Scrape

Turn any URL into clean, LLM-ready markdown

Paste a URL and watch it come back as markdown: rendered in a real browser, stripped of navigation and boilerplate, ready for a prompt or a knowledge base.

Scrape Crawl Map Search Parse Monitor Crawl tree

1 call
From a URL to markdown, HTML, links, screenshots and page metadata
Real browser
JavaScript apps are rendered before they are read, so you get the page a person sees
~500
Tokens per page of text on average, against 1 billion free

What a scrape gives you

One request, one page, every format you asked for. The result also lands in a knowledge base, so it is chunked, embedded and searchable without a second step.

Markdown that keeps structure

Headings, lists, tables, links and code blocks survive the conversion. Navigation, footers, sidebars and scripts do not.

Real browser rendering

Single-page apps and client-rendered content are rendered before they are read, so you get the page a person sees.

Several formats at once

Ask for markdown, HTML, links and a screenshot in the same call, with title, description and status as JSON metadata.

Linked documents included

With include_docs, PDFs and office files the page links to are parsed too, with OCR for scans.

Straight into a knowledge base

Every scrape is stored, chunked along headings and embedded, ready for keyword or semantic search.

Public pages only

Private and internal addresses are refused before anything is fetched. You choose what to scrape; we never train on it.

How scraping works

01 / 03
Send a URL
POST /api/jobs/scrape with the URL, the formats you want and the knowledge base to keep it in.
02 / 03
We render and clean it
The page is rendered, the main content is extracted and converted to markdown, and linked documents are queued if you asked.
03 / 03
Read or search the result
Fetch the page back as markdown or JSON, or search it alongside everything else in the knowledge base.

Scrape from your code

Jobs are asynchronous: a scrape returns a job id at once, and the page is ready when the job completes, usually within seconds.

curl -X POST https://crawl.sarv.com/api/jobs/scrape \
  -H "x-api-key: YOUR_API_KEY" -H "Content-Type: application/json" \
  -d '{"kb_id":"kb_abc123","url":"https://example.com","formats":["markdown","html"],"include_docs":true}'

Scrape API reference

Questions

What is a scrape API?

An HTTP endpoint that fetches one web page for you and returns its content in a usable format. SarvCrawl returns markdown, HTML, links, screenshots and metadata, and keeps the result in a searchable knowledge base.

Does it work on JavaScript-heavy sites?

Yes. Pages are rendered in a real browser before the content is read, so single-page apps come back as markdown rather than an empty shell.

Can I scrape a page without signing up?

Yes, the demo above scrapes any public page and shows a preview. A free account shows the full result, adds downloads and the API, and comes with 1 billion tokens.

How is it priced?

By tokens consumed by parsing, chunking and embedding, roughly 500 per page of text. Sign-up includes 1 billion tokens with no card.

How is scrape different from crawl?

Scrape reads one URL. Crawl starts at a URL and follows links across the site, scraping every page it finds within your depth and page limits.

Scrape your first page in a minute

1 billion tokens free on sign-up, no card. Roughly 500 tokens a page.

Start with 1 billion free tokens Read the API docs