MCP Catalogs
Home

cesium-mcp

by gaopengbin·103·Score 48

AI-powered CesiumJS 3D globe control with 49 tools for camera, entities, layers, animation & spatial analysis via MCP.

developer-toolsai-llmother
18
Forks
0
Open issues
1 mo ago
Last commit
2d ago
Indexed

Overview

Cesium MCP provides a comprehensive bridge between AI systems and CesiumJS 3D globe visualization. It exposes 58 tools across 12 categories that enable natural language control over 3D GIS operations including camera movements, entity manipulation, layer management, animations, and spatial analysis. The project offers three entry points: a browser agent for zero-backend implementation, function calling integration for web applications, and MCP runtime for Claude Desktop/Cursor/Dify integration. The architecture is built around a protocol-agnostic dispatcher that supports multiple AI drivers while maintaining a consistent tool interface.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

you:Natural language control of 3D Earth visualizations in web applications
you:AI-powered GIS analysis through conversational interface
you:Integration of 3D globe capabilities into AI assistants like Claude and Cursor
you:What is the difference between cesium-mcp-bridge and cesium-mcp-runtime?
you:Can I use this with other AI systems besides Claude Desktop/Cursor?

When to choose this

Choose Cesium MCP when you need AI-powered natural language control over CesiumJS 3D globes, especially for GIS applications, data visualization, or integrating AI assistants with geographic systems.

When NOT to choose this

Not suitable for projects requiring high-performance rendering of massive geographic datasets or those needing write access to CesiumJS internal state, as this server focuses on providing tools for AI interaction rather than raw API access.

Tools this server exposes

12 tools extracted from the README
  • flyTo

    Moves the camera to a specific location or entity

  • addMarker

    Adds a marker entity at specified coordinates

  • addModel

    Loads and displays a 3D model at specified location

  • addPolygon

    Creates a polygon entity with specified coordinates

  • setBasemap

    Changes the base map imagery of the globe

  • screenshot

    Captures and saves the current view as an image

  • measure

    Measures distances or areas on the globe

  • createAnimation

    Creates an animation along a specified path

  • load3dTiles

    Loads and displays 3D tilesets

  • addGeoJsonLayer

    Adds a layer from GeoJSON data

  • geocode

    Converts location names to geographic coordinates

  • setView

    Sets the camera view with specific options

Comparable tools

mapbox-mcpopenlayers-mcpcesium-mcp-server

Installation

Installation

Option 1: Browser Agent (Recommended for quick start)

  1. Visit the [live demo](https://cesium-browser-agent.pages.dev/)
  2. Fork the [examples/browser-agent](https://github.com/gaopengbin/cesium-mcp/tree/main/examples/browser-agent) folder to deploy your own

Option 2: Function Calling in web app

npm install cesium-mcp-bridge

Option 3: MCP Runtime

# stdio mode (Claude Desktop, VS Code, Cursor)
npx cesium-mcp-runtime

# HTTP mode (Dify, remote/cloud MCP clients)
npx cesium-mcp-runtime --transport http --port 3000
Claude Desktop Configuration
{
  "mcpServers": {
    "cesium": {
      "command": "npx",
      "args": ["-y", "cesium-mcp-runtime"]
    }
  }
}

FAQ

What is the difference between cesium-mcp-bridge and cesium-mcp-runtime?
cesium-mcp-bridge is the core protocol-agnostic command dispatcher with 60+ tools. cesium-mcp-runtime is the MCP server implementation that exposes the bridge via stdio or HTTP transport for Claude Desktop/Cursor/Dify integration.
Can I use this with other AI systems besides Claude Desktop/Cursor?
Yes, the browser-agent example works with any OpenAI-compatible API. The cesium-mcp-bridge can be integrated into any web app that supports function/tool calling, regardless of the AI provider.

Compare cesium-mcp with

GitHub →

Last updated · Auto-generated from public README + GitHub signals.