The rising prevalence of mental health challenges has increased the need for advanced, personalized digital screening tools. This study introduces DASS-21.EvaLLM, an interactive system that integrates Large Language Models (LLMs) with the DASS-21 framework to generate context-aware, emotionally sensitive recommendations. It also evaluates the advisory quality of four leading LLMs ChatGPT, Gemini, Mistral, and LLaMA across clarity, accuracy, and empathy. The system aims to support early detection, reduce stigma around mental health screening, and promote help-seeking behavior through tailored, compassionate feedback.