I've never really been a fan of proofs without words; they've always felt way too slippery to me, for lack of a better term. A well worded proof with nice explanatory diagrams hits the spot for me instead.
For the more geometry-based ones where you have move triangles around and so, it's often not obvious to me that two angles that look the same really always are the same, and that things that add up to rectangle do so reliably, independently of the actual angles used in the examples.
I guess in these cases, a more parameterized, interactive version would work better, where you can use sliders to adjust some of the angles and lengths used. That should make it much more obvious that it's not just an artifact of particular angles used in an example.
On the linked page, a lot of the proofs are essentially proofs by induction that stop at some (pretty small) n. Maybe there’s a way to make it rigorous by visually showing the induction step that proves n+1 given n, but if there is, it’s not shown.
This can be great for building intuition for a statement known to be true by other means, but I wouldn’t consider them to be proofs.
This problem exists not only for visual proofs, but for standard written ones too.
The problem is that it is far too easy to convince someone of something which is not true via visual means.