Free Website to Markdown Converter
Turn a web page URL or raw HTML into clean Markdown that keeps headings, links, lists, tables and code blocks. Ready for an LLM prompt, a docs site or a notes app.
By the Scrapier team · Free · No signup · URL or HTML · MD download · Last updated
Sample result
The Wikipedia article "Web scraping" on 24 Oct 2026 returned 4,254 words of Markdown in 104 ms: one # title, 6 ## and 12 ### headings, and 291 links as absolute URLs.
| Input URL | en.wikipedia.org/wiki/Web_scraping |
| Mode | URL, main content |
| Fetch time | 104 ms |
| Words | 4,254 |
| Headings | 1 #, 6 ##, 12 ### |
| Links | 291, absolute URLs |
| Removed | Navigation, header, footer, sidebars, scripts, styles, forms |
# Web scraping - Wikipedia Source: https://en.wikipedia.org/wiki/Web_scraping … **Web scraping**, **web harvesting**, or **web data extraction** is [data scraping](https://en.wikipedia.org/wiki/Data_scraping) used for [extracting data](https://en.wikipedia.org/wiki/Data_extraction) from [websites](https://en.wikipedia.org/wiki/Website). … ## History … After the [birth of the World Wide Web](https://en.wikipedia.org/wiki/History_of_the_World_Wide_Web) in 1989, the first web robot, … ## Techniques ### Human copy-and-paste
Excerpt. Wikipedia's [edit] links and citation markers stay in the full output; the title and Source line are added by the tool. Run on .
How to use the website to markdown converter
- Paste the page URL, or switch to Paste HTML to convert HTML directly.
- Press Convert to Markdown. Navigation, scripts and footers are removed; the content keeps its structure.
- Copy the Markdown or download a .md file.
What you get
Clean Markdown of the page's main content, with a # title and a Source line at the top.
| Field | What it contains | Example |
|---|---|---|
| Headings | H1 to H6 become # to ###### | ## History |
| Links and images | Absolute URLs, so they work outside the page | [data scraping](https://en.wikipedia.org/wiki/Data_scraping) |
| Lists and tables | Bullets, numbered lists and pipe tables | - item, 1. item, | a | b | |
| Code | Fenced code blocks, with the language when the page marks it | ```python … ``` |
Main content
Converts <main> or <article> and drops navigation, headers, footers, sidebars and forms. Pick it for LLM context, docs and notes.
Full page
Converts all visible content, menus and footers included. Pick it to archive a page as it stands.
Who it is for
Developers and AI builders
Markdown keeps the structure an LLM needs while using far fewer tokens than HTML. Use it for prompts, agents and RAG chunks.
Technical writers and docs teams
Move existing web pages into a Markdown docs site, a wiki or a static site generator without rewriting them.
Researchers and note-takers
Save articles into Obsidian, Notion or a Git repo with links and headings intact.
Common uses
- Give an AI agent or chatbot a web page as context: Markdown keeps headings and lists the model can follow, at a fraction of the HTML token count.
- Migrate pages from an old CMS into a Markdown-based docs site such as Docusaurus, MkDocs, Hugo or a GitHub wiki.
- Clip articles into Obsidian, Notion or Logseq with working links.
- Convert HTML email or newsletter archives into Markdown for a repository.
- Turn documentation pages into Markdown files an agent can read with a file tool.
How it compares
This tool: Keeps headings, links, lists and tables; drops page chrome. Watch out for: One page at a time. Other methods, and the case where each one fits:
HTML to Markdown libraries
Good at: Scriptable (turndown, markdownify). Watch out for: Convert the whole page, menus included, unless you pre-clean it. The same job in a few lines of Python, for when you need to run it on many pages.
import requests
from bs4 import BeautifulSoup
from markdownify import markdownify # pip install markdownify
html = requests.get("https://example.com", timeout=15).text
soup = BeautifulSoup(html, "html.parser")
for tag in soup(["script", "style", "nav", "header", "footer", "aside", "form"]):
tag.decompose()
main = soup.find("main") or soup.find("article") or soup.body
print(markdownify(str(main), heading_style="ATX"))Copy and paste into an editor
Good at: Fast for a paragraph. Watch out for: Loses tables and links; brings in junk.
