This is Remark 7.36 of Görtz/Wedhorn. We have a scheme $X$ and a quasi-coherent $\mathcal{O}_X$-module of finite type $\mathscr{F}$, and we define $\operatorname{Ann}(\mathscr{F})$ as the kernel of the canonical homomorphism of $\mathcal{O}_X$ into the endomorphism sheaf of $\mathscr{F}$ over $\mathcal{O}_X$. Then we assume that $\mathscr{I}$ is a quasi-coherent ideal of $\mathcal{O}_X$, and we thus get an induced closed subscheme $V(\mathscr{I})$ of $X$, whose inclusion into $X$ we denote by $i$. If we assume further that $\mathscr{I}\subseteq \operatorname{Ann}(\mathscr{F})$, then $\mathscr{F}$ is naturally an $\mathcal{O}_X/\mathscr{I}$-module. The authors then say that $\textbf{this}$ shows that the canonical morphism coming from adjuntion $\mathscr{F}\rightarrow i_*(i^*\mathscr{F})$ is an isomorphism, which I don't really get at the moment.
I think that since $i$ is an inclusion of a closed subscheme, we should get isomorphism on stalks, but I don't see how this uses the argument $\textbf{this}$, and also I think my thought is invalid, because otherwise for every closed subscheme we would have this isomorphism. I would be grateful if someone could give me just a little hint.
Hint: work affine locally and describe the map $\mathscr{F}\to i_*i^*\mathscr{F}$ explicitly. What is it? How can you use the fact that $\mathscr{I}\subset \operatorname{Ann}(\mathscr{F})$ with this description?
Full solution: