Connect the Arcadium Data scrapers to Claude and Cursor through one MCP endpoint

An AI assistant answers questions about reviews, prices or job openings from memory unless it has a tool that reads the live page. Adding one scraper at a time to an MCP client means one config entry and one input schema per scraper.

arcadium/scrapers-mcp is a remote MCP server with one tool per question an agent asks: reviews of a hotel or place, software reviews, hotel prices, Google Trends, Google News, YouTube transcripts and videos, company hiring, job search, Amazon, Etsy, 1688.com, TikTok Shop, Zillow, Trustpilot and used cars in 8 EU countries. Each tool has a small input schema, an example and its price in the description. Answers are compact JSON with no reviewer, recruiter or seller names.

What you need

• An Apify account and its API token: Apify Console, Settings, API & Integrations. The Free plan works. The token is required on every request.
• Claude Code, Claude Desktop (with Node.js for the mcp-remote bridge), Cursor or any MCP client that speaks Streamable HTTP.

Step 1: Claude Code

One command. $APIFY_TOKEN is read from your shell. The endpoint is https://arcadium--scrapers-mcp.apify.actor/mcp (Streamable HTTP, on Apify).

claude mcp add --transport http arcadium-scrapers https://arcadium--scrapers-mcp.apify.actor/mcp \
  --header "Authorization: Bearer $APIFY_TOKEN"

Step 2: Claude Desktop

Put this in claude_desktop_config.json (Settings, Developer, Edit config) and restart Claude Desktop. The mcp-remote bridge adds the bearer header.

{
  "mcpServers": {
    "arcadium-scrapers": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://arcadium--scrapers-mcp.apify.actor/mcp",
        "--header",
        "Authorization: Bearer ${APIFY_TOKEN}"
      ],
      "env": {
        "APIFY_TOKEN": "<your Apify API token>"
      }
    }
  }
}

Step 3: Cursor

Put this in ~/.cursor/mcp.json, or in .cursor/mcp.json inside a project.

{
  "mcpServers": {
    "arcadium-scrapers": {
      "url": "https://arcadium--scrapers-mcp.apify.actor/mcp",
      "headers": {
        "Authorization": "Bearer <APIFY_TOKEN>"
      }
    }
  }
}

Step 4: LangChain or LangGraph (Python)

With langchain-mcp-adapters, pass the tools to create_react_agent(model, tools). Any other MCP client: Streamable HTTP, the URL above and the bearer header.

from langchain_mcp_adapters.client import MultiServerMCPClient

client = MultiServerMCPClient({
    "arcadium-scrapers": {
        "transport": "streamable_http",
        "url": "https://arcadium--scrapers-mcp.apify.actor/mcp",
        "headers": {"Authorization": f"Bearer {APIFY_TOKEN}"},
    }
})
tools = await client.get_tools()  # pass to create_react_agent(model, tools)

Ask in plain words

Once connected, ask for example: "Use the arcadium-scrapers tools to show me the Google Trends interest in claude over the past 12 months, and the latest news about it." The assistant picks the tool, fills its input and answers from the rows. Every answer carries status, the compact results, free errors with a code, the run’s chargedUsd and a datasetUrl with all rows of the run on your account. A call that answers "status": "running" is finished later with the free get_run_results tool.

What it costs

The server itself is free. Each tool call runs one Arcadium Data scraper on your own Apify account and is billed there per result, from $0.30 per 1,000 results, at the price shown in the tool’s description. Error rows (not found, blocked, invalid input) are always free.

Tools

One tool per question; the price is the Apify Free plan price per 1,000 results of the scraper the tool runs.

Limitations

• Every call returns at most 50 rows and has a spend cap of $0.10-0.50, set per tool; all rows stay in the run’s dataset on your account.
• At most 3 runs in flight and 120 calls an hour per token.
• The server stores no tool inputs, tokens or IP addresses; it logs one row per call (tool, channel, status, USD) for usage statistics.

Links

• Arcadium Data Scrapers MCP Server
• arcadium/scrapers-mcp on the Apify Store
• MCP setup for single actors
• All tutorials