<![CDATA[Prolonged use of digital devices triggers the risk of Computer Vision Syndrome (CVS) and neck posture abnormalities such as Forward Head Posture. Conventional detection systems with static thresholds often trigger false alarms due to differences in individuals' natural postures. This study aims to develop a real-time neck posture monitoring web application to prevent CVS adaptively. The system was developed using the Prototyping method based on a Progressive Web App (PWA). Visual processing is executed entirely on the client-side using MediaPipe Pose Landmarker and WebRTC, while dynamic threshold calibration utilizes the Personalized Baseline method. Monitoring data and gamification history are managed locally through Local Storage. The test results indicate that the system is capable of extracting neck coordinates, monitoring posture inclination, and providing ergonomic notifications accurately without data transmission to a server. The Personalized Baseline method proved effective in reducing the false alarm rate, making this system an adaptive, lightweight CVS preventive solution capable of ensuring user privacy. Keywords: Computer Vision Syndrome; Forward Head Posture; Local Storage; MediaPipe; Personalized Baseline. Abstrak Penggunaan perangkat digital yang berkepanjangan memicu risiko Computer Vision Syndrome (CVS) dan kelainan postur leher seperti Forward Head Posture. Sistem deteksi konvensional dengan ambang batas statis sering memicu peringatan palsu akibat perbedaan postur alami individu. Penelitian ini bertujuan mengembangkan aplikasi web pemantau postur leher real-time untuk mencegah CVS secara adaptif. Sistem dikembangkan menggunakan metode Prototyping berbasis Progressive Web App (PWA). Pemrosesan visual dieksekusi sepenuhnya di sisi klien menggunakan MediaPipe Pose Landmarker dan WebRTC, sedangkan kalibrasi ambang batas dinamis menggunakan metode Personalized Baseline. Data pemantauan dan riwayat gamifikasi dikelola secara lokal melalui Local Storage. Hasil pengujian menunjukkan sistem mampu mengekstrak koordinat leher, memantau kemiringan postur, dan memberikan notifikasi ergonomi secara akurat tanpa transmisi data ke peladen. Metode Personalized Baseline terbukti efektif menekan tingkat peringatan palsu, menjadikan sistem ini sebagai solusi preventif CVS yang adaptif, ringan, dan mampu menjamin privasi pengguna.]]>
