This study presents a comprehensive, open-source methodology for the automatic classification of masonry pathologies in historical buildings using 3D point cloud data and supervised machine learning. The proposed pipeline integrates Terrestrial Laser Scanning (TLS) with a geometric feature-based analysis, followed by classification using a Random Forest algorithm executed in a cloud-based Python environment. The workflow was tested on Palazzo Labriola, a historic structure located in Tursi, Italy, characterized by diverse surface degradation phenomena. Point cloud preprocessing and feature extraction were performed using CloudCompare, focusing on curvature and point density as proxies for surface anomalies. A rule-based labeling strategy was used to generate training data, and classification results were visualized through color-coded outputs, facilitating expert interpretation. All operations were implemented within open-source frameworks, ensuring methodological transparency, computational accessibility, and cross-platform reproducibility. The results demonstrate the viability of the approach for supporting preventive conservation through scalable, automated diagnostics. By reducing dependency on manual inspection and commercial software, this workflow contributes to a more inclusive and standardized practice in digital heritage management. Future work will explore multimodal feature integration and broader applicability across architectural typologies.