# GIS MCP Server (summary)

Use this as concise context for LLMs about GIS MCP.

- What: MCP server exposing Shapely, GeoPandas, Rasterio, PyProj, PySAL
  plus optional data helpers (admin boundaries, climate, ecology, movement,
  land cover, satellite imagery) and visualization.
- Transports: stdio (default), HTTP, SSE. Endpoints (HTTP/SSE): `/mcp` or
  `/sse` and `/storage`. Storage path env/flag: `GIS_MCP_STORAGE_PATH` or
  `--storage-path` (default `~/.gis_mcp/data`). `--debug` enables verbose logs.
- Quick start: `gis-mcp` (stdio) | `GIS_MCP_TRANSPORT=http ... gis-mcp` |
  `GIS_MCP_TRANSPORT=sse ... gis-mcp`.
- Install: `pip install gis-mcp`; extras `[...]` for visualize,
  administrative-boundaries, climate, ecology, movement, land_cover,
  satellite_imagery. Docker: build `docker build -t gis-mcp .`, run
  `docker run -p 9010:9010 gis-mcp` (HTTP mode on 9010).
- Capabilities:
  - Shapely geometry ops, buffering, overlays, validation, length/area/dist.
  - PyProj CRS transforms, EPSG/UTM helpers, geodetic distance/area.
  - GeoPandas vector IO, spatial join/overlay, dissolve/clip, nearest.
  - Rasterio raster IO, crop/reproject/resample/mosaic, NDVI, hillshade,
    focal stats, reclassify, tile, zonal stats.
  - PySAL spatial stats (Moran’s/Geary/Getis-Ord), weights, Markov, OLS+diag.
  - Visualization static/export and interactive web maps (Folium/PyDeck).

Docs: https://gis-mcp.com/
Full context: see `llms-full.txt`.
