URL to Markdown API
URL to Markdown, in one API call.
POST a URL and get back LLM-ready Markdown with the navigation, cookie banners, and scripts stripped. Add a schema and the same call returns validated JSON.
1,000 free credits every month · No card required · Blocked requests cost nothing
$ curl -X POST https://api.datamint.xyz/v1/mint \
-H "Authorization: Bearer mint_live_xxx" \
-H "Content-Type: application/json" \
-d '{
"url": "https://example.com/article"
}'# Example article
The page's main content, with navigation, cookie banners, and scripts removed.
## Section heading
- Lists stay lists
- Links keep their targets
- Heading order is preservedThe output
The page without the furniture.
A browser needs the navigation, cookie banner, scripts, and footer. Your model does not. The API returns the main content as Markdown and keeps the structure that matters intact.
One POST replaces the fetch, the parser, and the cleanup step. Pass a schema and the same endpoint returns typed JSON instead.
Boilerplate stripped
datamint removes navigation, cookie banners, scripts, and footers before it returns the response.
Structure preserved
Headings, lists, and links keep their order and targets, so the Markdown drops straight into a context window.
Typed when you need it
Add a schema and the same endpoint returns JSON in that shape, with a confidence score alongside it.
The call
One endpoint for any language.
It is a plain HTTP POST. These three examples do the same thing: send one URL and read the Markdown back.
curl
curl -X POST https://api.datamint.xyz/v1/mint \
-H "Authorization: Bearer mint_live_xxx" \
-H "Content-Type: application/json" \
-d '{"url": "https://example.com/article"}'Python
import requests
response = requests.post(
"https://api.datamint.xyz/v1/mint",
headers={"Authorization": "Bearer mint_live_xxx"},
json={"url": "https://example.com/article"},
)
response.raise_for_status()
print(response.text) # clean MarkdownTypeScript
const response = await fetch(
"https://api.datamint.xyz/v1/mint",
{
method: "POST",
headers: {
Authorization: "Bearer mint_live_xxx",
"Content-Type": "application/json",
},
body: JSON.stringify({
url: "https://example.com/article",
}),
},
);
const markdown = await response.text();Set js: true for client-rendered pages (5 credits), stealth: true for bot-defended targets (15 credits), or schema for JSON (+2 credits). The full parameter reference is in the docs.
Choosing an approach
When a library is enough, and when you need an API.
There are three ways to turn a URL into Markdown. The right one depends on who does the fetching.
Local library
Readability + Turndown
Best when you already have the HTML for a static page. Free, and you keep full control.
You still own the fetch, JavaScript rendering, proxy rotation, and parser upkeep.
Hosted reader endpoint
r.jina.ai and similar
A quick way to turn a public page into Markdown.
Published rate limits, and it is not built to get through bot defences. The sourced limits are in the Jina Reader guide.
Read the Jina Reader guidedatamint
One API for the whole path
Fetch, render, stealth routing, and typed output in one call. datamint bills each request at the tier that succeeded.
Paid per page: 1 to 17 credits, with 1,000 free every month.
See pricingIf the target is a public, static page and you already have the HTML, a library is the cheapest option. The API handles the part after that: JavaScript, bot defences, and a fixed output shape.
Pricing
Priced per page in credits, not seats.
The transport tier decides the cost. datamint bills you for the tier that succeeded, and blocked requests cost nothing.
1,000 free credits every month · No card required
Questions
Before you send your first request.
What does the API return?
Without a schema, the response body is clean Markdown with navigation, scripts, and boilerplate removed. With a schema, it returns JSON in that shape, plus a confidence score.
Does it render JavaScript?
Yes. Set js to true to render the page in a headless browser before extraction. That tier costs 5 credits instead of 1.
Can it reach bot-defended sites?
Set stealth to true to route the request through residential proxies with fingerprint mutation. That tier costs 15 credits. It is built for publicly accessible data only: no logins, paywalls, or access-control circumvention.
How much does a page cost?
1 credit for a static fetch, 5 for a headless render, 15 for stealth routing, and 2 more for a schema pass: 1 to 17 credits per page. Credits are $1 per 1,000, and every account gets 1,000 free each month.
Do failed requests cost credits?
No. datamint charges credits only when a request succeeds. If a target blocks the request and nothing usable comes back, you pay nothing.
Do I need an SDK?
No. It is a plain HTTP endpoint, so curl, Python, TypeScript, or any HTTP client works. There is also a hosted MCP server for Claude, Cursor, and other agents.
Compare web scraping API alternatives or read the sourced Jina Reader guide.
Stop maintaining scrapers. Start shipping.
Point datamint at the targets your team keeps losing. Get typed data back, and stop rebuilding parsers every sprint.
1,000 credits free every month · No card required · Blocked requests cost nothing