URL to Markdown API

About the URL to Markdown API

This URL to Markdown and website to markdown API converts a webpage to Markdown in one call: send any public URL and get back the page title, the Markdown text and a word count, ready for LLM, RAG and docs pipelines. Headings, paragraphs, lists, tables, code blocks, quotes and emphasis are kept as Markdown, while navigation, headers, footers, sidebars, cookie banners and scripts are removed, so what is left is the main content of the page.

Teams use it to feed articles and documentation into language models, build knowledge bases from help centres, archive pages as plain text, or turn a competitor's changelog into something a script can read. The response also counts the images and links found in the content, which helps you decide whether a page is worth processing further.

On the Free plan links are reduced to their text and images are left out. Paid plans can turn on includeLinks and includeImages to keep full Markdown link syntax and image references with absolute URLs.

How it works

The API fetches the page through a scraping service that tries a plain HTTP request first and falls back to a headless browser, retrying through a proxy when the site answers with a blocking status. The browser waits only for the DOM to load and skips images, fonts and media to stay fast. Non-public addresses are refused before any fetch, and pages that block the request or return an error status are reported as failures instead of being converted.

The HTML is parsed and script, style, iframe, nav, header, footer, aside and common sidebar, menu, advert, cookie-banner and popup classes are removed. Conversion starts from the first main, article or common content container, or from body if none is present. Headings become # to ######, pre blocks become fenced code, tables become pipe tables with the first row as header, and lists, quotes, bold and italic map to their Markdown forms. Runs of blank lines and spaces are collapsed. title comes from the page <title>, then og:title, then the first h1.

includeLinks writes [text](url) and includeImages writes ![alt](src), resolving relative URLs against the page address and reading lazy-load data-src attributes. Both default to false and are ignored on the Free plan. imageCount and linkCount count elements in the content area whatever those settings are, excluding # and javascript: links. wordCount splits the final Markdown on whitespace. Results are cached for one hour per URL and option combination, and a page that yields under 10 characters of Markdown returns an error.

What people use it for

Static Site Migration

Developers moving legacy blogs to static site generators extract page titles and converted text to generate clean Markdown source files automatically.

Offline Documentation Readers

Technical writers turn public product docs into offline Markdown manuals, tracking extracted word counts across pages to structure revision bundles.

Knowledge Base Ingestion

Feed live help center URLs to internal knowledge bases, stripping unnecessary HTML markup while preserving document headings and lists for local search.

Web Page Archiving

Before third-party guides go offline, research teams store target articles as version-controlled Markdown text plus recorded image and link counts.

Ways to call it

One endpoint, many ways in — REST with JSON, XML, YAML and CSV, plus GraphQL and an MCP interface for AI agents.

JSON

Default REST response

XML

Markup format

YAML

Human-readable

CSV

Tabular export

Beta

GraphQL

Query language

New

MCP

For AI agents

Request parameters — POST /v1/urltomarkdown

Sent as JSON in the request body. Premium parameters are accepted on every plan but only take effect on plans that include them. Anything not listed here is dropped rather than passed through.

Required

Optional

1
2
3
4

curl -X POST "https://api.apiverve.com/v1/urltomarkdown" \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url":"https://example.com/","includeImages":true,"includeLinks":true}'

Authentication

Send your key in the x-api-key header. That is the only auth step — no token exchange, and no per-endpoint scope to configure.

Response — POST /v1/urltomarkdown

Every APIVerve endpoint returns the same three top-level keys, so one response handler covers your whole integration: status, error and data. Only data changes shape.

1
2
3
4
5
6
7
8
9
10
11
12

{
  "status": "ok",
  "error": null,
  "data": {
    "url": "https://example.com/",
    "title": "Example Domain",
    "markdown": "# Example Domain\n\n\nThis domain is for use in documentation examples without needing permission. Avoid use in operations.\n\n\n[Learn more](https://iana.org/domains/example)",
    "wordCount": 20,
    "imageCount": 0,
    "linkCount": 1
  }
}

Response fields

Paths are relative to data. Premium fields are absent rather than zeroed on plans that do not include them, so check for presence instead of comparing to 0.

Response headers

Every response carries your balance, so your own code always knows where it stands without polling anything. All four are exposed to browsers through CORS.

Errors

Read the HTTP status first, then error for the specific reason. The body names the parameter that has to change.

What a call costs — 10 credits per call

Credits are shared across every APIVerve API — one balance, one key, one bill. Failed requests never count against it, so a rejected input costs nothing.

Calls are what the credits buy at this API's rate — spend them here, on any of the other APIs, or across both. Nothing is reserved per endpoint.

Zero-code embed — one snippet, no backend

Drop an interactive URL to Markdown form onto any page. The widget calls the API for you, so no key ever appears in your markup and there is nothing to deploy.

Changelog — URL to Markdown API

Every change to this endpoint, newest first. Breaking changes ship as a new version and the previous one keeps serving — response fields are added, never removed or retyped in place, so an integration written against v1 keeps working.

Currently v1Data refreshed realtime