Black loses game with 0 inaccuracies

There's something wrong with the analysis. Black's 12th move loses 3 pawn points and his 14th move loses 7 pawn points. Therefore both moves usually would qualify as a blunder.

Its a Bug

#1
This problem was reported yesterday here:
lichess.org/forum/lichess-feedback/move-seems-to-meet-definition-of-inaccuracy-or-worse-but-not-evaluated-as-such#2

Definitely a bug.

Lol. The computer may say 0 inaccuracies, but man, there is a massive developing disparity early on and a strong attack. You are 2600, you should see that white is in a very commanding position and black has nothing no show for the extra pawn and eventually got mated.

Either, the analysis is wrong or you just solved chess in this line where despite black making the best moves white still wins. Id say the analysis is wrong.

another example

@Notrashopenings0 said in #6:
> another example
I assume you are talking about 10...Ng5 being -14.0. However, that is the horizon effect. The depth Lichess gave Stockfish was not deep enough for Stockfish to see that White has a 7 move forced mate!