Google Scraping Suite Mcp Server

A high-performance Model Context Protocol (MCP) server connecting Claude Desktop, Cursor, Antigravity IDE, Windsurf, and any MCP-compatible AI agent directly to the suite of 9 specialized Google scraping, intelligence, news, and AI Actors powered by Apify.

• Instant GitHub Launch (stdio - Recommended): Run directly from this public repository via uvx with zero cloning, zero manual venv management, and native low-latency stdio transport.
• Zero Scraping Maintenance: No proxy rotations, headless browsers, or anti-bot bypasses to manage.

---

What is Google Suite MCP Server?

Google Suite MCP Server provides an AI-native interface to Google's public data ecosystem. Rather than maintaining custom web scrapers, residential proxies, or headless browsers, your AI assistant can directly query live Google services using standard natural language prompts:

• Live AI Overviews: Retrieve direct AI Mode answers, synthesized overviews, and source references.
• Comprehensive Search SERP: Access organic search rankings, Knowledge Graph cards, and People Also Ask questions.
• Full-Text News Intelligence: Crawl and extract clean, paywall-free full articles from global news sources.
• Trends & Cultural Momentum: Query real-time search trends enriched with Gemini marketing intelligence, plus historical multi-year interest graphs.
• Local Business Data: Extract Google Maps merchant profiles, coordinates, phone numbers, and ratings at scale.
• Travel & Commerce: Query live Google Flights airfares and Google Shopping merchant listings and pricing.
• Fast Generative Intelligence: Execute high-speed summarization, data extraction, and multi-modal tasks via Gemini Flash Lite.

---

The 9 Exposed Tools

Category	Tool Name	Apify Actor	Description	Example Prompt	Default Timeout
Search	google_search_ai_mode	ilborso/google-search-ai-mode	Synthesized AI answers, text overviews, and source citations from Google AI Mode.	"What are the latest breakthroughs in solid-state batteries according to Google AI?"	50s
Search	google_search_serp	ilborso/google-search-fast-complete-result	Comprehensive Google SERP: organic rankings, Knowledge Graph, and People Also Ask.	"Find the top 5 ranked articles for quantum computing frameworks."	50s
News	google_news_content_crawler	ilborso/ai-news-content-crawler	Clean, full-text news articles and content from Google News without paywalls or subscriptions.	"Extract the full text of today's top 3 news stories about semiconductor manufacturing."	50s
Intelligence	google_trending_now	ilborso/google-trending-now	Real-time trending searches with optional Gemini marketing intelligence enrichment.	"Show me what is trending right now in Italy with AI marketing analysis."	50s
Intelligence	google_trending_historical	ilborso/google-trending-historical-data	Historical Google Trends: interest over time, regional interest, and related queries.	"Compare search interest trends between Python and Rust over the last 12 months."	50s
Intelligence	google_maps_search	ilborso/google-maps-fast-scraper	Fast extraction of Google Maps local business data, ratings, addresses, and contacts.	"Find top rated Italian restaurants in Zurich with their address and rating."	50s
Commerce	google_flights_search	ilborso/google-flights-api	Live flight fares, airlines, layovers, durations, and carbon emission insights.	"Find one-way flights from Rome (FCO) to New York (JFK) on November 20."	50s
Commerce	google_shopping_search	ilborso/google-shopping-now	Real-time product search, merchant pricing, reviews, and discount tracking.	"Search for Sony WH-1000XM5 headphones and list prices across different sellers."	50s
AI	gemini_flash_lite_generate	ilborso/gemini-flash-lite-latest	High-speed text generation, summarization, and data extraction using Gemini Flash Lite.	"Summarize this technical specification into 3 concise bullet points."	50s

---

How to Connect in 3 Steps

Connecting your AI assistant to the Google Suite MCP Server takes less than two minutes.

Step 1: Obtain Your Apify API Token

1. Sign in to your Apify Console. If you do not have an account, you can create one for free.
2. Navigate to Settings -> Integrations.
3. Copy your personal API Token.

Step 2: Add the Server to Your MCP Client

Method 1: Local stdio from GitHub via uvx (Recommended ⭐)

The recommended connection method uses uvx to run the server over standard input/output (stdio). uvx automatically fetches the code from GitHub, installs dependencies in an isolated cache, and runs the server at maximum speed with zero manual maintenance:

{
  "mcpServers": {
    "google-suite": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/ilborso/google-suite-mcp-server.git",
        "google-suite-mcp-server"
      ],
      "env": {
        "APIFY_TOKEN": "YOUR_APIFY_TOKEN"
      }
    }
  }
}

Step 3: Start Prompting

Open your AI assistant and prompt naturally:

"Use Google Search AI Mode to summarize the latest developments in generative robotics."

Your assistant will query the MCP server, trigger the Actor on Apify, process the trimmed result, and formulate its response.

---

Client Configuration Examples

Option A: Local stdio via uvx directly from GitHub (Recommended ⭐)

No local git clone or manual environment setup required. uvx pulls the latest release from GitHub and executes in high-performance stdio mode.

