Skip to main content
The official Claude Code plugin gives Claude Code everything it needs to read pages that block it: the MCP tools, a one-shot /scrape-url command, and reference material it pulls in when writing scraping code for your project. You bring your own API key. It is entered once, when the plugin is enabled, and stored in your operating system’s keychain - not in a settings file, not in an environment variable.

​
Install

Run both commands inside Claude Code:
Claude Code asks for your API key while enabling the plugin. Get one from the dashboard if you don’t have it yet.

​
What you get

MCP tools - Claude calls these on its own whenever a page is blocked or needs a real browser: Command - for when you want an explicit one-off fetch:
Skills - reference material loaded on demand while Claude writes code against the API:

​
Example prompts

  • “Read https://example-shop.com/product/123 and tell me the price.” - Claude reaches for fetch_html by itself once the page blocks a plain request.
  • “Get the parsed data for https://www.amazon.com/dp/B08N5WRWNW and summarise the product.”
  • “Write me a Python script that scrapes these 40 listing URLs.” - Claude pulls in the skills and writes it against the SDK, with retries and country targeting handled properly.

​
Managing the plugin

​
Plugin or MCP server?

Both run the same server underneath, so the tools are identical. Use the plugin in Claude Code; use the MCP server everywhere else.

​
Source