Molecular glues are promising protein-degrading agents that hold great therapeutic potential but face significant challenges in rational design, effective synthesis, and precise targeting of tumor sites. In this study, we first overcame some of these limitations by introducing a fumarate-based molecular glue handle onto specific ligands of therapeutic kinases (TBK1, FGFR, and Bcr-Abl), resulting in the effective degradation of these important cancer targets. Despite the broad applicability of the strategy, we unexpectedly discovered potent and widespread cytotoxicity across various cell lines, including noncancerous ones, rendering it less effective in cancer therapy. To address this critical issue, we next developed a self-assembled nanoparticle-based molecular glue (nano-mGlu) based on one of the newly discovered Bcr-Abl-degrading molecular glues (H1-mGlu). We showed that the resulting nano-mGlu (named Cle-NP) was able to release H1-mGlu in vitro in the presence of a high concentration of GSH or H2O2 (commonly found in the tumor microenvironment). Subsequent in vivo antitumor studies with a K562-xenografted mouse model indicated that Cle-NP was highly effective in tumor-specific degradation of endogenous Bcr-Abl expressed in K562 cells, leading to eventual tumor regression while maintaining good biosafety profiles. With key advantages of generality in molecular glue design, targeted delivery (e.g., H1-mGlu), potent antitumor activity partially induced by target-specific degradation, and minimized collateral damage to healthy tissues, our self-assembled nano-mGlu strategy thus provides a novel approach that might hold a significant promise for effective and personalized cancer therapy.