The Hubble Space Telescope (HST) has been providing tremendous survey efficiency via its pure-parallel mode, by observing another field in parallel with the primary instrument in operation for the primary observation. In this study, we present a new archival project, SuperBoRG, which aims at compiling data taken in extragalactic parallel programs of HST with WFC3 in the past decade; including pure-parallel (BoRG, HIPPIES, and COS-GTO) and coordinated-parallel (CLASH and RELICS) programs. The total effective area reaches $sim0.41$deg$^2$ from 4.1Msec, or 47days, of observing time, which is the largest collection of optical-NIR imaging data of HST for extragalactic science. We reduce all data in a consistent manner with an updated version of our data reduction pipeline. When available, infrared imaging data from the Spitzer Space Telescope are included in photometric analyses. The dataset consists of 316 independent sightlines and is highly effective for identification of high-$z$ luminous sources ($M_mathrm{UV}<-21$mag) at $zsim7$ to $12$, helping to minimize the effects of cosmic variance. As a demonstration, we present three new $z>7$ source candidates, including one luminous galaxy candidate at $z_mathrm{phot}sim10.4$ with $M_mathrm{UV}sim-21.9$ mag; for this object the best-fit spectral energy distribution implies a large amount of stellar mass ($log M_*/M_odot sim 10$) and moderate dust attenuation ($A_V sim 1.4$mag), though the possibility of it being a low-$z$ interloper cannot completely be rejected ($sim23%$) with the current dataset. The dataset presented in this study is also suited for intermediate and low-$z$ science cases.