You need the direct and indirect effects and their p-values. In AMOS you can see the indirect effects in a tab (when you choose it in outputs). You also have to calculate your model twice (3, 4... times depending on the amount of mediators). You present results of your model without a mediator and with the mediator. And you look at the differences, first, in significance level and, second, in the regression weight.
Here are two good examples:
https://youtu.be/0artfnxyF_A
https://youtu.be/ICnh3s2FG14
You can also look in our paper how we described the mediation effects: