And again...
https://www.chess.com/analysis/game/daily/399670935
I was rather more surprised by this one. Firstly, that it doesn't disapprove of 4...Qh4. Secondly, Black's moves up to 12...Be6 are all 'theory' (or at least given in my book: https://everymanchess.com/collections/openings-books/products/a-disreputable-opening-repertoire). And thirdly, 13...Nxe5 is hardly "brilliant". I'd expect to see that in a bullet game.
But anyway... 
https://www.chess.com/analysis/game/daily/393888511
The chess.com analysis engine assesses my play in this game as 100% accurate, which is a first for me. Okay, I've only really played one move (12 b3). The first eleven had all been seen before (G.Van der Stricht-J.Van der Fraenen, Belgian League 2017); 12 b3 (attacking the defender of d5) is pretty obvious; and the rest followed automatically. But still... 100%