Connect any MCP-compatible AI agent to CesiumJS. Camera, layers, entities, spatial analysis — all through natural language.
Camera control, GeoJSON layers, 3D Tiles, terrain, imagery, markers, heatmaps, viewshed analysis, and more — ready to use out of the box.
Claude Desktop, VS Code Copilot, Cursor, Windsurf, or any standards-compliant MCP client. No vendor lock-in.
Bridge (browser SDK), Runtime (MCP server), and Dev (coding assistant) — each independently installable via npm.
Sub-second command execution. Tool call flows from AI agent through the runtime to the browser bridge, and results return instantly.
One command to start: `npx cesium-mcp-runtime`. Add the bridge to your CesiumJS app. Connect your AI agent. That is it.
Major.minor tracks CesiumJS releases (1.139.x targets Cesium ~1.139.0). Patch versions iterate independently for MCP features.
┌──────────────┐ stdio ┌──────────────────┐ WebSocket ┌──────────────────┐
│ AI Agent │ ◄────────► │ cesium-mcp- │ ◄─────────► │ cesium-mcp- │
│ (Claude, │ MCP │ runtime │ JSON-RPC │ bridge │
│ Cursor…) │ │ (Node.js) │ │ (Browser) │
└──────────────┘ └──────────────────┘ └──────────────────┘
│
┌──────▼──────┐
│ CesiumJS │
│ Viewer │
└─────────────┘1. Your AI agent sends a natural language request, which the MCP client translates into a tool call.
2. cesium-mcp-runtime receives the call via stdio and forwards it as a WebSocket command.
3. cesium-mcp-bridge executes the command on the CesiumJS Viewer in the browser.
4. Results flow back through the same pipeline to the AI agent.
| Package | Description | Version |
|---|---|---|
cesium-mcp-bridge | Browser SDK — embed in your CesiumJS application |  |
cesium-mcp-runtime | MCP Server — 24 tools and 2 resources for AI agents |  |
cesium-mcp-dev | IDE MCP Server — Cesium API docs and code generation |  |