Give web scraping powers to your AI agents
Connect Ultimate Web Scraper to Claude, Claude Code, Cursor or any MCP-compatible AI and scrape whole websites, query the results and run automations — right from the conversation.
Ultimate Web ScraperPulled the whole store — one row per variant:
Low-Top Sneaker, White$120
High-Top Sneaker, Red$135
Turf Cleat, Neon Green$95Ultimate Web ScraperScraped all 186 — here's the table:
Steel Watch, Red Dial18
Gold Day-Date Watch42
Digital Sport Watch, Teal7
CursorUltimate Web ScraperCollected 264 emails:
Ultimate Web Scraper38 products match:
Rattan Arch Bed, Walnut$1,290
Curved Headboard Bed$980
Solid Wood Dining Table$1,540Works with MCP-compatible AI apps
Claude
ChatGPTCursor
Perplexity
Hermes
OpenClaw
Connect your agent in 3 steps
Cursor, Perplexity, Hermes — the same three steps in any app that supports MCP connections.
Install the extension
Connect MCP
Part of the paid cloud plans.
Paste it and ask
Copy it, paste it into your agent.
Claude, Claude Code and ChatGPT skip the token
They connect by signing in instead — nothing to copy but a link.
Claude Code — one line in your terminal
claude mcp add --transport http ultimate-web-scraper https://mcp.ultimatewebscraper.com/mcp
Not sure about your app? Look in its settings for “MCP” or “custom connectors” — if they’re there, it’ll work.
Scraping tools your agent can use
Scrape Shopify stores, product pages, emails, phone numbers, social profiles and map listings — or map entire sites from their sitemaps.
Page Scraper
scrape_pagesExtract one row per URL from similar pages of one site — products, listings, articles, job and real-estate pages. Structured data captured automatically, no selectors; page metadata and phone numbers can join as extra columns.
Shopify Extractor
extract_shopify_storeExtract an entire Shopify store, or just the collections you name, into one row per variant — price, SKU, options, images, availability.
WooCommerce Extractor
extract_catalogExtract an entire WooCommerce store, or the categories you name, into one row per variant — price, SKU, stock, images, variations.
Contact Extractor
extract_contactsBuild lead lists across many different sites — one row per site with emails, social profile links, or both in one run. Deep scan follows internal links to reach contact and team pages.
Map Places Extractor
extract_map_placesPull name, rating, review count, address, phone, website and opening hours from map place pages — one row per place.
Sitemap Explorer
discover_sitemap · select_sitemap_urlsMap every page on a site from its own sitemap, grouped into sections, then pick which ones to feed the Page Extractor. No URL list needed.
Website Analyzer
analyze_websiteCheck any site before you scrape it — detects the platform, finds the sitemap, picks the approach that works.
Automation Control
run_automation · update_automationLet your AI agent run any saved automation — cost estimated from its run history first — reschedule or pause it, and read its latest results.
Data Cleanup
deduplicate_rows · delete_rows · merge_columnsLet your AI agent clean up your data after the scrape — rename or merge columns, drop the ones you don’t need, delete unwanted rows and remove duplicates.
Plus it runs any point-and-click scraper you’ve built in the extension, keeps any job on a daily schedule, and answers “how do I…?” straight from our docs.
Why connect us to your AI agent
On its own, an AI agent reads a handful of pages, one at a time. Connected to us, it scrapes thousands — real browsers, proxies in the location you pick and a clean table back.
Without
I could only reach a few pages — most were blocked or came back empty. You’d need to paste the rest in yourself, a handful at a time.
- Reads a handful of pages before the conversation is full
- One page at a time, in the chat, while you wait
- Pages that build themselves with JavaScript come back empty
- Every request from one address, one page at a time
- Results are text in a chat — gone when you close it
- To analyse it, you paste the data back in and hope it fits
- Want it again tomorrow? Ask again from scratch
With
Ultimate Web ScraperScraped all 186 product pages into one table.
Steel Watch, Red Dial18Gold Day-Date Watch42Digital Sport Watch, Teal7- Thousands of pages in a single job — up to 10,000 sites for contact data
- Runs on parallel cloud browsers — close the tab, it keeps going
- Real browsers that render the page the way you'd see it
- Runs from many locations, retries built in
- Results land in a saved table you can filter, export and re-run
- Ask questions of a 100,000-row table without it entering the chat
- Put it on a daily schedule and the table stays fresh
Shown with Claude — the story is the same in any MCP client.
