With the increasing prevalence of mental health issues, especially among university students, depression has emerged as a concern requiring effective support. To partially mitigate this, we develop a Vietnamese dialogue system that automates mental healthcare services for students facing significant stress. The system has a hybrid architecture that combines a large language model-based chatbot with a finite-state dialogue system to address two crucial tasks: (1) addressing inquiries related to student mental health issues and (2) assessing depression levels using standardized questionnaires. Furthermore, a user intent detector is employed to enable flexible switching between these two tasks during the conversation. Due to the lack of Vietnamese benchmarks regarding student mental health, we carefully collected and manually annotated a dataset for the testing phase. The dataset can be a student mental health benchmark for testing Vietnamese dialogue systems. The conducted experiments on the generated dataset demonstrate the system’s ability to provide meaningful psychological support and deliver accurate depression assessments, representing a promising approach to addressing mental health challenges among college students.