Predicting the superconducting transition temperature (T_c) from crystal structure and composition remains a central challenge in condensed-matter physics, reflecting the absence of a broadly predictive framework connecting microscopic bonding to macroscopic quantum behavior. Here, we introduce GP-T_c, an interpretable, structure- and chemistry-aware Gaussian process model that enables uncertainty-quantified T_c prediction from experimentally accessible inputs. By encoding local bonding environments as graphlet histograms, we find that the predictive space collapses to a compact set of descriptors: the distribution of electron-affinity (EA) differences between neighboring atoms, together with interatomic distances and simple elemental features, suffices to predict T_c across disparate superconducting families—identifying an overlooked chemical control parameter that underscores the essential role of local structure beyond composition-only approaches. Our results demonstrate that the EA differences serves as an accessible window into electronic structure providing a mechanism-agnostic physical basis that captures T_c across conventional and unconventional families, including doped charge transfer insulators. GP-T_c reproduces the experimentally reported T_c range of the infinite-layer nickelate Nd_(0.8)Sr_(0.2)NiO₂, and we predict and experimentally confirm superconductivity in stoichiometric PtPb₃Bi (T_c ≈ 3~K). To facilitate broad community use, GP-T_c is made available through a web interface for crystal-structure-based prediction, and the same framework identifies additional high-priority superconducting candidates—including SrNiO₂ and K(PRh)₂---that provide concrete targets for ongoing and future experimental exploration.