mcp-google-map MCP — Cas d'usage, Installer & Démo en direct

Pourquoi l'utiliser

Fonctionnalités clés

14 atomic tools: geocode, reverse-geocode, directions, distance-matrix, places-search, place-details, elevation, timezone, weather, air-quality, static-maps, and more
4 composite tools for one-shot exploration, routing, comparison, and SEO rank-tracking
Tool subset filtering via GOOGLE_MAPS_ENABLED_TOOLS — trim context cost
stdio / HTTP / CLI transports
Runs with npx @cablate/mcp-google-map

Démo en direct

Aperçu en pratique

google-map.replay ▶ prêt

0/0

Installer

Choisissez votre client

~/Library/Application Support/Claude/claude_desktop_config.json · Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "google-map": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-google-map"
      ],
      "_inferred": true
    }
  }
}

Ouvrez Claude Desktop → Settings → Developer → Edit Config. Redémarrez après avoir enregistré.

~/.cursor/mcp.json · .cursor/mcp.json

{
  "mcpServers": {
    "google-map": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-google-map"
      ],
      "_inferred": true
    }
  }
}

Cursor utilise le même schéma mcpServers que Claude Desktop. La config projet l'emporte sur la globale.

VS Code → Cline → MCP Servers → Edit

{
  "mcpServers": {
    "google-map": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-google-map"
      ],
      "_inferred": true
    }
  }
}

Cliquez sur l'icône MCP Servers dans la barre latérale Cline, puis "Edit Configuration".

~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "google-map": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-google-map"
      ],
      "_inferred": true
    }
  }
}

Même format que Claude Desktop. Redémarrez Windsurf pour appliquer.

~/.continue/config.json

{
  "mcpServers": [
    {
      "name": "google-map",
      "command": "npx",
      "args": [
        "-y",
        "mcp-google-map"
      ]
    }
  ]
}

Continue utilise un tableau d'objets serveur plutôt qu'une map.

~/.config/zed/settings.json

{
  "context_servers": {
    "google-map": {
      "command": {
        "path": "npx",
        "args": [
          "-y",
          "mcp-google-map"
        ]
      }
    }
  }
}

Ajoutez dans context_servers. Zed recharge à chaud à la sauvegarde.

claude mcp add google-map -- npx -y mcp-google-map

Une seule ligne. Vérifiez avec claude mcp list. Supprimez avec claude mcp remove.

Cas d'usage

Usages concrets : mcp-google-map

Track your Google Maps ranking for key searches across zip codes

👤 Local SEO specialists, restaurant / service businesses ⏱ ~20 min intermediate

Quand l'utiliser : You want to know where your business ranks for 'best pizza' in 20 zip codes around town, refreshed weekly.

Prérequis

Google Maps API key with Places (New) enabled — Google Cloud Console > Enable APIs

Déroulement

Use the composite local-rank-tracker tool

Run local-rank-tracker for business 'Tony's Pizza' (place_id: [id]) on query 'pizza near me' across these 20 coordinates [list lat/lng].✓ Copié

→ Per-location rank (or 'not in top 20')
Store history

Save results to ./rank-history/YYYY-MM-DD.json. Diff against last week.✓ Copié

→ Week-over-week delta

Résultat : Weekly local rank report without paying a SaaS SEO tool.

Pièges

Google's Places API cost adds up — ~$17 per 1000 calls — Batch the locations; limit to key zip codes; cache for 7 days

Combiner avec : filesystem

Plan a multi-stop day trip with driving time and elevation

👤 Travelers, travel bloggers ⏱ ~10 min beginner

Quand l'utiliser : Planning a road trip and want optimized stop order + realistic drive times.

Déroulement

Describe the stops

I want to visit Yosemite Valley, Mariposa Grove, Tunnel View, Glacier Point — in one day starting from Merced. Optimize the order and give drive times.✓ Copié

→ Ordered stops + per-leg drive minutes
Add context

For each stop, get elevation and the current weather.✓ Copié

→ Annotated itinerary

Résultat : A sane itinerary with real-world durations.

Pièges

Routes API uses traffic-aware estimates only at query time — Re-query the morning of the trip for realistic drive times

Compare candidate locations for a new retail store

👤 Real estate analysts, franchise planners ⏱ ~30 min intermediate

Quand l'utiliser : You're deciding between 3 candidate addresses for a new storefront. Foot traffic, competitor proximity, demographic proxies matter.

