Available tools | PyScrappy

The server exposes 22 tools. Every tool returns a ScrapeToolResult with the same shape: data (the items), count, scraper, source_urls, and errors.

Tool	Description
`scrape_url`	Scrape any URL: text, links, images, tables, metadata
`scrape_wikipedia`	Fetch a Wikipedia article (`full` / `paragraphs` / `headers`)
`scrape_stock`	Yahoo Finance quotes, history, and profiles
`scrape_news`	RSS/Atom feeds, auto-discovered site feeds, or a single article
`search_images`	Image search (returns URLs + metadata)
`search_youtube`	YouTube video search
`search_linkedin_jobs`	Public LinkedIn job listings
`search_github`	GitHub repository search (stars, language, …)
`search_hackernews`	Hacker News story search (points, comments)
`search_books`	Book search via Open Library (title, author, year)
`get_weather`	Current weather for a place (no key)
`get_crypto`	Cryptocurrency prices and market data (CoinGecko)
`convert_currency`	Exchange rates and currency conversion
`define_word`	Word definitions and examples
`search_amazon`	Amazon product search
`search_newegg`	Newegg electronics / computer hardware search
`search_ikea`	IKEA furniture / home search
`search_soundcloud`	SoundCloud track search (uses the browser backend)
`lookup_movie`	Movie/TV info from IMDB by title or id (via OMDb; needs `OMDB_API_KEY`)
`scrape_zomato`	Restaurant listings by city
`search_ubereats`	Uber Eats restaurants by city
`get_ubereats_menu`	An Uber Eats restaurant's full menu (from its store URL)

The result envelope

Every tool returns the same typed structure, which gives agents a stable schema:

{
  "data": [ { "...": "item fields depend on the source" } ],
  "count": 5,
  "scraper": "crypto",
  "source_urls": ["https://api.coingecko.com/..."],
  "errors": []
}

data: the scraped items. The item shape depends on the source (a movie, a stock quote, an article), so it's a list of free-form objects.
errors: non-fatal problems, each with a url and a message.

JavaScript-rendered pages

scrape_url does a fast static fetch by default: it reads the HTML the server sends and does not run JavaScript. Many modern pages (GitHub's contribution graph, dashboards, infinite-scroll feeds) ship an almost-empty HTML shell and build their content with JS in the browser, so a static scrape comes back with little or no data.

When that happens, scrape_url adds a hint to errors telling you the page looks JS-rendered. You have two ways to handle it:

Render the JavaScript. Pass render_js=true so PyScrappy loads the page in a headless browser first, then scrapes the rendered result:
```
{ "url": "https://github.com/torvalds", "render_js": true }
```
This needs the browser extra on the server (pip install 'pyscrappy[browser]' plus playwright install chromium). If it isn't installed, render_js returns an error.
Scrape the data endpoint directly. JS pages fetch their data from an API or a dedicated URL that is static. For example, GitHub's contribution graph lives at https://github.com/users/<name>/contributions, which scrape_url reads with a plain fetch, no browser needed. This is usually faster and more reliable than rendering.

24 scrapers, 22 tools?

Two scrapers aren't exposed as MCP tools because they're blocked without a proxy (Instagram, Twitter/X, Spotify), and the Uber Eats scraper provides two tools (search_ubereats + get_ubereats_menu). The net is 22. See the scraper catalog for the full mapping.