AI Integration & MCP

Every Maptoolkit API - maps, routing, search, elevation, static images, and more - works out of the box with any major LLM. Whether you’re building with Claude, ChatGPT, Gemini, or an AI-powered IDE like Cursor or Windsurf, you can go from idea to working map code in a single prompt.

The recommended way to connect your AI tool to Maptoolkit is via our MCP Server.

What is the Maptoolkit MCP?

Model Context Protocol (MCP) is an open standard that lets AI assistants connect to external tools and APIs. Once connected, your AI assistant knows about all Maptoolkit endpoints, parameters, and capabilities - no copy-pasting of API documentation required.

Connect the Maptoolkit MCP Server

The Maptoolkit MCP server is available as a remote server - no installation needed, just add the URL to your AI tool.

https://mcp.maptoolkit.net

{
  "mcpServers": {
    "maptoolkit": {
      "url": "https://mcp.maptoolkit.net"
    }
  }
}

{
  "mcpServers": {
    "maptoolkit": {
      "url": "https://mcp.maptoolkit.net"
    }
  }
}

{
  "mcpServers": {
    "maptoolkit": {
      "url": "https://mcp.maptoolkit.net"
    }
  }
}

Using Maptoolkit on Localhost

When your application runs on localhost or 127.0.0.1, all Maptoolkit APIs and the MCP Server work without an API key. You can start building and testing without signing up first.

Start any local development server - the API requests will go through automatically:

# Vite
npm run dev

# Create React App / Next.js
npm start

# Webpack Dev Server
npx webpack serve

# Simple static file server
npx serve .

Note: opening an HTML file directly in the browser (file://) does not count as localhost. Your app needs to be served through a local server.

When you’re ready to deploy to production, get an API key and add it to your app configuration.

Using Prompts from the Documentation

Throughout this documentation you’ll find ready-to-use prompts alongside code examples. Each prompt is written to produce the shown result when used with an MCP-connected AI assistant.

Look for the Prompt tab in code examples - copy it directly into your AI tool and it will generate working map code, complete with your API key placeholder and the correct Maptoolkit endpoints.

Example: On Vector Tiles → MapLibre you’ll find prompts that produce a full MapLibre map with 3D terrain in one shot.

What the MCP Server Knows

Once connected, your AI assistant has access to:

• All Maptoolkit API endpoints with full parameter documentation
• Authentication requirements and API key usage
• Tile URL formats for vector and raster tiles
• Routing, isochrone, geocoding, and elevation APIs
• Style URLs and configuration options

No need to paste documentation into your prompt - just ask.

Example Prompts to Get Started

Once the MCP server is connected, try these:

Show me a full-page MapLibre map using Maptoolkit vector tiles
with the terrain style, centered on Vienna.

Create a routing example that finds the fastest route between
two coordinates and displays it on a Maptoolkit map.

Build a geocoding search box that shows results on an
interactive Maptoolkit map.