Déroulement

Reverse-geocode each candidate

For these 3 addresses, get coordinates and neighborhood.✓ Copié

→ 3 geocoded results
Use compare-places

For each location, count competitors within 1km (category: coffee shop). List top 10 nearby businesses.✓ Copié

→ Per-site competitive density + anchor tenants
Assess traffic corridors

For each, get drive-time from the nearest highway exit and major residential zip code.✓ Copié

→ Accessibility score

Résultat : Data-backed site recommendation you can defend in a meeting.

Pièges

Google Places is not a substitute for actual foot-traffic data — Use as a first-pass filter; combine with Placer.ai or county permit data for final decision

Combinaisons

Associez-le à d'autres MCPs pour un effet X10

google-map + filesystem

Store weekly rank-tracker snapshots to disk, diff over time

Run local-rank-tracker for my business, save to /reports/rank/YYYY-MM-DD.json.✓ Copié

google-map + postgres

Enrich customer addresses in your DB with lat/lng for map dashboards

For each row in customers where lat IS NULL, geocode the address and UPDATE with lat/lng. Stop at 500/day to respect quota.✓ Copié

Outils

Ce que ce MCP expose

Outil	Entrées	Quand appeler	Coût
geocode	address: str	Address → coordinates	$5 / 1000 calls
reverse_geocode	lat, lng	Coordinates → address	$5 / 1000
directions	origin, destination, mode?, waypoints?	Turn-by-turn or drive-time	$5-10 / 1000
places_search	query, location?, radius?	Find businesses by keyword near a point	$17-32 / 1000 (New Places)
place_details	place_id	After places_search to get hours, phone, website	$17+ / 1000
explore-area	center, radius, categories[]	Composite — one shot neighborhood survey	bundled
plan-route	stops[], mode, optimize?: bool	Multi-stop trip planning	bundled
compare-places	places[], categories[]	Site selection or evaluation	bundled
local-rank-tracker	query, place_id, locations[]	Local SEO audits	N * places_search cost

Coût et limites

Coût d'exécution

Quota d'API: Google: $200/mo free credit covers ~12k geocodes or ~10k Places (New) Autocomplete/Details
Tokens par appel: Tool results 500-3000 tokens; trim via GOOGLE_MAPS_ENABLED_TOOLS to reduce schema overhead
Monétaire: Pay Google directly. See cloud.google.com/maps-platform/pricing.
Astuce: Places (New) is the expensive one. Use geocode (cheap) when you just need coordinates, not business data. Cache place_id aggressively.

Sécurité

Permissions, secrets, portée

Portées minimales : Enable only the APIs you actually use

Stockage des identifiants : GOOGLE_MAPS_API_KEY env var. Restrict the key to your IP / referrer.

Sortie de données : All queries to maps.googleapis.com

Ne jamais accorder : Unrestricted key on a shared machine — anyone can burn your quota

Google bills monthly — set a budget alert in GCP Billing before enabling APIs.
Places (New) supersedes legacy Places; make sure you enabled the (New) version.

Dépannage

Erreurs courantes et correctifs

REQUEST_DENIED / API key not valid

API isn't enabled or key restrictions block your IP. Check GCP Console > APIs & Services > Credentials.

Vérifier : curl 'https://maps.googleapis.com/maps/api/geocode/json?address=NYC&key=$KEY'

OVER_QUERY_LIMIT

You've blown the $200 monthly free tier, or Google is throttling per-second. Back off + batch.

Vérifier : GCP Billing > Reports

Empty results for obvious queries

Places (New) requires explicit field masks on some calls. Update to latest npm version.

Alternatives

mcp-google-map vs autres

Alternative	Quand l'utiliser	Compromis
OpenStreetMap Nominatim MCP	Free, no API key, non-commercial use OK	Less rich data; rate-limited to 1 rps
Mapbox MCP	You prefer Mapbox's styling and routing	Different pricing and data model

Plus

Ressources

📖 Lire le README officiel sur GitHub

🐙 Voir les issues ouvertes

🔍 Parcourir les 400+ serveurs MCP et Skills