Identifying collective variables (CVs) that are both discriminative and interpretable remains a central challenge for enhanced sampling and mechanistic analysis of chemical systems. We present Covalent (Collective variables learnt by a computer), a supervised machine learning-based CV discovery pipeline that combines a filter-wrapper-substitution feature funnel with an improved, Riemannian-optimized variant of harmonic linear discriminant analysis (GDHLDA) and a post hoc subspace rotation to concentrate pairwise transition information. We demonstrate the distinguishability and interpretability of CVs generated by Covalent and its applicability across diverse systems, ranging from polycyclic aromatic hydrocarbons to membrane proteins. Beyond identifying low-dimensional CVs that separate metastable states, Covalent also provides geometric insight into how different linear CV spaces can be compared.