Claude Desktop & Antigravity IDE

Edit your configuration file:

• Antigravity IDE: ~/.gemini/config/mcp_config.json
• Claude Desktop (Windows): %APPDATA%\Claude\claude_desktop_config.json
• Claude Desktop (macOS): ~/Library/Application Support/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "google-suite": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/ilborso/google-suite-mcp-server.git",
        "google-suite-mcp-server"
      ],
      "env": {
        "APIFY_TOKEN": "YOUR_APIFY_TOKEN"
      }
    }
  }
}

Cursor IDE

Open Cursor Settings -> Features -> MCP Servers or edit your workspace .cursor/mcp.json:

{
  "mcpServers": {
    "google-suite": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/ilborso/google-suite-mcp-server.git",
        "google-suite-mcp-server"
      ],
      "env": {
        "APIFY_TOKEN": "YOUR_APIFY_TOKEN"
      }
    }
  }
}

Alternative: One-Time pip Installation from GitHub

If you prefer standard Python pip without uv:

$pip install git+https://github.com/ilborso/google-suite-mcp-server.git

Then configure your client to run the installed CLI command directly:

{
  "mcpServers": {
    "google-suite": {
      "command": "google-suite-mcp-server",
      "args": ["--transport", "stdio"],
      "env": {
        "APIFY_TOKEN": "YOUR_APIFY_TOKEN"
      }
    }
  }
}

Alternative: Docker Container from GitHub

{
  "mcpServers": {
    "google-suite": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "-e", "APIFY_TOKEN=YOUR_APIFY_TOKEN",
        "-e", "MCP_TRANSPORT=stdio",
        "ghcr.io/ilborso/google-suite-mcp-server:latest"
      ]
    }
  }
}

---

Client Compatibility & Timeouts

Live web scraping and multi-page news crawling involve dynamic browser rendering, proxy allocation, and network queries. While simple searches finish in 5 to 15 seconds, deep crawls or multi-result maps queries may take between 20 and 45 seconds.

Many MCP clients ship with an aggressive default timeout of 10 to 30 seconds. If your client closes the connection before the scrape completes, it will throw an interrupted connection or authentication error.

To ensure uninterrupted operation, configure your client's request timeout to 60 seconds or higher.

Recommended Timeout per Client

Client	Default Timeout	Recommended Timeout	Configuration Method
Claude Desktop	30s	60s+	Set "timeout": 60000 in claude_desktop_config.json
Cursor IDE	30s	60s+	Set "requestTimeoutMs": 60000 in mcp.json
Windsurf	60s	60s+	Default configuration is sufficient
Continue.dev	30s	60s+	Set requestTimeoutMs: 60000 in MCP server block
Cline (VS Code)	30s	60s+	Extension Settings -> MCP Timeout
LangChain MCP	None	60s+	Set timeout=60 in client initialization
@modelcontextprotocol/sdk	30s	60s+	Set requestTimeoutMs: 60000 in client options

---

Best Practices to Avoid Timeouts

1. Sequential Tool Invocations: Execute tool calls sequentially rather than firing multiple concurrent requests. Concurrent calls can trigger client transport closure before all streams finish.
2. Use Conservative Pagination Defaults: Start queries with focused parameter limits (e.g., max_results: 3 to 5 or days_back: 1 for news). Increase result counts only when extensive analysis is needed.

---

Troubleshooting Common Issues

Observed Error	Root Cause	Solution
"Invalid or expired MCP authentication"	The MCP client timed out before the scraping response arrived. This is a client-side timeout issue, not an authentication failure.	Increase your MCP client timeout to 60,000 ms (60 seconds) or more.
"Interrupted network connection"	The client closed the transport socket while waiting for the scraper.	Ensure client timeout is configured to 60s+ and avoid issuing parallel requests.
"Tool call returned no content"	Overlapping requests or dropped SSE stream.	Structure prompts so the assistant awaits each tool result sequentially.
"AUTHENTICATION_REQUIRED"	The APIFY_TOKEN environment variable was not set or is invalid.	Verify that APIFY_TOKEN is correctly configured in your MCP client env block and that your token is active.

---

Payload Optimization & Graceful Degradation

Context Window Protection

To prevent context window bloat and excessive LLM token usage, all scraper responses are automatically trimmed by the server:

• Raw HTML documents, base64 data URIs, and debug payloads are stripped.
• High-value semantic fields are retained: title, url, snippet, price, rating, ai_overview, and sources.
• Long text articles and descriptions are automatically condensed to essential passages.

50-Second Graceful Degradation

All tools enforce a strict internal timeout limit of 50 seconds. If an Apify scraping task exceeds this limit, the MCP server does not crash or raise an unhandled exception. Instead, it returns a structured JSON status payload:

{
  "status": "RUNNING",
  "actor_id": "ilborso/google-maps-fast-scraper",
  "run_id": "W4k2Lm98QpLzx",
  "run_url": "https://console.apify.com/view/runs/W4k2Lm98QpLzx",
  "message": "Execution exceeded the strict low-latency timeout of 50 seconds. The scraper task is still executing safely in the background on Apify. The user can track its live progress or download results once finished at: https://console.apify.com/view/runs/W4k2Lm98QpLzx"
}

This ensures the AI assistant can explain to the user that the task is still progressing safely in the cloud, providing the direct Apify Run URL without breaking the MCP session.

---