Kepler planets (including super-Earths and sub-Neptunes) are likely formed before the gaseous proto-planetary disks have dissipated. Together with gas giants, we call these generation-I planets, to differentiate them from planets that form after disk dispersal (generation-II planets, e.g., terrestrial planets in the Solar system). If the metal content in these disks resembles that in the host stars, one naively expects Kepler planets to occur more frequently, and to be more massive, around metal-rich stars. Contrary to these expectations, we find that the radii of Kepler planets (a proxy for mass) are independent of host metallicity, and their occurrence rate rises only weakly with metallicity. The latter trend is further flattened when the influence of close binaries is accounted for. We interpret the first result as that the mass of a Kepler planet is regulated by a yet unknown process, as first suggested by \citet{Wu2019}. We explain the second result using a simple model, wherein the masses of proto-planetary disks have a much larger spread than the spread in stellar metallicity, and disks that contain more than $\sim 30$ Earth masses of total solid can form Kepler planets. Hosts for these planets, as a result, are only mildly more metal-rich than average. In contrast, the formation of a giant planet requires some $5$ times more solid. Their hosts, which also harbour Kepler planets, are significantly more metal-rich. This model also predicts that stars more metal-poor than half-solar should rarely host any gen-I planets.