Can write $Y_2 = \sqrt{1-\rho^2}Z + \rho Y_1$ where $Z$ is an independent $N(0,1).$ Then $$Y_2^3 = (1-\rho^2)^{3/2}Z^3 + 3 (1-\rho^2)\rho Z^2Y_1 +3\sqrt{1-\rho^2}\rho^2 ZY_1^2 + \rho^3Y_1^3.$$ Taking the expected value conditional on $Y_1^2$ gives zero. This is because $E(Y_1^n|Y_1^2) = 0$ for any odd $n$, and the terms with even powers of Y's all have odd powers of Z.
Then we square $Y_2^3$ and take the conditional EV to get the variance. Here I won't write all the terms, only the ones that contribute to the expectation:
$$
Y_2^6 = (1-\rho^2)^3Z^6 + 9(1-\rho^2)^2\rho^2Z^4Y_1^2 + 9(1-\rho^2)\rho^4Z^2Y_1^4+\rho^6Y_1^6+ 6(1-\rho^2)^2\rho^2Z^4Y_1^2
+6(1-\rho^2)\rho^4Z^2Y_1^4
$$
where the last two terms are is the only indirect terms from the square to contribute. Since $E(Y_1^n|Y_1^2) = Y_1^n$ for Y even, we just have
$$
E(Y_2^6|Y_1^2) = (1-\rho^2)^3E(Z^6) + 15(1-\rho^2)^2\rho^2E(Z^4)Y_1^2 + 15(1-\rho^2)\rho^4E(Z^2)Y_1^4+\rho^6Y_1^6
$$
Since $E(Z^6)=15,$ $E(Z^4)=3,$ and $E(Z^2)=1,$
$$
Var(Y^3_2|Y_1^2)= E(Y_2^6|Y_1^2) = 15(1-\rho^2)^3 + 45(1-\rho^2)^2\rho^2Y_1^2 + 15(1-\rho^2)\rho^4Y_1^4+\rho^6Y_1^6
$$