Visual-spatial abilities (VSA) are considered a building block of early numerical development. They are intuitively acquired in early childhood and differentiate in further development. However, when children enter school, there already are considerable individual differences in children’s visual-spatial and numerical abilities. To better understand this diversity, it is necessary to empirically evaluate the development as well as the latent structure of early VSA as proposed by the 2 by 2 taxonomy of Newcombe and Shipley (2015). In the present study, we report on a tablet-based assessment of VSA using the digital application (app) MaGrid in kindergarten children aged 4–6 years. We investigated whether the visual-spatial tasks implemented in MaGrid are sensitive to replicate previously observed age differences in VSA and thus a hierarchical development of VSA. Additionally, we evaluated whether the selected tasks conform to the taxonomy of VSA by Newcombe and Shipley (2015) applying a confirmatory factor analysis (CFA) approach. Our results indicated that the hierarchical development of VSA can be measured using MaGrid. Furthermore, the CFA substantiated the hypothesized factor structure of VSA in line with the dimensions proposed in the taxonomy of Newcombe and Shipley (2015). Taken together, the present results advance our knowledge to the (hierarchical) development as well as the latent structure of early VSA in kindergarten children.