Mental stress increasingly affects students and professionals due to academic pressure, workload, and prolonged digital interaction, while existing detection methods rely on intrusive questionnaires or medical sensors. This paper presents a non-medical, non-intrusive stress detection system that analyzes everyday typing behavior to identify early stress patterns without user input or content storage. Features such as typing speed, pause time, backspace frequency, error rate, and typing consistency are extracted and analyzed using machine learning models — Support Vector Machine (SVM) and Random Forest — to classify stress levels as Low, Medium, or High. The system runs silently in the background, preserves user privacy, and provides early stress indications rather than medical diagnoses, making it a practical and cost-effective solution for long-term stress monitoring in academic and workplace settings.

Keywords: Stress Detection, Typing Behavior, Keystroke Dynamics, Machine Learning, SVM, Random Forest, Passive Monitoring, Privacy-Preserving.