ScrapeUnblocker plugin for Claude Code and Cursor

Official Claude Code and Cursor plugin for ScrapeUnblocker - fetch any web page from behind anti-bot protection (Cloudflare, DataDome, PerimeterX, Akamai, Shape) without writing a scraper.

Install

In Claude Code:

/plugin marketplace add ScrapeUnblocker/claude-code-plugin
/plugin install scrapeunblocker@scrapeunblocker

You will be asked for your ScrapeUnblocker API key when the plugin is enabled. It is stored in your OS keychain, not in a settings file. Get a key at app.scrapeunblocker.com.

Install in Cursor

The same plugin ships a Cursor manifest (.cursor-plugin/). Its MCP server is the hosted one (https://mcp.scrapeunblocker.com/mcp/oauth): you sign in with your ScrapeUnblocker account once, and there is no key to paste.

Until it is listed in the Cursor Marketplace, install it locally:

git clone https://github.com/ScrapeUnblocker/claude-code-plugin
mkdir -p ~/.cursor/plugins/local
cp -r claude-code-plugin/plugins/scrapeunblocker ~/.cursor/plugins/local/scrapeunblocker

Then run Developer: Reload Window, open Customize, find the scrapeunblocker MCP server and click Connect to sign in. The skills load with it; the /scrape-url command is Claude Code only.

What you get

MCP tools - available to Claude automatically:

Tool	What it does
fetch_html	Fully rendered HTML of any page, anti-bot bypassed
fetch_parsed	AI-parsed structured JSON (products, articles, listings) instead of raw markup
google_search	Organic Google results as structured data
fetch_markdown	A page as clean Markdown, usually 2-25% the size of the HTML - the cheap way to read a page
search_and_read	A Google search plus its top 1-5 result pages as Markdown, in one call
list_elements	A page's links, inputs and buttons with ready-to-use selectors
browse_page	Click, type, select and scroll on a page in a real browser, then return the HTML (not read-only: it acts on the website)
list_plugins / run_plugin	100+ ready-made site scrapers (Amazon, Zillow, Google Maps, YouTube, App Store...) that return clean JSON
amazon_search, amazon_product, amazon_reviews, walmart_search, ebay_search, google_shopping, google_local, maps_place_details, google_images, youtube_search, youtube_transcript, zillow_search, indeed_jobs, trustpilot_reviews, booking_hotels	The most-used site plugins as tools of their own

Command:

/scrapeunblocker:scrape-url https://example.com
/scrapeunblocker:scrape-url https://example.com parsed
/scrapeunblocker:scrape-url https://example.com markdown
/scrapeunblocker:scrape-url https://example.com country=US

Skills - instructions the agent loads when a task matches, for using ScrapeUnblocker well and for writing scraping code into your project:

Skill	What it teaches the agent
best-practices	Endpoints, parameters, retries, choosing the right call
page-source	Rendered HTML or Markdown (markdown=true), render waits, country targeting
parsed-data	AI-parsed structured JSON and how to fix a bad parse
serp	Google results as JSON: organic with real URLs, ads, AI overview, People also ask, local, shopping
structured-data	Ready-made site plugins (Skyscanner, eBay, Google Local) instead of hand-written parsers
mcp	Setting up the ScrapeUnblocker MCP server (local and hosted) in any MCP client
competitive-intel	Competitor pricing, features, reviews and hiring signals from their own sites
price-comparison	Live prices and stock across retailers, ranked into one recommendation

Agent Skills for any agent

The skills are plain Agent Skills (SKILL.md folders), so they also work without the plugin - in Codex, Gemini CLI, VS Code Copilot, Windsurf, Cursor or Claude Code. Install all eight with the skills CLI:

npx skills add ScrapeUnblocker/claude-code-plugin

Or pick one:

npx skills add ScrapeUnblocker/claude-code-plugin --skill page-source

You can also copy a folder from plugins/scrapeunblocker/skills/ into your agent's skills folder (.claude/skills/, .agents/skills/, .cursor/skills/...).

Without the plugin, the skills call the HTTP API with your key: put it in the SCRAPEUNBLOCKER_KEY environment variable (the SDKs read it from there). Get a key at app.scrapeunblocker.com. For tool calls instead of HTTP, add the MCP server too.

Why

Anything a normal fetch or curl returns as 403, a captcha, or an empty JS shell comes back as the real page. Requests go through a real browser and a proxy pool, so Claude can read pages it otherwise cannot: pricing pages, marketplaces, listings, SERPs.

Links

• Docs: docs.scrapeunblocker.com
• MCP server (standalone, for other MCP clients): scrapeunblocker-mcp
• SDKs: Python, Node.js, Ruby, PHP

License

MIT