Reproducing the orbits and masses of the terrestrial planets Mercury, Venus, Earth, and Mars remains elusive. A successful model for the formation of the terrestrial planets is crucial to better understand key questions related to these planets, such as: building blocks in the primordial protoplanetary disk (feeding zones), origin, timing and amount of water and other materials delivered, number and timing of any giant impacts experienced, and formation timescales. To address these questions, we investigated terrestrial planet formation by performing extensive N-body simulations of primordial protoplanetary disks representative of typical models in the literature. Those disks consisted of embryos and planetesimals placed up to ~1-4 au (depending on the disk model) and the newly formed giant planets. First, we employed our classification algorithm to identify the systems that formed planet analogs of the Venus-Earth pair (in terms of orbit and mass) plus Mercury or/and Mars in the same system. From the analysis of the terrestrial planets formed in such 3- or 4-planet system analogs, we obtained detailed properties regarding the formation and evolution of analogs for each of the four terrestrial planets. These properties include a-e-i-mass distributions, delivery of water/volatiles, number of giant impacts, accretion histories, formation timescales, etc. We also found that compared to the real planets, typical truncated or narrow disks have serious difficulties in reproducing the four terrestrial planets. Worth noting, the formation of Mercury remains an outstanding problem in planetary sciences. Nevertheless, our results can constrain the main properties of the primordial protoplanetary disk that formed the terrestrial planets. In particular, analysis of the results and initial conditions allowed us to better understand the conditions that the disk must fulfill in order to explain the formation of the four terrestrial planets consistently.