Here: https://xtools.wmflabs.org/articleinfo/en.wikipedia.org/Sieve%20of%20Eratosthenes
XTools version: 3.1.23-d4e0364
The top edit by added text is actually vandalism:
Max. text added: 2007-03-28, 23:46 • 207.172.220.53 • 548,937
Top 10 by added text -- graph is affected as well.
I'm not sure there's a way to tell a "vandalism" edit with a "good edit," at least with the data XTools can see...
The one option might be to look at the next edit and previous edit and see if they match. But then that wouldn't catch any partial rollbacks.
See T179996 for revert detection, which is already in the pipeline.
The end result of this task would be to add a notation if the top edit by added text is reverted, adding a notation to the "Max. text added" line. I'll take this task and give it a stab.
T179996: Investigate using rev_sha1 to detect reverted edits is done, so we should have better revert detection now.
Made some tests, and actually that edit isn't counted as reverted. Probably because a few other edits happened before the revert.
(For reference, current status is that reverted edits are not shown at all. Also, the precise on-wiki page history url is https://en.wikipedia.org/w/index.php?title=Sieve_of_Eratosthenes&action=history&dir=prev&offset=20060913102251|75479087&limit=50.)
The issue is on what we call "reverted". Neither XTools nor MW consider that edit reverted. Seeing this is was reverted would require a) seeing that the revert a few edits downstream was one, and b) marking all edits between the identical SHA1s as reverted, which will require a change in our revert workflow.