Yatmo neighbourhood data for real estate MCP server

Yatmo MCP server: neighbourhood data for AI assistants and agents

Yatmo exposes its neighbourhood intelligence to AI clients through a remote MCP server. Connect Claude, Cursor, VS Code or your own agent, and it can answer "what is around this property?" with real data: nearest school, nursery, supermarket, public transport, train station, motorway access, distances with travel times, and a quotable summary paragraph, in 25 countries.

This repository holds ready-to-copy client configurations and a dependency-free quick start. The server itself is operated by Yatmo at https://mcp.yatmo.com/mcp/v1 (Streamable HTTP). Full documentation: documentation.yatmo.com/mcp.

The four tools

All tools take latitude, longitude, country (BE, FR, UK...) and language, and return the same envelope: text (a summary the assistant can quote), facts (structured data), warnings and metadata. Categories use one vocabulary everywhere: school, nursery, supermarket, publictransport, trainstation, motorway. Details: tools reference.

Authentication

Every request carries your Yatmo backend key in the LicenceKey header (LicenseKey is accepted too). Keep the key in the client's secret store, never in a public web page: the key grants access to your Yatmo account. Get a key at yatmo.com; keys explained.

Connect a client

Generic remote-MCP configuration (configs/mcp.json), accepted by most clients, sometimes under a slightly different parent key:

{
  "mcpServers": {
    "yatmo": {
      "url": "https://mcp.yatmo.com/mcp/v1",
      "headers": { "LicenceKey": "YOUR_KEY" }
    }
  }
}

Claude.ai and Claude Desktop: Settings > Connectors > Add custom connector, URL https://mcp.yatmo.com/mcp/v1, custom header LicenceKey: YOUR_KEY. The four tools appear once the connector is enabled.

Claude Code:

claude mcp add --transport http yatmo https://mcp.yatmo.com/mcp/v1 --header "LicenceKey: YOUR_KEY"

Cursor: copy configs/cursor/mcp.json to .cursor/mcp.json in your project (or to ~/.cursor/mcp.json for every project).

VS Code (Copilot agent mode): copy configs/vscode/mcp.json to .vscode/mcp.json.

Your own agent (Anthropic, OpenAI or any LLM SDK): use an MCP client library (TypeScript, Python, .NET, Go) with the Streamable HTTP transport, add the LicenceKey header to the transport, and forward the four tool definitions to the model. quickstart.py shows the raw protocol without any library.

Try it from the command line

export YATMO_KEY=your_backend_key
python3 quickstart.py          # initialize, list the tools, call yatmo_get_location_summary on a Brussels address
sh quickstart.sh               # the same three calls with curl

Prompts that work well

Shortened. The full README is on GitHub.