MCP Server
Connect ScrapeMapsAPI to AI assistants like Claude, Cursor and Claude Code through the Model Context Protocol.
Connect to Claude or ChatGPT (no install)
Add ScrapeMapsAPI as a custom connector with this URL, then sign in with your ScrapeMapsAPI account and click Allow. No API key to copy, nothing to install, and it works on the web, desktop and mobile apps.
https://api.scrapemapsapi.com/mcp- Claude (claude.ai, Desktop, mobile): Customize → Connectors → + Add → Add custom connector, paste the URL, then sign in and allow access. In a chat, turn ScrapeMapsAPI on under + → Connectors.
- ChatGPT: open chatgpt.com/plugins on the web, click + → Add custom MCP server, paste the URL as Server URL, choose OAuth authentication, create it, then sign in and allow access. Your account or workspace must allow custom MCP servers.
- Claude Code:
claude mcp add --transport http scrapemapsapi https://api.scrapemapsapi.com/mcp, then/mcpto sign in.
Calls use your credits exactly like the API (1 credit per successful call); failed calls are free. See or disconnect your apps in Dashboard → Connected apps. Any MCP client that supports remote servers with OAuth (Streamable HTTP) can use the same URL.
What is MCP?
The Model Context Protocol (MCP) lets AI assistants call external tools. With the ScrapeMapsAPI MCP server your assistant can search Google Maps, read place details and analyse every review of a place, billed to your API key like any other call.
Setup
Two equivalent packages: scrapemapsapi-mcp on PyPI (run with uvx) and scrapemapsapi-mcp on npm (run with npx). Both read SCRAPEMAPSAPI_API_KEY.
Claude Desktop
Add this to claude_desktop_config.json and restart Claude Desktop.
{
"mcpServers": {
"scrapemapsapi": {
"command": "uvx",
"args": ["scrapemapsapi-mcp"],
"env": {
"SCRAPEMAPSAPI_API_KEY": "key_your_api_key"
}
}
}
}Cursor
Add the same block to ~/.cursor/mcp.json (or .cursor/mcp.json in a project).
{
"mcpServers": {
"scrapemapsapi": {
"command": "uvx",
"args": ["scrapemapsapi-mcp"],
"env": {
"SCRAPEMAPSAPI_API_KEY": "key_your_api_key"
}
}
}
}Claude Code
claude mcp add scrapemapsapi -e SCRAPEMAPSAPI_API_KEY=key_your_api_key -- uvx scrapemapsapi-mcp
# or with Node
claude mcp add scrapemapsapi -e SCRAPEMAPSAPI_API_KEY=key_your_api_key -- npx -y scrapemapsapi-mcpAvailable Tools
Once connected, your AI assistant can use these tools:
search_placesSearch Google Maps by text, optionally near a point (1 credit per call, up to 20 places)get_placeDetails of a place: address, hours, contacts, attributes (1 credit)get_reviewsReviews with sorting, language filter, translation and cursor paging (1 credit per page of up to 20)resolve_placeTurn a Google Maps link or any identifier into a place ID (1 credit)Review authors are returned as pseudonymous IDs only, so personal data never reaches the model.
Example Prompts
Once the MCP server is connected, try asking your AI:
“Find the five best-rated trattorias near Santa Croce in Florence”
“Summarize the last 100 reviews of https://maps.app.goo.gl/x7Qp2Nf9TzVb8LkE6: what do people complain about?”
“Compare the opening hours and ratings of these three restaurants”
“Which reviews of this place mention the service, and how did the owner reply?”
“Translate the Italian reviews of this place and group them by topic”
Costs and limits
The MCP server runs on your computer and calls the same API as your code, with your key: every tool call is billed like the matching API call, and errors cost nothing. When a tool fails, the assistant gets a short message that says what to do next, such as topping up credits or resolving a link first, instead of a raw error.
Review texts are shortened for the model, so a long list of reviews fits in its context. For bulk exports, use the SDKs or the Reviews API directly.