Google Maps Grounding MCP

该 Skill 通过 mcporter 配置并调用 Google Maps Grounding Lite MCP，提供 search_places、lookup_weather、compute_routes 三个工具，用于地点/商家/地址搜索、天气查询以及驾车或步行路线距离与耗时估算。需要 GOOGLE_MAPS_API_KEY，输出需包含 Google Maps 链接。

ryanbaumann

热度 96.11

最近更新 2026-05-11 15:50:57

分类：生活依赖：代码执行 / MCP 服务 / 第三方 API应用场景：旅行 / 推荐

安装技能

复制下面的话术发给你的 Agent，它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill： npx secure-skills-cli add google-maps-grounding-lite-mcp

Metadata

name: grounding-lite
description: Google Maps Grounding Lite MCP for location search, weather, and routes via mcporter.
homepage: https://developers.google.com/maps/ai/grounding-lite
metadata: {"clawdbot":{"emoji":"🗺️","requires":{"bins":["mcporter"],"env":["GOOGLE_MAPS_API_KEY"]},"primaryEnv":"GOOGLE_MAPS_API_KEY","install":[{"id":"node","kind":"node","package":"mcporter","bins":["mcporter"],"label":"Install mcporter (npm)"}]}}

Grounding Lite

Google Maps Grounding Lite MCP provides AI-grounded location data. Experimental (pre-GA), free during preview.

Setup

Enable the API: gcloud beta services enable mapstools.googleapis.com
Get an API key from Cloud Console
Set env: export GOOGLE_MAPS_API_KEY="YOUR_KEY"

Configure mcporter:

mcporter config add grounding-lite \
  --url https://mapstools.googleapis.com/mcp \
  --header "X-Goog-Api-Key=$GOOGLE_MAPS_API_KEY" \
  --system

Tools

search_places: Find places, businesses, addresses. Returns AI summaries with Google Maps links.
lookup_weather: Current conditions and forecasts (hourly 48h, daily 7 days).
compute_routes: Travel distance and duration (no turn-by-turn directions).

Commands

# Search places
mcporter call grounding-lite.search_places textQuery="pizza near Times Square NYC"

# Weather
mcporter call grounding-lite.lookup_weather location='{"address":"San Francisco, CA"}' unitsSystem=IMPERIAL

# Routes
mcporter call grounding-lite.compute_routes origin='{"address":"SF"}' destination='{"address":"LA"}' travelMode=DRIVE

# List tools
mcporter list grounding-lite --schema

Parameters

search_places: textQuery (required), locationBias, languageCode, regionCode

lookup_weather: location (required: address/latLng/placeId), unitsSystem, date, hour

compute_routes: origin, destination (required), travelMode (DRIVE/WALK)

Notes

Rate limits: search_places 100 QPM (1k/day), lookup_weather 300 QPM, compute_routes 300 QPM
Include Google Maps links in user-facing output (attribution required)
Only use with models that don't train on input data