GDAL

GDAL MCP enables geospatial analysis through a Model Context Protocol server that requires you to justify methodological choices before execution. This reflection-based workflow improves transparency, auditability, and cross-domain reuse of reasoning while delivering raster and vector capabilities through a unified MCP interface.

How to use

You interact with the GDAL MCP server through an MCP client or orchestration layer. The server supports pre-execution reflection to justify choices like coordinate reference systems (CRS) and resampling methods, then carries out geospatial operations in a way that is auditable and reproducible. You can chain raster and vector operations in a single workflow and reuse core justifications across data domains to keep methodological consistency.

Typical usage patterns include preparing data (inspecting metadata, choosing projection or resampling strategies), triggering reflective justification, and then executing the requested operation. The system caches justified methodologies so that repeated steps in multi-operation workflows execute quickly while preserving the documented reasoning.

How to install

Prerequisites: you need a Python-enabled environment and access to an MCP client or launcher that can speak the GDAL MCP protocol. The server is designed to run with the MCP tooling that can initiate an stdio-based workflow.

Install the MCP client tooling from its distribution and ensure it is available on your system PATH. The client will start the server in stdio mode so you can interact programmatically or via automation pipelines.

Start the server in stdio mode using the following command (run from your terminal):

uvx --from gdal-mcp gdal --transport stdio

Configuration and quick start with Claude Desktop

If you configure a Claude Desktop integration, add an MCP server entry that uses the stdio transport. This enables Claude to launch the MCP workflow directly from your desktop environment.

{
  "mcpServers": {
    "gdal_mcp": {
      "command": "uvx",
      "args": ["--from", "gdal-mcp", "gdal", "--transport", "stdio"],
      "env": {
        "GDAL_MCP_WORKSPACES": "/path/to/your/geospatial/data"
      }
    }
  }
}

Additional notes

The GDAL MCP server exposes a comprehensive toolset for raster and vector operations, including reflection-enabled workflows that justify CRS selection and resampling methods on first use and cache those rationales for future operations.

Available tools

raster_info

Inspect raster metadata including CRS, resolution, bands, and nodata values.

raster_convert

Convert raster formats with options for compression and overviews, including COG support.

raster_reproject

Reproject rasters to a new CRS with optional interpolation methods and warping controls.

raster_stats

Compute statistics and histograms for raster datasets.

vector_info

Inspect vector metadata such as CRS, geometry types, and attributes.

vector_reproject

Reproject vector layers to a target CRS with appropriate handling of geometries.

vector_convert

Migrate between vector formats like SHP, GPKG, and GeoJSON.

vector_clip

Subset vectors by geometry or extent for focused analyses.

vector_buffer

Perform proximity analysis by creating buffers around geometries.

vector_simplify

Simplify geometries to reduce complexity while preserving topology.

store_justification

Cache epistemic reasoning to enable rapid reuse in subsequent steps.

advisory_prompts

Provide guidance prompts for CRS selection and resampling methods during reflection.