Atrial fibrillation (AF) imposes a growing burden on global healthcare, driving demand for non-invasive markers of atrial remodelling. The P-wave of the surface electrocardiogram (ECG) offers a window into atrial conduction and substrate, yet its quantitative analysis remains fragmented: studies rely on heterogeneous, often undisclosed signal processing pipelines, and even fundamental quantities such as amplitude and terminal force lack consensus definitions. This algorithmic opacity undermines cross-study comparability and has contributed to contradictory findings on post-ablation P-wave changes. We present an open-source library that addresses this gap by grounding atrial activity-derived metrics—from duration and amplitude to fragmentation and frontal-plane axis—in explicit mathematical mappings from annotated waveform to feature space. We demonstrate the library on a pilot cohort of twenty subjects spanning a spectrum of atrial disease, illustrating how preprocessing assumptions alone can shift extracted biomarkers. By decoupling physiological signal from methodological artefact, this framework provides the infrastructure for reproducible, large-scale atrial phenotyping.