollers can't be expected to review edits in real-time, especially on smaller wikis. Once human judgement is available you either approve the page or revert it, in either case FlaggedRevs is not necessary anymore.