The Limits of Agnosticism

I'm a big fan of agnosticism, in T.H. Huxley's original sense:

In matters of the intellect, follow your reason as far as it will take you,
without regard to any other consideration. And negatively: In matters of the
intellect, do not pretend that conclusions are certain which are not
demonstrated or demonstrable,

but today I'm having a problem with it.

Robert Anton Wilson, one of my favourite writers, got me into agnosticism with "model agnosticism", which says that since everything we know about "reality" is filtered through our sensory systems, any model which describes reality really describes our experience of it, and so all of them ought to be taken with a pinch of salt. The scientific model seems the best so far, because it's the most internally consistent and successfully predictive.

In other words, the scientific method provides the most rational explanation for things, which implies that one has already accepted reason as the best way of knowing.

Which is what I've realised about agnosticism: at the level of "how do we know things?" it's useless.

If agnosticism is the refusal to pass any judgement on things we don't have good enough evidence for, then we must first specify what sort of evidence is good evidence. Some people think that experimental data is good evidence. Others think that their personal experience is good evidence. It makes no sense to accept both types of evidence as good evidence, because they conflict - rationally, the fact that I have seen pixies does not imply that pixies exist. I might be mad.

So before we can begin thinking about whether we have good enough reasons for believing something, we have to choose whether we're rationalists or 'experientialists.' Which, I suppose, is why so many people, theists and atheists alike, think agnosticism is a weak option.

The agnostic answer, I suppose, would be that the question of how to know things is an unknowable thing, and so we shouldn't conclude either way. Which means that we can't know anything at all, because we have no basis for it.

So agnosticism boils down to either solipsism or rationalism?