Google Maps Mcp Server

Google Maps Platform MCP server

This uses the googlemaps Python package (which hasn't been updated since 2023, but does appear to still be active)

• documentation

Setup

For this example, you will need a Google Maps Platform API key. I restricted my APIs to the following:

• Directions API
• Distance Matrix API
• Geocoding API
• Places API

You may need to expand on this if you wish to add more functionality. Here is the Google Maps Platform API Picker to help you decided what you need. You will also need an set an additional environment variable, GOOGLE_MAPS_API_KEY.

Of course, there will be a cost to using the Maps API as well. It's got a pretty generous free-tier, depending on what you're using, just don't go crazy with it ;)

Running the FastMCP server

fastmcp run server.py --transport stdio

You can substitute stdio with streaming-http or sse (deprecated) depending on your deployment method.

NOTE: if you use the fastmcp option, it ignores the if __name__ == "__main__", so you need to pass --transport and any other settings you wish to override.

Unit tests

pytest -v tests/

Client test

Run server with:

fastmcp run server.py --transport streamable-http

Run client with:

python tests/client_list_tools.py

• sample output:

[Tool found]: get_directions
[Tool found]: get_distance
[Tool found]: get_geocode
[Tool found]: find_place
[Tool found]: place_nearby
[Tool found]: place_details

How to use/connect

Agent Development Kit (ADK)

model="gemini-2.5-flash"
google_maps_api_key=os.getenv("GOOGLE_MAPS_API_KEY")

maps_agent = LlmAgent(
    name="maps_agent",
    model=model,
    instruction=(
        "You are an AI Google Maps assistant. Your primary function is to find places that meet the user's criteria. Include with your findings the name of the place, its rating, and its address. Limit the results to a maximum of 10. Use only the tools provided to you."
    ),
    tools=[
        MCPToolset(
            connection_params=StdioServerParameters(
                command='fastmcp',
                args=[
                    "run",
                    "maps_agent/server.py"
                ],
                env={
                    "GOOGLE_MAPS_API_KEY": google_maps_api_key
                },
            ),
            #tool_filter=['tool1', 'tool2']
        )

    ],
)

And here's a more detailed prompt example I give my agent:

MAPS_AGENT_PROMPT="""
System Role: You are an AI Google Maps assistant. Your primary function is to find places that meet the user's criteria. You achieve this by finding a list of places within a 4km radius of the user's location unless otherwise specified. Include with your findings the name of the place, its rating, and its address. Limit the results to a maximum of 10.

When asked for directions, provide the route we will be travelling via, total distance, total time and step-by-step directions in a list format.

If the user asks for "within walking distance", any nearby places returned should be within a 500m radius of the origin location.

Use only the tools provided to you.
"""

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: Neutrollized
• Source: Neutrollized/google-maps-mcp-server
• License: Apache-2.0
• Homepage: https://medium.com/@glen.yu/a-google-maps-platform-mcp-server-in-python-featuring-google-labs-jules-c070ae6156ac

Install and usage instructions live in the source repository linked above.