One of the downsides of White's plan seems to be that the knights become entangled on the Queenside.

And this is the reason for the stockfish evaluation. White has Nd5 here, mixing it up.

Supporting the c4 pawn and intending Q side castling

Which white stops.

I considered Nge7 here followed by castling but thought it would be better to break the potential pin to d8 now. I preferred to castle long as the rook would be active on d8 immediately and the the King would be able to defend against Nb5 should white decide to play it.

I was half expecting Bf4 which may have been more active.

Keeping tempo as the rook can now recapture should white exchange.

Following endgame advice by keeping the king in the centre rather than castling.

With the idea of going to d3 if the opportunity presented itself and freeing the c pawn to start pushing on the Queenside with the extra pawns. The computer prefers Nb4.

If f4 then Bg4+ is a skewer. This keeps the knight and bishop out of g4.