Website to Markdown

A modern web page is mostly not the article. Around the words you came to read sits a thick layer of interface: a top navigation bar, a cookie consent banner, a newsletter modal, related-post grids, social buttons, comment widgets, ad units, and a sprawling footer. Copy-pasting a page drags all of that along with it.

This tool isolates the main content — the actual article, post, or documentation body — and throws the rest away. It uses established readability extraction to find the primary block of a page, the same class of technique reader modes use, then converts that block to Markdown.

What survives is the part that carries meaning: the title, the headings, the body paragraphs, ordered and unordered lists, block quotes, code blocks, and the inline links inside the text. What is dropped is the chrome around it — menus, ads, share bars, banners, and layout scaffolding. The result is a clean document whose word count is close to what a human would consider the article, with none of the noise a model or a notes app would have to wade through.

Every other converter on this site runs in your browser because it works on a file you already have. This one is different, and the reason is a browser security rule rather than a design choice.

Browsers enforce cross-origin resource sharing (CORS), which stops a page on one domain from fetching arbitrary content from another domain. That rule is what keeps the web safe, but it also means your browser physically cannot pull down an article from a site it does not control. Fetching therefore has to happen on a server.

So here is the exact deal, stated plainly. When you submit a URL, it goes to our server. Our server fetches that public page, extracts the content, converts it to Markdown, and returns it. It does not store the page or build a library of what you convert; it keeps only a short-lived cache of the result so that a link a lot of people convert loads instantly rather than being re-fetched every time. Nothing is uploaded from your device, because there is no file — just the link you pasted. If server-side fetching is a dealbreaker for a given URL, that is a signal the content is sensitive enough that you should not be routing it through any web tool.

The output is designed to be immediately useful in the two places people actually put it.

For LLM context, clean Markdown is close to ideal. Stripping the navigation and ads means you are not spending context-window tokens on menu labels and cookie notices, and the preserved headings give a model — or a retrieval pipeline — clear structure to work with. Paste a doc page or article straight into a chat, or feed several into a RAG index, without hand-cleaning each one first.

For notes and knowledge apps, Markdown is the native tongue. The .md output drops cleanly into Markdown editors and note vaults, so clipping an article becomes a matter of paste-and-save, with headings and links intact rather than a slab of styled HTML.

When doing this once at a time stops scaling — you are ingesting a whole documentation site, or building an agent that reads pages — a metered REST API and a hosted MCP server are on the roadmap. Same conversion, called from your code or an agent instead of a text box, so a research agent can turn a webpage into Markdown mid-task.

The readability approach is tuned for article-shaped pages, and it is worth knowing the edges.

Converts cleanly: blog posts, news articles, documentation pages, long-form essays, and most content built around a single main body of text. These are exactly what readability extraction was designed for, and they come out looking hand-written.

Converts partially or awkwardly: pages that are mostly interface rather than prose — dashboards, web apps, search-results pages, and image-first galleries — because there is no single article body to isolate. Heavily interactive pages that build their content with JavaScript after load may return less than you see in your own browser.

Does not convert: anything you cannot reach without signing in. Content behind a login, a paywall, or a members-only gate is invisible to the server, which fetches the page as an anonymous visitor would. If you cannot read it in a fresh private-browsing window, the tool will not be able to either.