You stay in charge: a new scrape always shows its price before it runs, and you can connect read-only so it can look but never scrape or change anything.
Built to handle scale
Your AI does the asking — our cloud does the scraping. Every job runs on our infrastructure, not your machine.
Ultimate Web ScraperCloud5 browsers running5 runningParallel browsers
Many sites scraped at once.
Proxies, built in
ISP & residential, any country.
Reliable at scale
JS rendered, retries built in.
Long, unattended runs
Up to 24 hours a job.
Cloud-only
Nothing to host or keep awake.
Common questions
MCP (Model Context Protocol) is the open standard that lets AI apps use other tools on your behalf. Ultimate Web Scraper speaks MCP — connect it and your AI can use our scraping, querying and export tools the same way it uses its own. There's nothing to install and nothing to run: you connect once and it just works.
Cursor, and any other app that supports MCP connections, connect with an access token you create from the Chrome extension's Settings → Connect MCP. Claude on the web and desktop, Claude Code, and ChatGPT are the exception: they connect by signing in, with no token to create.
Three steps. Install the free Chrome extension, open Settings → Connect MCP, and create an access token on the dashboard page it opens. The token arrives inside a ready-made settings block — paste that into your app (in Cursor: Settings → MCP → Add server) and you're connected. Each connection is scoped to one workspace, can be read-only, and can be revoked from the same page at any time.
Very likely. Look in its settings for “MCP”, “custom connectors” or “custom tools” — if your app lets you add your own MCP connection, it'll work with ours. And as more AI apps adopt MCP, they connect the same way, with nothing new to set up.
It's the easiest way, and it's what we recommend: the extension has the Connect Claude, Connect ChatGPT and Connect MCP buttons built in, and it's where you build the point-and-click scrapers your AI can then run. Strictly speaking the connection itself only needs a cloud account on a paid plan, so if you already have one you can also connect from your dashboard's Integrations page. Tables your AI creates live in your cloud workspace, viewable in the dashboard.
You can paste up to 500 links into a single request. To go bigger, point it at a table you already have: emails and social profiles run across up to 10,000 URLs that way, and product pages, phone numbers and map places up to 2,000. Product-page and phone-number jobs need all the URLs to be from the same website. Mapping a site's own sitemap can pull in far more pages than you'd ever paste by hand.
Ask for it in the chat: CSV, JSON or Excel, plus a Shopify-import file for Shopify catalogues. You get a download link that stays valid for 24 hours, and a link to browse the table online. Google Sheets works too once you've connected Sheets in your dashboard — each export creates a new spreadsheet.
The Chrome extension is free to install. Connecting an AI is part of the paid cloud plans. After that, reading, querying, cleaning and exporting your data cost nothing — credits are only used when your AI actually scrapes pages, at around 1 credit per page. Mapping a whole site's pages is a flat 25 credits, and free if you re-scan the same site within the hour.
Before it scrapes pages or pulls a Shopify catalogue, it shows you the estimate and waits for you to confirm — and it counts the rows first before deleting anything. Two things do charge as soon as you ask for them: mapping a whole site (a flat 25 credits) and re-running a scraper you already built (about 1 credit per page). If you'd rather it couldn't spend at all, connect with read-only access.
It can start five ready-made jobs — product details, emails, social profiles, phone numbers and map places — plus full Shopify catalogues and whole-site mapping, and it can run any scraper you've already built. Point-and-click scrapers, where you choose exactly which parts of a page to grab, are built by you in the extension. Image Downloader is extension-only too. LinkedIn pages can't be used as scrape targets.
Give your agents scraping powers.
Install the free extension, connect over MCP once, and every scrape after that is a sentence away — in whichever AI app you work.
Free to install · 100,000+ users · 4.6★ on the Chrome Web Store
Claude is a trademark of Anthropic, PBC. Ultimate Web Scraper is an independent product and is not affiliated with, endorsed by, or sponsored by Anthropic. All other product names and logos are trademarks of their respective owners, used for identification only.