Content extraction API
Distil fetches a page, renders the JavaScript, strips nav, ads and
chrome, and hands back the main content as tidy Markdown your model
can actually read. One POST — structured, deterministic,
ready for retrieval.
$ curl https://api.distil.dev/v1/extract \
-H "Authorization: Bearer $DISTIL_KEY" \
-H "Content-Type: application/json" \
-X POST -d '{
"url": "https://example.com/blog/rag-tips",
"format": "markdown",
"main_only": true
}'{
"title": "Seven RAG tips",
"word_count": 2184,
"markdown": "# Seven RAG tips..."
}$ npm i @distil/sdk
Built for pipelines
You don’t want raw HTML in your context window. Distil returns the part that matters — and tells you exactly what it kept.
Typed SDK
Point the client at a URL and stream structured Markdown straight into your splitter. Retries, render-wait and rate limits are handled for you.
import { Distil } from "@distil/sdk";
const distil = new Distil(process.env.DISTIL_KEY);
const doc = await distil.extract({
url: "https://example.com/docs",
format: "markdown",
render: "auto", // wait for JS only if needed
});
index.add(doc.markdown); // straight into your vector storeA readability model isolates the article body and drops nav, cookie banners, footers and related-post rails — so your embeddings learn signal, not chrome.
Same URL, same Markdown — every call returns a content hash, token count and the rules that fired, so a re-index is a diff, not a guessing game.
content_hash on every responseQuickstart
Install the SDK, point extract at a URL, and read clean
Markdown back. No headless browser to babysit, no HTML to scrub —
the happy path is the whole path.
import { extract } from "@distil/sdk";
const doc = await extract("https://example.com/blog/rag-tips");
console.log(doc.markdown); // → "# Seven RAG tips\n\n## 1. Chunk on…"Pricing
Metered per successful extraction. Failed fetches are never billed.
Free
$0
1,000 pages / month
Scale
$0.0008
per page · volume tiers
Enterprise
Custom
dedicated & on-prem