The therapeutic efficacy of programmed cell death protein 1/programmed cell death–ligand 1 (PD-1/PD-L1) blockade immunotherapy is extremely dampened by complex immunosuppressive mechanisms including regulatory T cells (Treg), M2 macrophages (M2), and prostaglandin E2 (PGE2). The pivotal roles of PGE2 have been recognized by directly inactivating CD8+ T cells and indirectly inducing Treg and M2. Therefore, PGE2 abolishment through inactivating cyclooxygenase-2 (COX-2) could be robust to sensitize tumour toward anti-PD-1/PD-L1 immunotherapy, which has gone into clinical trials. However, exploring this promising strategy in nanomedicine to enhance immunotherapy remains unrevealed. The key challenge to synergistically combine COX-2 inhibition and anti-PD-1/PD-L1 lies in the different pharmacokinetic profiles and the spatial obstacles since PD-1/PD-L1 interaction occurs extracellularly and COX-2 locates intracellularly. Thus, the programmed release nanoparticles (termed as Cele-BMS-NPs) are rationally designed, which are composed of pH-sensitive human serum albumin derivative, BMS-202 compound as PD-1/PD-L1 inhibitor, glutathione (GSH)-activatable prodrug of celecoxib (COX-2 inhibitor). The in vitro experiments demonstrate that this smart Cele-BMS-NPs could extracellularly release BMS-202 under the acidic tumour microenvironment, and the intracellularly release of celecoxib in response to the elevated GSH concentration inside tumour cells. After systemic administration, the intratumoral infiltration of CD8+ T cells is significantly enhanced and meanwhile immunosuppressive M2, Treg, and PGE2 are reduced, thereby eliciting the anti-tumour immune responses toward low immunogenic tumours and postsurgical tumour recurrences.