The cosmic web forms an intricate network of matter spanning a wide range of scales and densities, with galaxy superclusters as its largest, overdense yet generally gravitationally unbound structures. Because of their large extents, superclusters may retain information from early stages of structure formation, though their evolution remains poorly understood due to limited samples. We present the largest catalogue of superclusters to date, identified from SDSS-BOSS DR12 spectroscopic data over 0.15 < z < 0.7 using a modified friends-of-friends algorithm that accounts for survey selection effects. Superclusters are defined as regions exceeding five times the mean cosmic density, and we show that a luminosity-density field better traces the underlying matter than a number-density field. Our method recovers known systems such as the BOSS Great Wall, the Einasto supercluster, and the Saraswati supercluster. Typical sizes are ~40 Mpc, with the largest spanning several hundred Mpc. We also introduce a novel method of estimating the confidence of each supercluster. Unlike recent supercluster catalogues based on galaxy cluster samples, our galaxy-based approach provides a more complete and accurate characterisation of supercluster morphology and extent, and the developed methods can be applied to surveys like 4MOST and DESI.