Global climate change increases extreme rainfall events, significantly elevating the frequency of geological hazards in granite residual soil (GRS) regions. The stress path of slope soils under extreme rainfall conditions differs from conventional triaxial scenarios, characterized by nearly constant deviatoric stress (CQ) and continuously varying mean effective stress ( \(p^{\prime }\) ). To investigate the instability and deformation behaviors of GRS under cyclic variations in \(p^{\prime }\) along the CQ path, consolidated drained (CD) tests, constant shear drained (CSD) tests, \(p^{\prime }\) cyclic tests along the CQ path and stepped stress level tests along the CQ path were conducted. The test results demonstrate that GRS under the CSD path exhibits instability characterized by dilative volumetric behavior and a rapid increase in axial strain. \(p^{\prime }\) cyclic tests along the CQ path reveal the existence of a potential instability stress ratio ( \(\eta_{{I_{p} }}\) ) less than instability stress ratio ( \(\eta_{IS}\) ), at which specimens undergo instability under multiple \(p^{\prime }\) cycles. Instability-type specimens demonstrate progressive accumulation of both deviatoric strain and volumetric strain with increasing cycle numbers, while exhibiting delayed deformation. Specifically, dilation occurs when \(p^{\prime }\) increases. In contrast, stability-type specimens primarily undergo elastic deformation. Stepped stress level test results indicate that abrupt changes in strain rate drive this delayed deformation pattern, leading to a corresponding delay in the \(\eta_{IS}\) values determined by current instability criteria. Deformation behaviors under cyclic loading show that the stability of GRS is critically dependent on the evolution of plastic volumetric strain increment ( \(\Delta \varepsilon_{v}^{p}\) ). Consequently, a methodology utilizing the strain increment ratio ( \(\Delta \varepsilon_{v} /\Delta \varepsilon_{q}\) )—stress ratio ( \(\eta\) ) curve from CSD tests is proposed to determine \(\eta_{{I_{p} }}\) . The results demonstrate that this method provides accurate predictions of instability under cyclic loading. The research findings provide critical references for geological hazard prevention and mitigation in GRS regions under extreme rainfall conditions.