Here's my attempt. As first suggested by JG89, first show that [tex]\frac{2 - (x_n - 1)^2}{3} < 1[/tex] holds. This will show that [tex]x_n[/tex] is bounded above by 1 for all [tex]n \in \mathbb{N}[/tex].
Next, we need to show that [tex]x_n \geq 0[/tex] for all [tex]n \in \mathbb{N}[/tex]. For contradiction, suppose there exists some [tex]N+1 \in \mathbb{N}[/tex] such that [tex]x_{N+1} < 0[/tex]. So we have that [tex]0 < \frac{1 + 2x_N - x_N^2}{3}[/tex], implying that [tex]0 < 2 - (x_N - 1)^2[/tex]. Do some further rearranging, get [tex]2 > (x_N - 1)^2 \geq 0[/tex] and finally, get [tex]\sqrt{2} + 1 > x_N \geq 1[/tex] --- contradiction to the prior where we had shown that [tex]x_n < 1[/tex] for all [tex]n \in \mathbb{N}[/tex].
Next, we show that the sequence [tex](x_n)[/tex] is monotonically increasing. To show that [tex]x_{n+1} \geq x_n[/tex] for all [tex]n[/tex] is equivalent to showing [tex]x_{n+1} - x_n \geq 0[/tex]. So, consider that [tex]x_{n+1} - x_n = \frac{1 + 2x_n - x_n^2}{3} - x_n = \frac{1 + 2x_n - x_n^2 - 3x_n}{3} = \frac{1 - x_n - x_n^2}{3} \geq \frac{1 - x_n}{3} > 0[/tex] and the last inequality holds because we had previously shown that [tex]0 \leq x_n < 1[/tex] holds.
Hence, by the monotone convergence theorem, the sequence [tex](x_n)[/tex] converges.
In fact, we can do better than that --- we will find the limit. Note that if [tex]\lim x_n = x[/tex], then [tex]\lim x_{n+1} = x[/tex] (this is easy to prove). Hence, from the original recurrence relation, we can have [tex]x^2 = \frac{1 + 2x - x^2}{3}[/tex]. Using the typical quadratic formula, we solve to get [tex]x = \frac{-1 \pm \sqrt{5}}{2}[/tex] and we reject the negative root since we had shown that [tex]x_n \geq 0[/tex], which implies [tex]\lim x_n \geq 0[/tex]. Hence, [tex]\lim x_n = (-1 + \sqrt{5})/2[/tex].