HEADLESS GIS ANALYST · MODEL CONTEXT PROTOCOL

Your GIS analyst,
inside Claude.

GeoEDGE MCP turns Claude Desktop into a spatial analyst. Ask in plain English — it plans the workflow, runs 130+ deterministic GIS tools, and hands back correct, standards-compliant maps and data. No GIS commands, no Python, no wrestling with projections.

One-click install into Claude Desktop. A license key (issued by GeoEDGE) unlocks the tools — find yours here.

130+GIS tools, one interface
Vector · Raster · MCDAFull analyst toolkit
OGCStandards-compliant output
HeadlessNo new software to learn
// WHY IT'S DIFFERENT

The heavy spatial work never touches your context window.

Spatial data is enormous — a single Sentinel-2 tile is over 100 MB, thousands of times larger than an entire LLM context. Asking a model to reason through coordinate math or read raw features token-by-token is slow, expensive and error-prone. GeoEDGE flips that: the model orchestrates, the tools compute.

Tools do the math, natively

Geometry and raster operations run in optimised C libraries (GDAL, GEOS), not in tokens. The model never hand-rolls a buffer or a reprojection — so it never spends context "thinking" through spatial arithmetic.

Big data stays on disk

Datasets are held in a stateful layer registry and referenced by name. Rasters and feature collections never get pasted into the prompt — the conversation stays tiny no matter how large the data.

Compact, structured results

Each tool returns a short, validated summary and a file path — a map, a GeoPackage, a table — instead of dumping thousands of raw coordinates or pixel values back into the chat.

One call, not ten round-trips

Multi-step analyses run as a single guided workflow. Fewer model turns means fewer tokens, faster answers, and a fully recorded provenance trail.

The result: you can work with gigabyte rasters and thousands of features while your token usage — and your bill — stays flat.

// EVERYTHING A GIS ANALYST NEEDS

One agent. The whole geospatial workflow.

Vector operations

Buffer, clip, overlay, dissolve, spatial joins, centroids, distances and areas — precise geometry powered by GEOS/Shapely.

Raster & imagery

NDVI and band math, reproject, clip, resample, merge, hillshade and zonal statistics on gigabyte rasters — via rasterio/GDAL.

Coordinate systems

Reproject and transform between any CRS with pyproj. Everything lands in the right projection, automatically.

Suitability & MCDA

Multi-criteria decision analysis and weighted overlays for site selection — combine constraints into a defensible suitability surface.

Spatial statistics

Spatial autocorrelation (Moran's I), hot-spot analysis (Getis-Ord), clustering and neighbourhood models with PySAL/esda.

Topology & validation

Validate and repair geometries, detect overlaps and gaps, and run proximity checks before results reach a decision.

Data connectors

Read from PostGIS and ArcGIS REST services, and pull public datasets — Sentinel-2 / STAC imagery, OpenStreetMap and more.

Cartographic outputs

Publication-quality static maps and interactive web maps — choropleth, graduated symbols and heatmaps, plus tabular reports.

Standards & schema

OGC-conformant outputs validated against a company schema at the boundary — you get clean, compliant data, not raw model text.

Guided workflows

Templated multi-step analyses run as one deterministic call — full provenance recorded so every result is traceable.

// HOW IT WORKS

Ask. It reasons, acts, and verifies.

1
Describe the task

Type a plain-English question. No syntax, no tool names to remember.

2
It plans & calls tools

Claude picks the right deterministic GIS tools and runs them in order.

3
Results are validated

Every output is coerced to the schema and checked before you see it.

4
You get maps + a plain answer

A short narrative, plus the formal map, data file and report.

// TRY ASKING
>_Buffer these bore locations by 500 m, clip them to the catchment, and map the overlap with the protected zones.
>_Compute NDVI from the latest Sentinel-2 tile and flag low-vegetation areas inside the project boundary.
>_Build a solar-farm suitability surface: slope under 5°, within 2 km of a substation, outside flood zones.
>_Run a hot-spot analysis on incident counts by suburb and give me a choropleth I can drop into the report.
// GET STARTED

How to get access

  1. A GeoEDGE administrator issues you a license key.
  2. Create an account with the same email and find your key on your account page.
  3. Download the extension and open it in Claude Desktop → Settings → Extensions.
  4. Paste your key when prompted. The GIS tools unlock — and stay current automatically.

Self-serve subscriptions are coming soon. For now, licenses are provisioned by GeoEDGE staff.

Ready to map at the speed of a sentence?

Download the extension, add your license key, and start analysing.