GeoLibre v2.3.0 brings self-writing legends and browser-based geoprocessing

GeoLibre, the open-source geographic information system that runs in a browser but never sends your data to a server a principle shared by Bruno 4.0's BYOK approach, shipped version 2.3.0 this week. The release focuses on making map-making less tedious and geoprocessing faster by keeping everything on the client side.

The headline feature is an automatic legend that reads the symbology from visible layers and builds itself. For graduated, categorized, rule-based, and expression-driven styles, it draws class rows in the correct colors. Heatmaps and raster colormaps get gradient bars. If the raster has an attribute table, it pulls land-cover labels directly. You can rename, hide, reorder, or add your own entries, and the legend saves with the project file.

The Layers panel also got a visual upgrade. Each row now shows a dot, line, square, or image glyph in the layer's actual color, so a tall stack of layers is readable at a glance without clicking into each one.

Browser-native geoprocessing at scale

Version 2.3.0 includes more than 200 geoprocessing tools written in Rust, compiled to WebAssembly, and running entirely in the browser like the local inference acceleration from Sipp.sh. No server round-trips, no uploads. Among the new additions is an Emerging Hot Spot Analysis that builds a space-time cube from timestamped points and classifies each cell into seven categories, new, intensifying, persistent, diminishing, sporadic, oscillating, or historical, all computed client-side.

The Time Slider now steps through MosaicJSON and STAC collections that contain many Cloud Optimized GeoTIFFs per date, on either a GPU or the WebAssembly rendering engine. That makes time-series analysis of satellite imagery more practical without standing up a backend similar to how open web agents operate in-browser without server calls.

GeoLens catalog browser and smarter data loading

GeoLibre v2.3.0 introduces the GeoLens catalog browser, which connects to a self-hosted GeoLens server, searches its catalog, and adds datasets as vector tiles, GeoJSON, or rendered raster tiles. That gives teams a straightforward way to browse and pull in private or large geospatial datasets without external dependencies.

Data import got smarter too. Users can now pick which layers to load from a multi-layer GeoPackage, import CSVs whose coordinates are in any projected coordinate reference system, and check a raster's real CRS, pixel size, and extent directly from the metadata dialog.

Copy-paste styles, shareable tool links, and multiple AI profiles

Styling a set of layers consistently no longer requires repeating the same clicks. The new copy and paste style commands let users give a whole project one look without restyling each layer individually. Tool setups are also shareable: any Whitebox tool can be deep-linked with a ?tool= URL that opens the dialog preselected, and a copy link button builds that URL automatically.

The assistant panel now supports multiple AI profiles similar to multi-model routing in Claude Code. Users can define several provider, model, and credential setups, pick a default, and switch between them while working.

GeoLibre is free software, built on MapLibre, and available at web.geolibre.app. The project is on GitHub under the Open Geospatial Solutions organization joining other open-source platforms like Nvidia's DeepStream. The release notes and a documentation site are linked from the source announcement.