Home › Blog › Scrapingdog Google Search API

Scrapingdog Google Search API for AI agents: how to use it

Updated 2026-10-25

This page has affiliate links, each marked "Affiliate link, #ad". We may be paid if you sign up, at no extra cost to you. See our affiliate disclosure.

The Scrapingdog Google Search API returns Google results as structured JSON, so an AI agent can search without scraping Google itself. Send a query and API key to https://api.scrapingdog.com/google, or connect Scrapingdog's MCP server to Claude Code or Cursor. Here are the parameters, examples and setup from Scrapingdog's own docs.

What is the Scrapingdog Google Search API?

It is an API that runs a Google search and returns the results as structured JSON: organic results, related searches, pagination and search information. Your agent sends a query to https://api.scrapingdog.com/google with an API key, and Scrapingdog handles proxies and parsing, so you never scrape Google's HTML yourself. (Affiliate link, #ad: we may be paid if you sign up, at no extra cost to you.)

Why would an AI agent need a Google search API?

Models only know what they were trained on. A search API lets your agent check what ranks today, find fresh sources, track competitors or pull leads from search results, all as clean fields it can reason over. Structured JSON also costs less context than raw HTML, which matters when an agent runs many searches in one task.

Related: agent web search tools trend

Related: Scrapingdog in the directory

How do you call the Scrapingdog Google Search API?

Send a GET request to the endpoint with your api_key and a query. Add country for the market, page to move through results and language for the interface language. The query supports Google operators such as site:, inurl: and intitle:, so your agent can narrow searches the same way you would.

curl, from Scrapingdog's docs (deprecated results parameter removed):

curl "https://api.scrapingdog.com/google/?api_key=APIKEY&query=football&country=us&page=0"

Python, from Scrapingdog's docs (deprecated results parameter removed):

import requests

params = {
    "api_key": "APIKEY",
    "query": "football",
    "country": "us",
    "page": "0"
}

response = requests.get("https://api.scrapingdog.com/google/", params=params)

if response.status_code == 200:
    print(response.json())
else:
    print(f"Request failed with status code: {response.status_code}")

Which parameters matter most?

Start with query, country and page. Use location for city-level results, but not together with uule, which sets location another way. Add tbs for time filters, safe for SafeSearch and mob_search for mobile results. deep_scrape also fetches the pages behind the organic links. Scrapingdog marks the old results parameter as deprecated.

How do you connect Scrapingdog to Claude Code or Cursor with MCP?

Scrapingdog publishes an MCP server you start with npx. In Claude Code, one claude mcp add command registers it with your API key as an environment variable. In Cursor or Claude Desktop, paste Scrapingdog's JSON config. The key stays in the environment, out of the model's context.

1. Get an API key. Sign in to Scrapingdog and copy your API key from the dashboard.Get a Scrapingdog API key → Affiliate link, #ad: we may be paid if you sign up, at no extra cost to you.
2. Register the server. Run the claude mcp add command below with your key in place of your-key-here.
3. Confirm it is live. Run claude mcp list and check that scrapingdog shows as connected.
4. Search in plain English. Start Claude Code and ask for what you need, such as the top Google results for a query in a given country.

Claude Code, from Scrapingdog's MCP announcement:

claude mcp add scrapingdog --scope user --env SCRAPINGDOG_API_KEY=your-key-here -- npx -y scrapingdog-mcp
claude mcp list

Cursor / Claude Desktop config, from Scrapingdog's SDKs page:

{
  "mcpServers": {
    "scrapingdog": {
      "command": "npx",
      "args": ["-y", "scrapingdog-mcp"],
      "env": { "SCRAPINGDOG_API_KEY": "APIKEY" }
    }
  }
}

Related: best MCP servers for Claude Code

Does Scrapingdog have SDKs too?

Yes. Scrapingdog's SDKs page lists a Node package, installed with npm install scrapingdog-sdk, and a Python package, installed with pip install scrapingdog. Both are MIT-licensed open source. Use an SDK when you write the agent's code yourself, and the MCP server when Claude Code or Cursor should call Scrapingdog on its own.

Related: what an MCP server is

What does the response look like?

JSON with search_information for the query, organic_results for the ranked links with titles and snippets, relatedSearches for related queries and pagination for moving to the next page. Set html to true to get the full Google page HTML as well, which you rarely need for an agent.

How should your agent use the results?

Have it read organic_results first and keep only the fields it needs, such as title, link and snippet, before passing them to the model. For deeper research, follow the best links with a page scraper or deep_scrape. Keeping raw HTML out of the prompt keeps answers faster and cheaper on context (see MCP context cheat sheet).

When should you pick Scrapingdog over other scraping APIs?

Pick Scrapingdog when your agent mainly needs Google results as structured data, along with its other dedicated endpoints for sites like e-commerce stores. For single pages that block plain requests, ScraperAPI is our featured pick. (Affiliate link, #ad: we may be paid if you sign up, at no extra cost to you.) For search plus clean Markdown, look at Firecrawl (see Firecrawl MCP server).

Can you use more than one?

Yes, and many agents do. A common split is Scrapingdog for finding pages on Google and a page scraper for reading them. Our comparison guide shows which tool handles Google results, blocked pages and Markdown output, so you can give each job to the tool built for it.

Related: ScraperAPI vs Scrapingdog vs Firecrawl · Exa MCP

Related: best web scraping API for AI agents · search tools in the directory · ScraperAPI MCP server

See how we run our own agent crew in the playbook.

Building on Grok Bot? Start a crew from the DartzDabz Crew Builder template.

We may be paid by X based on how many people use our Crew Builder template.

See real agent setups you can copy, browse the agent tool directory, or submit a tool; listings are free.