What would you base an argument on that claims that observation is sufficient to establish the conditions of the cosmos.
I think you are asking, how do we know about the known world's known things? If so: Falsifiabilty. Prediction. The current hard-won
method of science, in short.
Epistemological problems are not settled cases, and new ones will arise as we -
if we - continue to more robustly model the world.
It's no small thing to know that stairs are hazardous terrain, and often learned first through pain, and then through trepidation, but learn it most of us do. We learn that we must
walk differently, which has the effect of confirming that the more usual modes of walking are generally sufficient, but each of us must still
learn to do both, and having learned, can know with some certainty (predictability) that walking will produce expected outcomes.
It is this - learning to learn - that lets most every person who matures out of early childhood know that
what they know about the rules of existence (the feeling of gravity, heat, wetness, hunger, thirst, depth, distance, time's passage and the variability of its felt pace, etc) are largely reliable. This is the foundation of the knowable world, that
it behaves in expected ways.
Not randomly, and quite curiously, the claims about the transcendent or supernatural rest precisely
on them not.