Use EuroScrape data from Claude, ChatGPT, Cursor and other AI agents (MCP)

Every EuroScrape Actor on Apify can be called by an AI agent through the Model Context Protocol (MCP). You do not need to write a scraper or an API client: you tell your assistant “give me the companies registered in Texas this week” or “what does a 3-night stay in Rome cost on every booking site”, and it runs the right Actor, waits for the result and reads the dataset. Apify hosts the MCP server; EuroScrape provides the Actors; you pay per result, and the first results of every run are free.

This page gives the exact configuration for the main clients as of 8 November 2026, one worked example, and the limits to know before you connect a paying agent.

What you get

• One server, 30+ data sources: company registers (France, Spain, UK, US states, Switzerland), building permits, energy ratings, EU public tenders, hotel and flight prices, marketplace listings (Vinted, Kleinanzeigen, Marktplaats, Subito, Wallapop, willhaben), app reviews, VAT numbers, fuel and electricity prices.
• Structured JSON for each row, the same fields as the Actor’s dataset, so the model can filter, count and compare instead of reading a web page.
• Tools the agent can call: search the Store, read an Actor’s input schema, run it, fetch the dataset. With the actors parameter you pin the Actors you trust and the agent sees only those.
• The same price as the Store: for example US New Business Filings is $3 per 1,000 companies, Google Hotels Prices from $0.002 a hotel. No subscription on top.

Configuration

Apify’s remote MCP server is https://mcp.apify.com. It authenticates with OAuth in clients that support it, or with your Apify API token in an Authorization: Bearer header. The actors query parameter preloads specific Actors as tools; without it, the agent gets a search tool and can discover Actors itself. Check Apify’s MCP documentation for the current client list; the shapes below are the standard MCP configuration formats.

Claude Desktop (claude_desktop_config.json, via the mcp-remote bridge):

{
  "mcpServers": {
    "euroscrape-data": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://mcp.apify.com?actors=euroscrape/us-new-companies,euroscrape/france-companies,euroscrape/google-hotels-prices,euroscrape/eu-public-tenders",
        "--header",
        "Authorization: Bearer YOUR_APIFY_TOKEN"
      ]
    }
  }
}

Cursor (.cursor/mcp.json) and Windsurf use the same idea with a direct URL:

{
  "mcpServers": {
    "euroscrape-data": {
      "url": "https://mcp.apify.com?actors=euroscrape/us-new-companies,euroscrape/spain-companies,euroscrape/vinted-scraper",
      "headers": { "Authorization": "Bearer YOUR_APIFY_TOKEN" }
    }
  }
}

ChatGPT: connectors that accept a remote MCP server URL (developer mode, paid plans) take https://mcp.apify.com and the OAuth flow. Claude.ai web: Settings → Connectors → add a custom connector with the same URL. Your own agent (Python or TypeScript MCP SDK, LangChain, Vercel AI SDK): the streamable HTTP transport with the URL above.

Replace YOUR_APIFY_TOKEN by a token from the Apify Console (Settings → Integrations). Never commit it. A free Apify account is enough to start: every EuroScrape run returns its first results at no charge, so the first questions your agent asks cost nothing.

A real example

Ask Claude, with the server connected: “Which companies were registered in Texas in the last 7 days with ‘solar’ in their name?”

1. The agent picks euroscrape/us-new-companies, reads its input schema and calls it with {"states": ["TX"], "daysBack": 7, "keywords": ["solar"]}.
2. The run finishes in about a minute; the agent fetches the dataset: one row per company with name, entity type, registration date, state, city, activity and the official register link.
3. It answers with the list and can export it as CSV, or schedule the same question every Monday.

For scale: over 31 October to 6 November 2026, the same Actor counted 4,598 new companies in New York State (72% LLCs, Brooklyn 700, Albany 588), straight from the state’s open data. The US New Business Filings example page shows the exact input and a sample of the output; the French companies page does the same for SIRENE. Both are the data your agent receives, nothing is summarised on our side.

How to do it in three steps

1. Create an Apify token (free account, Console → Settings → Integrations) and paste it into one of the configurations above.
2. Pin your Actors with the actors parameter: start with one or two, so the model cannot wander. Our Store page lists them all with their inputs.
3. Ask a question that names the data, not the tool: “new LLCs in Colorado this week”, “cheapest Rome hotels for 12–15 December”, “EU tenders for IT services published yesterday”. Check the first run in the Apify Console (Runs tab) to see the exact cost, then automate.

Pricing

Each Actor keeps its Store price; the MCP server adds nothing. Current examples: US New Business Filings $3 per 1,000 companies plus $0.003 per run; Spain New Companies $4 per 1,000; Google Hotels Prices from $0.002 per hotel; EU Public Tenders $3 per 1,000 notices. First results of every run free. Apify’s own platform usage (compute) is billed by Apify on the free tier’s $5 monthly credit first.

FAQ

Does the agent need a browser or a scraper of its own? No. The Actor runs in Apify’s cloud and returns JSON; the agent only calls tools.

Can I limit what the agent can run and spend? Yes: pin Actors with the actors parameter, set a maximum cost per run in each Actor’s input (maxItems, and Apify’s per-run spending limit), and use a dedicated Apify token you can revoke.

Which client works best? Any client that supports remote MCP servers over streamable HTTP. Claude Desktop needs the mcp-remote bridge as shown; Cursor and Windsurf take the URL directly.

Is the data fresh? Each run reads the source at the moment you ask: state registers are read live, price Actors query the sites live. Data pages on this site are rebuilt daily.

What about personal data? The company Actors return company data only (no officer names where the source lists them, no e-mails unless the official register publishes them). You remain responsible for your use under the GDPR.

Open the EuroScrape Store page on Apify → · More guides: Google Hotels API, businesses without a website, all guides.