Overview

Google runs an official, Google-managed MCP server for Google Maps as part of the Maps Grounding Lite API. It is hosted by Google at a single remote endpoint — no self-hosting, no local builds. You connect any MCP client directly to https://mapstools.googleapis.com/mcp and give your AI assistants access to places search, weather, route computation, and map URL resolution.

The server is listed in github.com/google/mcp (the catalog of Google's official MCP servers, 4.5k stars) and is generally available (GA) as a Google Cloud remote MCP server.

Hosted by Google:

This is a remote MCP server managed by Google — the endpoint and tool surface are maintained by Google, with an SLA and platform status page. You only manage an API key, not the server.

Key Features

🔍

Place Search

Find places, businesses, addresses, and points of interest with text queries and location bias

🌤️

Weather Lookup

Current conditions, hourly and daily forecasts, precipitation, UV index, and wind

🗺️

Route Computation

Driving and walking routes between any origin and destination

🔗

Map URL Resolution

Resolve shared map links and place names into canonical Google Maps Place IDs

Available Tools

The mapstools.googleapis.com MCP server exposes five tools:

Tool	Purpose
`search_places`	Search for places, businesses, addresses, and points of interest
`lookup_weather`	Current conditions, hourly and daily forecasts for a location
`compute_routes`	Compute driving or walking routes between origin and destination
`resolve_names`	Resolve a batch of place names/addresses into canonical Place IDs
`resolve_maps_urls`	Resolve shared Google Maps URLs into canonical Place IDs

Example — Places Search

{
  "text_query": "restaurants in San Francisco",
  "location_bias": {
    "circle": {
      "center": { "latitude": 37.7749, "longitude": -122.4194 },
      "radius_meters": 5000
    }
  }
}

Example — Compute Route

{
  "origin": { "address": "Eiffel Tower, Paris" },
  "destination": { "place_id": "ChIJt_5xIthw5EARoJ71mGq7t74" },
  "travel_mode": "DRIVE"
}

Server Setup

This is a Google Cloud remote MCP server. Before using it:

Enable MCP servers for your Google Cloud project
Set up authentication (API key or OAuth)
Ensure the Google Maps Platform services you use (e.g. Places API) are enabled and billing is active

Client configuration points to the remote endpoint:

{
  "mcpServers": {
    "google-maps": {
      "url": "https://mapstools.googleapis.com/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_GOOGLE_MCP_TOKEN"
      }
    }
  }
}

Attribution Required:

Results from this server are grounded in Google Maps data — the response includes an attribution field that must be shown to users, per the Maps Grounding Lite attribution guidelines.

Use Cases

Travel Planning

Ask AI to find hotels, restaurants, and attractions near a destination, then compute walking routes between them.

Local Discovery

Natural-language place discovery with location bias — "pet-friendly parks in Manhattan" or "date night restaurants in Chicago".

Weather-Aware Agents

Combine current conditions and forecasts with route planning for delivery, logistics, or travel assistants.

Link & Place Resolution

Normalize shared map links or landmark names into canonical Place IDs for downstream mapping workflows.

Legacy Reference Server Archived:

The original reference server in the Anthropic/MCP servers collection (@anthropic-ai/mcp-server-google-maps) has been archived in modelcontextprotocol/servers. Google's hosted Maps Grounding Lite server is the current, supported option. For a community-run alternative with a broader tool set, see the Google Maps MCP (npx) page.

Google Maps MCP Server: Google's Official MCP Integration