Bug Report: Stockfish castles by moving king 3 squares after king moved

lichess.org/editor?fen=rnbk1b1r/ppp2ppp/5n2/4p3/4P3/2N5/PPP2PPP/R1B1KBNR_w_KQkq_-_0_1

Try playing as white from this position. When you threaten to take the f7 pawn with a Bc4 move, stockfish will castle by jumping his king 3 squares from the d8 all the way to g8....

To reproduce reliably, start with Nf3 as white, then Bc4, then Ng5. Once this f7 pawn is thoroughly targeted, stockfish will resort to cheating...

I think that since you ticked the boxes about castling, stockfish thinks black can castle either side independently from where the black king is. Try generating the same position without black's castling powers, or without one side, and see if it changes anything. I think it will.

Here, (@) Deadban was right, Stockfish castles because the checkbox of black casting is ticked.
When I cancel it,

That's it! So Stockfish is just a dummy playing according to your choice.

Could this by a Chess960 position, you sneaky guy?

In chess 960 the king never starts on a1 or h1. So the white king must have moved already.

@butterbrain said in #1:
> lichess.org/editor?fen=rnbk1b1r/ppp2ppp/5n2/4p3/4P3/2N5/PPP2PPP/R1B1KBNR_w_KQkq_-_0_1
>
> Try playing as white from this position. When you threaten to take the f7 pawn with a Bc4 move, stockfish will castle by jumping his king 3 squares from the d8 all the way to g8....

You give castling rights explicitly and wonder when it gets castled?