GIS MCP Server

Instructions

Implementation Reference

• src/gis_mcp/rasterio_functions.py:1046-1090 (handler)

  The core handler function for the compute_ndvi MCP tool. It reads specified red and NIR bands from the input raster, computes NDVI using the formula (NIR - Red) / (NIR + Red), handles division by zero, copies metadata, and writes the result to a new GeoTIFF file.

  @gis_mcp.tool()
  def compute_ndvi(
      source: str,
      red_band_index: int,
      nir_band_index: int,
      destination: str
  ) -> Dict[str, Any]:
      """
      Compute NDVI (Normalized Difference Vegetation Index) and save to GeoTIFF.
  
      Parameters:
      - source:            input raster path.
      - red_band_index:    index of red band (1-based).
      - nir_band_index:    index of near-infrared band (1-based).
      - destination:       output NDVI raster path.
      """
      try:
          import rasterio
          import numpy as np
  
          src_path = os.path.expanduser(source.replace("`", ""))
          dst_path = os.path.expanduser(destination.replace("`", ""))
  
          with rasterio.open(src_path) as src:
              red = src.read(red_band_index).astype("float32")
              nir = src.read(nir_band_index).astype("float32")
              ndvi = (nir - red) / (nir + red + 1e-6)  # avoid division by zero
  
              profile = src.profile.copy()
              profile.update(dtype="float32", count=1)
  
          os.makedirs(os.path.dirname(dst_path) or ".", exist_ok=True)
  
          with rasterio.open(dst_path, "w", **profile) as dst:
              dst.write(ndvi, 1)
  
          return {
              "status": "success",
              "destination": str(dst_path),
              "message": f"NDVI calculated and saved to '{dst_path}'."
          }
  
      except Exception as e:
          raise ValueError(f"Failed to compute NDVI: {e}")

• src/gis_mcp/rasterio_functions.py:11-35 (registration)

  MCP resource that lists 'compute_ndvi' among available rasterio operations, effectively advertising the tool's availability.

  @gis_mcp.resource("gis://operation/rasterio")
  def get_rasterio_operations() -> Dict[str, List[str]]:
      """List available rasterio operations."""
      return {
          "operations": [
              "metadata_raster",
              "get_raster_crs",
              "clip_raster_with_shapefile",
              "resample_raster",
              "reproject_raster",
              "weighted_band_sum",
              "concat_bands",
              "raster_algebra",
              "compute_ndvi",
              "raster_histogram",
              "tile_raster",
              "raster_band_statistics",
              "extract_band",
              "zonal_statistics",
              "reclassify_raster",
              "focal_statistics",
              "hillshade",
              "write_raster"
          ]
      }

• src/gis_mcp/main.py:66-72 (registration)

  Import statements in the main entry point that trigger the execution of decorators in rasterio_functions.py, registering the compute_ndvi tool with the FastMCP instance.

  from . import (
      geopandas_functions,
      shapely_functions,
      rasterio_functions,
      pyproj_functions,
      pysal_functions,
  )

• src/gis_mcp/mcp.py:1-6 (helper)

  Definition of the gis_mcp FastMCP instance used by the @gis_mcp.tool() decorator to register tools including compute_ndvi.

  # MCP imports using the new SDK patterns
  from fastmcp import FastMCP
  
  
  gis_mcp = FastMCP("GIS MCP")