Atmospheric water harvest using porous materials is an attractive solution to the water crisis. The inflection point (α) in the water adsorption isotherm is one of the crucial criteria to screen porous materials for their water-harvesting properties. Here, we propose a simulation protocol that can help predict α in covalent organic frameworks (COFs) via an implementation of the flat histogram atomistic Monte Carlo simulation method, NVT + W, with interpolation. The step positions for six COFs possessing diverse linkages and varied hydrophilicity as calculated from the simulations matched well with those from experiments. The simulations revealed two distinct types of pore-filling mechanisms in these COFs at water loadings before the inflection point that depended on both the binding strength and the number of hydrophilic binding sites. The work paves the way to use these structural attributes as descriptors which can be deployed for the large-scale screening of COFs as potential water harvesters.