shubhranshugupta/jts-mcp

jts-mcp

MCP server exposing selected library functions from library JTS (Java Topology Suite) to solve Geospatial problems using Agentic AI

Features

• Extensive geospatial tools exposed as Spring AI tools
• Endpoints for geometry creation, spatial queries, and transformations

Tools

• createPoint - Create a Point geometry from [x, y] coordinates and return WKT.
• createLine - Create a LineString geometry from a list of [x, y] coordinates and return WKT..
• createPolygon - Create a Polygon geometry from a list of [x, y] coordinates and return WKT.
• bufferGeometry - Buffer a geometry by a given distance and return WKT.
• intersects - Check if two geometries intersect..
• contains - Check if the first geometry contains the second.
• distance - Calculate the distance between two geometries.
• unaryUnionLines - Merge all LineStrings into one network using UnaryUnionOp.
• getBoundary - Extract the boundary of a geometry as a LineString.
• intersectGeometries - Intersect two geometries and return the result as WKT.
• mergeAndSelectLongestChain - Merge intersected boundary segments into continuous chains and pick the longest one.
• projectOnRing - Project a point onto a ring and return the index.
• extractSubLine - Extract a subline from a ring between two indices.
• makePolygonFromRing - Create a Polygon from a closed LineString WKT.
• rebuildPolygonRing - Rebuild a polygon ring by splicing source ring and target chain.
• isValidGeometry - Check if a geometry is valid.
• polygonize - Polygonize: Create all possible polygons from a network of LineStrings.
• calculateArea - Calculate the area of a polygon.
• calculateCentroid - Calculate the centroid of a polygon and return as Point WKT.
• hausdorffDistance - Calculate the Hausdorff distance between two polygons.

Integration with Agentic AI

• Use the SSE integration from the MCP Client to inform your LLM about available endpoints and their usage.
  • e.g configuration in MCP client:

    mcp:
      client:
        enabled: true
        name: jts-mcp-client
        version: 1.0.0
        type: SYNC
        request-timeout: 20s
        toolcallback.enabled: true
        sse:
          connections:
            jts-mcp-sse:
              url: http://localhost:8081
    

• Example workflow:
  1. User provides a geospatial problem statement.
  2. Spring AI app sends it to LLM.
  3. LLM returns actions (e.g., buffer, intersect, fix).
  4. Spring AI app calls MCP endpoints to execute actions and sends results back to LLM.
  5. LLM iterates until the problem is solved.
  6. Final result is presented to the user.

Running the Server

mvn spring-boot:run