Writing a scraper
Good at: Batch whole sites. Watch out for: You maintain fetching, cleaning and conversion.
Scrapier API for platform data
Typed JSON from 35 platforms at volume, or inside an AI agent: see Use the API instead below.
| Method | Pages | JavaScript pages | Keeps links and tables | Effort |
|---|---|---|---|---|
| This tool | 1 per run | No | Yes | None |
| HTML to Markdown libraries | Many | No | Yes | Pre-clean the HTML |
| Copy and paste into an editor | 1 | Yes | No | Manual cleanup |
| Writing a scraper | Many | Your code | Your code | Write and maintain |
Limits and tips
- Links and images come out as absolute URLs, so the Markdown still works after you move it.
- For a page that needs a login or renders with JavaScript, paste the HTML from your browser's Inspect tool.
- Want just the words? The Website to Text Converter returns plain text without Markdown syntax.
- Reads the HTML a server sends. Pages that build their content with JavaScript in the browser can come back short or empty; paste the rendered HTML instead (in your browser: right-click, Inspect, copy the
<html>element). - Public pages only: no logins, paywalls or private content.
- Respects robots.txt. If a site asks automated tools not to fetch a page, the tool stops and says so.
- One page per run, up to 3 MB of HTML. 30 runs per hour per visitor.
Use the API instead
The free tool reads one page at a time. Scrapier's API returns typed JSON from Instagram, TikTok, Amazon, Google Maps and 31 more platforms: 199 endpoints, billed per successful call. It does not fetch arbitrary pages; it gives you the platform data directly, over REST or as an MCP tool inside an AI agent.
curl -G "https://proxy.scrapier.io/api/v2/bing-web-search/" \ -H "X-API-Key: YOUR_API_KEY" \ --data-urlencode "query=web scraping api" # → typed JSON, 1 credit per successful call; failed calls are free
import requests r = requests.get("https://proxy.scrapier.io/api/v2/bing-web-search/", headers={"X-API-Key": "YOUR_API_KEY"}, params={"query": "web scraping api"}) print(r.json())
"Search Bing for ‘web scraping api’ and give me the top 10 results as a Markdown list."
→ 1 successful call, 1 credit each: 1 credit. Failed calls are free.
| Free tool | Scrapier API | |
|---|---|---|
| Reads | Any public web page, or HTML you paste | 199 endpoints on 35 platforms |
| Output | Markdown of the page | Typed JSON per endpoint |
| Pages | 1 per run | Many, up to 3 parallel requests per key |
| Rate limit | 30 runs/hour per visitor, 3 MB per page | Your plan's credits; 429 with Retry-After if you go too fast |
| Signup | None | API key; a free tier is available |
| Price | Free, no signup | Per successful call, failed calls free. From $29/mo for 15,000 credits |
Safe and legal use
Fetching a public page is the same request a browser makes. The tool respects robots.txt, refuses private and local network addresses, and does not store the pages it reads or the HTML you paste. The site's terms and copyright govern how you reuse what you extract. Not legal advice; see our Terms of Use and Privacy Policy.
Frequently asked questions
Is the website to markdown converter free?
Yes. No signup and no charge, up to 30 conversions an hour per visitor.
Can I convert HTML to Markdown directly?
Yes. Switch to Paste HTML, paste any HTML, and convert. Links in pasted HTML stay as they are written.
Why use Markdown for LLMs?
Markdown keeps the structure (headings, lists, tables) that helps a model understand a page, and it is much shorter than HTML, so it costs fewer tokens.
Does it work on JavaScript-heavy pages?
It converts the HTML the server sends and does not run JavaScript. For pages built in the browser, paste the rendered HTML instead.
What gets removed?
Scripts, styles, navigation, headers, footers, sidebars, forms and hidden elements. The main content (the page's <main> or <article> when there is one) is kept.
Can I convert a whole site?
The tool converts one page per run. For many pages, use the Python example on this page or the URL Extractor to collect the links first.
Structured data from 35 platforms, one key.
199 endpoints that return typed JSON. Billed per successful call; failed calls are free.