Locating a point in relation to a systematic division of sheets remains a key issue in computational geometry and GIS practice, especially in the context of for archival cartographic resources. The Krakow Local System (ULK), widely used in the past for mapping the city of Krakow, is still found across numerous analog and digital datasets. This paper presents spatial indexing algorithms for ULK: an encoder (point → sheet code) and a decoder (sheet code → sheet extent) that mirror the cascaded subdivision from 1:2000 down to 1:1000 and 1:500. The procedures are concise and constant-time O(1), which allows them to be used “on the fly”—for example, automatic derivation of the sheet code from coordinates (XY) and immediate attachment of the corresponding archival rasters, as well as computing a sheet frame prior to loading based on a code embedded in the filename. In combination with ULK ↔ PL-2000/1992 transformations, the approach supports resource consolidation, quality assurance (QA), and integration with GIS repositories and web services. In AI/ML workflows, the sheet code serves as a spatial label, facilitating automated data labelling, training set construction, and multimodal search. Importantly, these are not just sketches or pseudocode: the algorithms are provided as ready-touse implementations (VB6) and can be ported directly to a variety of environments—from GIS and CAD (e.g., MicroStation VBA), through Python/R scripts (code translator), database functions (e.g., SQL/PLpgSQL), and even spreadsheets (Excel/LibreOffice: formulas, Power Query). This makes them a practical foundation for indexing and managing large collections of archival rasters within modern GIS/AI pipelines.