Free tool · for LLMs, notes and docs
URL to markdown
Enter a URL and get the main content of the page as markdown: headings, lists, links, tables and code blocks, without navigation, ads and cookie banners. The page is rendered in a real browser first, so text that JavaScript builds is included.
What the markdown contains
- Headings, paragraphs and lists, including nested lists.
- Links and images with absolute URLs, so they still work outside the page.
- Tables as GitHub-flavored markdown tables.
- Code blocks as fenced blocks, with the language when the page marks it (
language-*classes).
Cookie banners and ads are blocked before the conversion, and hidden elements are left out.
Main content or the whole page
With Main content only (the default), you get the article of the page without navigation, sidebars, header and footer, starting with the page’s main heading. When no clear article is found, or the article detection would lose tables or code blocks, the main region of the page is converted instead. Untick it to convert the whole page body.
What people use it for
| Use | Why markdown |
|---|---|
| Context for ChatGPT, Claude and other LLMs | The text and structure of a page without the HTML around it |
| RAG and search indexes | Text with headings that splits into chunks for embedding |
| Notes and knowledge bases | Paste into Obsidian, Notion or a wiki, with links and tables as markdown |
| Documentation migration | Move pages from an old CMS into a docs-as-code repository |
Limits
| Limit | Value |
|---|---|
| Time per page | about 12 seconds to load and convert the page; slower pages are stopped |
| Page size | very large pages are refused: more than about 2 million characters of HTML, or too much memory to convert |
| Pages you can open | public http and https pages; no logins, no private network addresses |
Convert URLs to markdown automatically
The screenshotbase API does the same from your code: one GET request with format=markdown returns the markdown as text, or with response_type=json as JSON with the page’s title and description. AI agents can use it through the MCP server, where the take tool returns markdown as text. See the URL to markdown API.
Frequently asked questions
How do I convert a URL to markdown?
What is left out of the markdown?
Why is part of the page missing?
Is the markdown stored?
Why use markdown for LLMs?
How many pages can I convert for free?
Markdown API
Convert web pages to markdown from your code
GET /v1/take with format=markdown renders the page in a real browser and returns its content as markdown, as text or as JSON with the title and description. For LLM pipelines, search indexes and AI agents.
/v1/take documentation URL to markdown API All documentation
300 free requests every month, and each conversion counts as one. No credit card required.
GET https://api.screenshotbase.com/v1/take?url=https://example.com&format=markdown
curl -G "https://api.screenshotbase.com/v1/take" \
--data-urlencode "url=https://example.com" \
-d format=markdown \
-H "apikey: YOUR-API-KEY"
# Response: the markdown (text/markdown), shortened
# Example Domain
This domain is for use in documentation examples without needing permission. This is not a service; avoid relying on it for testing and monitoring purposes.
…
[Learn more](https://iana.org/help/example-domains)