maps-fetcher-mcp
maps-fetcher-mcp
MCPサーバー + CLI: リンクから任意のGoogleマップの場所を取得 — 完全なビジネスメタデータに加え、すべてのギャラリー写真と動画を最大解像度で取得します。
インストール
pip install maps-fetcher-mcp # or: uv pip install maps-fetcher-mcp
maps-fetcher-setup # one-time: browser download (~120MB)Related MCP server: Local Business Data MCP Server
MCPサーバーとして使用(任意のMCPクライアント)
{
"mcpServers": {
"maps_fetcher": {
"command": "maps-fetcher-mcp"
}
}
}対応環境: Hermes Agent、Claude Desktop、Cursor、opencode、Cline、Zed — MCP(stdio)を話せるものなら何でも動作します。
CLIとして使用
maps-fetcher "https://maps.app.goo.gl/XXXX" # everything
maps-fetcher "https://maps.app.goo.gl/XXXX" --include info # metadata only (~20s)
maps-fetcher "https://maps.app.goo.gl/XXXX" --include photos
maps-fetcher "https://maps.app.goo.gl/XXXX" --include videos
maps-fetcher --list # what's been fetched公開ツール(MCP)
ツール | 目的 | 標準的な時間 |
| メタデータのみ: 名前、電話番号、評価、住所、営業時間、座標 | ~20秒 |
| メディアのみ( | 40〜220秒 |
| 一度にすべて取得( | 60〜220秒 |
| 以前に取得した場所 | 即時 |
同じ場所を24時間以内に再取得すると、スクレイピングを再利用します(~2秒)。
対応内容
短縮リンクの解決と正規化; 検索リンクはきれいに拒否
写真に偽装されたビデオタイル; 音声とミックスされたDASH/HLSマニフェスト
期限切れURL、アバターのジャンク、モーションブラー枠、知覚的類似画像 (削除ではなく隔離)、連番の欠落のないファイル番号
非決定的なギャラリーの遅延読み込み: スクレイピングを再試行してマージ
ボット壁検出(reCAPTCHA / 異常なトラフィック)と明確なエラー; Patchrightによるステルスブラウザ、Playwrightフォールバック
セキュリティ: すべてのメディアダウンロードに対するgoogleホスト許可リスト、ファイルごとおよび 場所ごとのサイズ上限、解凍爆弾ガード、シークレットなし
出力
<root>/<place-slug>/
photos/photo_01.jpg ... videos/video_01.mp4 ... quarantine/
info_pw.json contact_sheet.jpgデフォルトのルート: /mnt/d/Projects/maps-data-fetcher/places(WSLレイアウト;
output_dir または MAPS_FETCHER_ROOT で上書き可能)。
制限
非公式のスクレイピング — 低ボリュームでのバッチ収集に適しています。Googleの マークアップ変更で壊れる可能性があります(修正は通常小規模)。高ボリュームの 本番用ではありません。その場合は有料のGoogle Places APIを使用してください。CAPTCHAは 検出して報告しますが、自動解決はしません。
開発
ソースレイアウト: maps_fetcher/core.py(エンジン)、mcp_server.py(MCP
ツール)、cli.py(CLI)、setup.py(ブラウザインストールヘルパー)。
Available Tools
4 toolsfetch_maps_placeA
Fetch a Google Maps place by link (maps.app.goo.gl/... or full google.com/maps URL). Downloads gallery media at max resolution plus business metadata (name, phone, website, rating, address, plus code, hours, coordinates).
Args: url: Google Maps place link (single place only, not search links). output_dir: optional root directory (default: /mnt/d/Projects/maps-data-fetcher/places). A per-place subfolder is created with photos/, videos/, info_pw.json, contact_sheet.jpg. include: "all" (default) | "info" (metadata only, fast) | "photos" | "videos" | "media" (photos+videos). meta_only: deprecated alias for include="info".
Returns JSON summary: place data, media counts, output paths. Typical run: info ~20-40s, full fetch 60-220s.
Inputs: url (required), include, meta_only, output_dir. Returns: result
get_place_infoA
Fast metadata-only fetch: name, phone, website, rating, address, plus code, hours, coordinates. No media downloaded. Use when the agent only needs business info. ~20-40s.
Inputs: url (required), output_dir. Returns: result
get_place_mediaA
Media-only fetch: downloads gallery photos and/or videos, skips verbose metadata in the response. Use when the agent only needs the images/videos.
Args: kind: "both" (default) | "photos" | "videos". Returns JSON: place name, counts, output dirs.
Inputs: url (required), kind, output_dir. Returns: result
list_maps_placesC
List previously fetched places under the output root with photo/video counts and metadata.
Inputs: output_dir. Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 4 tools
There is significant overlap: fetch_maps_place with include='info' does exactly what get_place_info does, and with include='media' does what get_place_media does. While the descriptions clarify the intended use cases, an agent may struggle to pick the right tool since fetch_maps_place is a superset. The tools are not clearly distinct in purpose.
The naming mixes conventions: 'get_place_info' and 'get_place_media' follow a get_place_N pattern, but 'fetch_maps_place' uses a different verb (fetch) and noun order, and 'list_maps_places' uses plural. The pattern is not consistent across all tools, though the tools are readable.
At 4 tools, the server is well-scoped for a maps fetcher. However, the presence of get_place_info and get_place_media is redundant given fetch_maps_place can achieve the same via its include parameter, suggesting the count could be trimmed to 2–3 tools without losing functionality.
The core lifecycle for fetching Google Maps place data is covered: retrieving metadata, media, or both, plus listing previously fetched places. There are no obvious dead ends for the stated purpose. Missing delete/update operations are not essential for a fetcher, so the surface feels complete.
Maintenance
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceEnables web searching via Google Search and AI Mode, plus advanced web scraping capabilities including content extraction in multiple formats, link extraction, and batch scraping of multiple URLs.14 npmMIT
- AlicenseBqualityDmaintenanceEnables access to Google Maps business data including search, reviews, photos, and geocoding. Supports searching businesses by location, area, or coordinates, retrieving detailed business information, reviews, and performing reverse geocoding operations.13MIT
- FlicenseNot gradedqualityDmaintenanceEnables advanced web crawling and content extraction with JavaScript support, AI-powered analysis, PDF/Office document processing, YouTube transcript extraction, Google search integration, and multi-format data export capabilities.2-
- FlicenseNot gradedqualityDmaintenanceEnables AI assistants to search and scrape Google Maps places data (name, rating, address, etc.) directly without an API key.1-