Let $C$ be a smooth projective curve defined over a local field $K/\mathbb{Q}_p$. Denote by $\text{Aut}(C)$ the geometric automorphism group of $C$, which consist of isomorphisms of $C\times_{\text{Spec}K}\text{Spec}\overline{K}$ with itself.
In Stoll's paper on rational points on twisting families of curves, see: paper, in Proposition 4.1, Stoll shows that if $\Gamma\le \text{Aut}(C)$ is a finite $K$-defined subgroup of the automorphism group of $C$, with $v(\#\Gamma) = 0$, then $I_K$ acts trivially on $\Gamma$.
My issue is this: I don't understand the statement. If $\Gamma$ consists of $K$-defined automorphisms of $C$, doesn't it mean that each $\gamma\in\Gamma$ is an isomorphism: $$\gamma: C/K\longrightarrow C/K,$$ and therefore, the algebraic equations defining $\gamma$ all have coefficients in $K$, which imply that $G_K$ acts on them trivially?