The move that cries out to be played is 2...O-O. Castling O-O-O isn't going to be possible anytime soon, and it is hard to see where the Black King will find shelter elsewhere. The problem of King safety is intertwined with the problems of developing the Queen and g-Rook. It's easy to imagine where the minor pieces will be developed, but not at all easy for the King and major pieces.
I had two problems with 2...O-O. The first problem was the attack 3.g4 and 4.h4. After a single move forward, the Pawns backed by their own Rook and Queen already present a real menace. The second problem was the development 3.b3 and 4.Bb2. The Bishops on the a1-h8 and b1-h7 diagonals, although not presenting any immediate threat, will be raking Black's King position for a long time to come.
I decided that it was better to delay the castling decision for as long as possible, preferably waiting until White has shown his own hand. I finally played 2...b6, a natural move that opposes the light-squared Bishop against the Queen on the a8-h1 diagonal.
One advantage to thinking deeply about castling O-O was not having to revisit the decision over the next few moves. White castled O-O on the 11th move and Black followed suit on the 12th, confident that the King would not be overwhelmed. Castling on the first or second move is always an attractive option in chess960, but that doesn't mean it's a good one.