Abstract: With the increasing demand for underground resources in China, deep geomechanical issues have become a critical constraint on the advancement of mining, oil, gas, and geothermal engineering into the deeper crust. Deep rocks are typically subjected to coupled in-situ multi-physics conditions, characterized by high stress, high pore pressure, and high temperature. Their mechanical behavior is therefore governed by the in-situ conditions. However, the conventional concept of rock mechanical experiments lacks correspondence with the in-situ conditions, making it difficult to reveal their intrinsic mechanical behavior. To address this gap, the concept of deep in-situ multi-physics conditions is introduced, and a series of stress-, pore pressure-, and temperature-reconstructed experiments on 1 000–6 400 m cores from the Songliao Basin are summarized. Stress–strain responses under reconstructed stress and pore pressure conditions are further predicted through numerical modeling. In addition, NMR experiments on 800 m coal samples from the Zhaolou Coal Mine in Shandong Province are used to validate the thermally induced response mechanisms of porous media under variable temperature conditions. The results show that: under in-situ stress-reconstructed experiments, applying time-dependent stress boundaries equivalent to the in-situ conditions can effectively suppress stress release effects in deep cores. This procedure promotes microcrack closure and intergranular contact, thereby enhancing structural compactness and load-bearing capacity. Compared to conventional triaxial experiments, the elastic modulus and peak deviatoric stress increase by approximately 10%, and the results further highlight the interplay between in-situ stress and diagenetic composition on the mechanical behavior of rocks at different depths. Under in-situ pore pressure-reconstructed experiments, the introduction of non-uniform pore pressure generally leads to a reduction in mechanical strength, with this weakening effect diminishing with increasing depth (confining stress). This suggests that in-situ pore pressure exerts a weakening influence on rock strength, though its magnitude is constrained by the confining stress. Under in-situ temperature-reconstructed experiments, coal and sandstone samples exhibit a non-monotonic variation in porosity, elastic modulus, and Poisson’s ratio across the temperature range from room to in-situ and over-in-situ conditions. These results confirm a thermoelastic recovery mechanism within the room-to-in-situ temperature range, with noticeable thermo-damage and thermo-plastic behavior occurring only beyond the in-situ temperature threshold. Overall, these findings demonstrate that in-situ condition reconstructions offer reliable methods for restoring in-situ boundaries at the laboratory scale, thereby effectively reducing the mechanical deviation caused by the mismatch of experimental boundaries. Based on this, a three-stage experimental approach, including “Reconstruction-Variable Loading-Response,” is proposed for deep solid resource development. In this framework, reconstructed in-situ conditions serve as the experimental baseline, while conditional variable loading acts as the intermediate process linking engineering disturbances to rock mechanical responses. This path-dependent approach enables the characterization of deep rock behavior across the full transition from stable in-situ states to disturbed and unstable failure regimes, providing an expandable experimental methodology for synchronized parameter testing, stability evaluation, and failure mechanism identification in the development of deep underground resources.