<- all tokdocs
2026.08.04@whitewhoadie

GeoLibre v2.3.0: A Full GIS Platform That Never Sends Your Data Anywhere

Coding
Watch on TikTok

View on TikTok ->

GeoLibre is a free, open-source GIS platform that runs in the browser, on desktop, on mobile, and inside Jupyter, and it does all of its processing on your own machine. The clip is a 32-second launch announcement for version 2.3.0 from @whitewhoadie, delivered straight to camera while screen recordings of the app play behind him. The whole pitch lands on one line from the transcript: "It keeps your local data private. Keeps your data local and private. Who's going to hate on that?"

One App, Four Places You Already Work

The transcript sets up the core claim early: "It runs everywhere you do. So like in a web browser, on the desktop, on mobile, and inside Jupyter Notebooks." That is not marketing shorthand. GeoLibre ships as a web app with no install, a native desktop build for Windows, macOS, and Linux, an Android app, and a geolibre Python package for Jupyter. The project files move between these installations, so the same map you build in a notebook opens on your phone.

The frames back this up. The screen behind the narrator shows a standard desktop GIS layout: a layered legend panel on the left, a dark-themed map canvas in the center, and a top toolbar. It looks like QGIS or ArcGIS rather than a stripped-down web toy, which is the point the video is making.

Local-First Is the Whole Argument

Most cloud GIS tools ask you to upload your data to their servers before you can analyze it. GeoLibre inverts that. The getting-started docs describe it as "processing everything client-side in your browser session," and the video repeats the local-and-private line twice for emphasis.

This matters for anyone working with sensitive or regulated geospatial data: parcel records, infrastructure, health data tied to location, field survey results. The narrator's closing question, "Who's going to hate on that?", is aimed squarely at that audience. If your data never leaves the machine, you sidestep a large category of compliance and vendor-lock-in problems.

What v2.3.0 Actually Added

The clip is thin on feature detail, but the caption and the release notes fill it in. Version 2.3.0 shipped on July 25, 2026, and added Emerging Hot Spot Analysis built on a space-time cube, mosaic time-series support in the time slider, multiple AI assistant profiles, auto-generated map legends from your symbology, and the GeoLens catalog browser for finding datasets.

The frames show two of these directly. Several frames feature an auto-generated legend built from the layer symbology, and one frame captures a "Move to Time Slider" dialog, which is the temporal visualization feature the release notes describe.

The Geoprocessing Engine Runs in WebAssembly

The caption claims "200+ Rust geoprocessing tools in the browser." The public repo describes it slightly differently: 1,000+ geoprocessing tools that execute in WebAssembly with no server, covering vector, raster, remote sensing, hydrology, terrain, LiDAR, and conversion operations. Treat the exact tool count in the caption as approximate; the confirmed number in the repo is higher.

The underlying stack is Tauri v2, React, TypeScript, MapLibre GL JS, and DuckDB-WASM Spatial. DuckDB-WASM is what powers the in-browser SQL workspace, so you can run spatial queries against your data without a database server. This is the machinery that makes the local-first promise possible: the analysis engine ships to your browser instead of your data shipping to a cloud.

Key Takeaways

  • GeoLibre is free, open source (MIT licensed), and runs in the browser, on desktop, on mobile, and in Jupyter from a single project format.
  • All processing happens client-side, so your geospatial data never leaves your machine.
  • Version 2.3.0 (July 25, 2026) added Emerging Hot Spot Analysis, mosaic time series, multiple AI profiles, auto-generated legends, and the GeoLens catalog browser.
  • The geoprocessing engine runs entirely in WebAssembly; the repo lists 1,000+ tools, more than the "200+" figure in the video caption.
  • The stack (Tauri, MapLibre GL JS, DuckDB-WASM Spatial) is what lets a full GIS run without cloud lock-in.
  • If you handle sensitive or regulated location data, the local-first model removes a whole class of upload-and-compliance friction.

Resources

Published August 4, 2026. Writeup generated from a favorited TikTok.