Supercritical CO2 (SC–CO2) would be a potential substitute for water-based fracturing fluid widely used in the present hydraulic fracturing of shale reservoirs. However, the field application of SC-CO2 fracturing is still in its infancy. Laboratory experiments are crucial for understanding the effects of various fracturing parameters and validating numerical simulations before conducting costly field trials. This paper first introduces the design of a self-developed experimental apparatus for fracturing cylindrical rock specimens and outlines the working principles of six functional modules. The entire experimental system is flexible because its functional modules can be used interchangeably. Subsequently, some key designs are emphasized. (1) A packer with the sealing of O and Y rings exhibits a good seal efficiency for a relatively small wellbore; moreover, this packer can be conveniently moved out from the fractured shale without damaging the specimen. (2) Downhole temperature measurement structure integrated into the packer can sensitively monitor the CO2 temperature variation at a high sampling rate during the entire SC-CO2 fracturing process. (3) The design of the triaxial core holder contributes to the realization of withstanding high axial-confining and fluid injection pressures and customized installation of acoustic emission (AE) sensors. Ultimately, some fracturing experiments with water and SC-CO2 as fracturing fluids and seepage tests with nitrogen (N2) under various test parameters are successfully conducted. The results indicate that, after the specimen fractures, the injection pressure would be approximately decreased to the confining pressure due to the confining chamber sealing design; there is a simultaneous sharp variation in the temperature at the bottom of the wellbore. SC-CO2 fracturing generates more fractures with greater bifurcation and higher tortuosity. The gas flow rate decreases with increasing confining pressure, whereas it increases significantly with increasing injection pressure. In general, these experimental results can validate the performance and feasibility of the self-developed experimental apparatus.