Very recently, MXene-based wearable hydrogels have emerged as promising candidates for epidermal sensors due to their tissue-like softness and unique electrical and mechanical properties. However, it remains a challenge to achieve MXene-based hydrogels with reliable sensing performance and prolonged service life, because MXene inevitably oxidizes in water-containing system of the hydrogels. Herein, catechol-functionalized poly(vinyl alcohol) (PVA-CA)-based hydrogels is proposed to inhibit the oxidation of MXene, leading to rapid self-healing and superior strain sensing behaviors. Sufficient interaction of hydrophobic catechol groups with the MXene surface reduces the oxidation-accessible sites in the MXene for reaction with water and eventually suppresses the oxidation of MXene in the hydrogel. Furthermore, the PVA-CA-MXene hydrogel is demonstrated for use as a strain sensor for real-time motion monitoring, such as detecting subtle human motions and handwriting. The signals of PVA-CA-MXene hydrogel sensor can be accurately classified using deep learning models.