Because of the low-permeability nature, stimulated reservoir zone (SRZ) must be created in order to effectively extract gas from shale gas reservoirs. When the gas is extracted, a multiple of physical processes is triggered in SRZ, non-SRZ and hydraulic fractures, and all of them are stress-sensitive. Although these stress-dependencies were investigated in previous studies, the complexity of interactions between different domains (SRZ, non-SRZ and hydraulic fractures) and among multiple physics in each domain has not been understood well. In this study, a fully coupled, multi-domain and multi-physics model is developed to thoroughly capture this complexity. Shale reservoir is characterized as an assembly of three distinctive components: kerogen (in both SRZ and non-SRZ), inorganic matrix (in both SRZ and non-SRZ) and hydraulic fractures. Furthermore, kerogen and inorganic matrix are defined as different double-porosity systems, respectively, while hydraulic fracture is simplified as a 1-D cracked medium. Under this framework, a series of coupled partial differential equations were derived to define different processes in each domain, including Darcy flow, non-Darcy flow effects, shale deformation and gas desorption. These processes are fully coupled through a set of property models such as porosity and permeability. The full set of partial differentiation equations (PDEs) and the associated property models were implemented and numerically solved by COMSOL Multiphysics, a commercial PDE solver. The fully coupled model is validated against an analytical solution of a simplified case, and verified through comparing modelling results with a set of gas production data from a shale reservoir. The results analysis was carried out to investigate the effects of stress dependency on gas recovery. The results of sensitivity analysis reveal how the stress dependency of interactions between domains (SRZ, non-SRZ, and hydraulic fractures) and among multiple physical processes in each domain affect the gas recovery, and demonstrate the unique capability of the proposed model for accurately predicting gas production.