Any to markdown: how I convert web pages into clean Markdown for AI

Any to markdown: how I convert web pages into clean Markdown for AI

"Any to markdown" sounds simple until you try to use it on real pages.

A clean article page is easy. A logged-in dashboard, a documentation page with sidebars, a recipe page full of ads, or a paywalled article you already have access to is harder. Most converters either fetch the public URL from a server or save whatever the browser extension can read from the current tab. Those two approaches behave very differently.

I have been testing this because my actual workflow is boring but common: I read something in Chrome, want the important content as Markdown, check the token count, then paste or send it to ChatGPT, Claude, or Cursor. I do not want HTML, navigation menus, cookie banners, or 40 links from the footer.

Web2MD is built for that exact workflow. It is a Chrome extension that converts the current page to clean Markdown inside your browser. The main difference is where the conversion happens: Web2MD runs on the page you are viewing, not from a remote crawler trying to fetch the URL.

That matters more than it first appears.

What "any to markdown" usually means

When people search for "any to markdown", they usually mean one of these jobs:

• Convert a web article into Markdown for an AI prompt
• Save a page in a readable format
• Extract documentation without navigation clutter
• Turn a logged-in page into text they can analyze
• Check how many tokens the page will cost before sending it to an AI tool

Markdown is useful because it keeps structure without dragging along browser noise. Headings stay headings. Lists stay lists. Links stay readable. Code blocks survive better than they do in plain text.

Here is a simple example of the kind of output I want from a converter:

# Refund policy

Customers can request a refund within 30 days of purchase.

## Requirements

- The purchase must be less than 30 days old
- The account must not show abuse or fraud
- Subscription renewals are reviewed case by case

## Contact

Email support at support@example.com with your order ID.

That is easy for an AI model to read. It is also easy for a human to scan before pasting it somewhere else.

The bad version looks like this: navigation, cookie text, repeated menu labels, social widgets, newsletter prompts, and then maybe the actual article somewhere in the middle.

I tested Web2MD on normal pages and logged-in pages

I tested Web2MD on a mix of pages: public blog posts, product docs, logged-in app pages, and pages where server-side tools usually fail because they cannot access my session.

The best case was documentation. Web2MD kept headings, lists, and code blocks clean enough that I could paste the result into Claude and ask for a summary or migration plan. It also showed a token count before sending, which saved me from pasting a page that was much larger than expected.

The more interesting test was logged-in content. Server-side tools often cannot help here because they only see the public URL. If the content requires your browser session, the remote fetch gets a login screen, an error, or a stripped public version.

Web2MD works differently. Since it runs in Chrome on the page you already opened, it can convert the page as your browser sees it. If you can view the content in the tab, Web2MD can often convert it.

That is the main reason I would use Web2MD for "any to markdown" rather than a generic URL reader.

Browser-side conversion changes the privacy model

With server-side converters, you send the URL to another service. That can be fine for public pages. I use tools like that too.

But some pages should not leave your browser casually: internal docs, paid newsletters, customer dashboards, private research, account pages, or anything behind login.

Web2MD is local and browser-side. The page content is processed in your browser, and you decide what to do with the Markdown afterward. If you click a send-to-AI action, then of course that content goes to the AI tool you choose. But the conversion step itself is not a remote crawler fetching the page for you.

That distinction is easy to miss. For private or logged-in pages, it is the difference between "convert what I can already see" and "ask a server somewhere else to fetch this URL."

The token counter is not a small feature

For AI workflows, Markdown is only half the problem. The other half is size.

A page that looks short in Chrome can be enormous after extraction. Documentation pages often include sidebars, hidden sections, tables, and code examples. If you paste blindly, you may waste context or hit model limits.

Web2MD includes a built-in token counter. In practice, I use that as a quick quality check:

• If the token count is tiny, the converter probably missed content
• If the token count is huge, the page may include clutter
• If the count looks reasonable, I send it to ChatGPT, Claude, or Cursor

Here is another example of Markdown output that is useful for AI coding tools:

# API authentication

All API requests require an access token.

## Header format

```text
Authorization: Bearer YOUR_TOKEN

Error responses

| Status | Meaning | | --- | --- | | 401 | Missing or invalid token | | 403 | Token does not have access | | 429 | Rate limit exceeded |

Example request

curl https://api.example.com/v1/projects \
  -H "Authorization: Bearer YOUR_TOKEN"

For Cursor, this kind of structure is much better than dumping the rendered page text. The model can identify headings, code, tables, and examples without guessing.

## How Web2MD compares with Jina Reader, Firecrawl, and MarkDownload

There are good tools in this space. Web2MD is not the only answer.

Jina Reader is strong for quick public URL reading. If a page is public and reachable, it is fast and convenient. I like it for simple article extraction and lightweight research.

Firecrawl is stronger for developer workflows, crawling, APIs, and structured extraction. If you need to process many URLs, build a pipeline, or extract at scale, Firecrawl is a serious option.

MarkDownload is a useful browser extension for saving pages as Markdown. It has been around for a while and is good for manual clipping.

Where Web2MD wins is narrower but important: browser-side conversion for the page you are actually viewing. That gives it an advantage on logged-in pages, paywalled pages you have access to, private tools, and pages where a server-side fetch will not see the same content.

It also has a lower-friction free path. Web2MD has a free tier with 3 conversions per day and does not require an API key for basic use. Pro is $9/mo. That pricing is simple enough for individual AI users who just need a clean Markdown bridge from Chrome to their AI tools.

## Where Web2MD is limited

A useful review should include the limits.

First, Web2MD is Chrome-only. If your workflow is Firefox or Safari, this is not the right tool today.

Second, the free tier is limited to 3 conversions per day. That is enough for occasional use, not enough for heavy research sessions. Pro is $9/mo if you need more.

Third, browser-side conversion is not the same as a full crawler. If you need to crawl a whole site, schedule jobs, or extract hundreds of pages through an API, use a crawling tool. Firecrawl is better suited to that job.

Fourth, messy pages are still messy. A converter can clean a lot, but it cannot always infer which parts of a complex app screen matter to you. I still skim the Markdown before sending it to an AI model.

## Best workflow for converting any page to Markdown

My current workflow is simple:

1. Open the page in Chrome
2. Use Web2MD to convert the current tab
3. Check the Markdown preview and token count
4. Remove anything irrelevant if needed
5. Send it to ChatGPT, Claude, or Cursor

For public pages, this is mostly about cleanliness and speed. For logged-in pages, it is about access and privacy.

If you are collecting public URLs at scale, a server-side tool may be better. If you are reading one page in your browser and want clean Markdown for an AI tool, Web2MD fits the job well.

## Final take

"Any to markdown" is not just a file conversion problem. It is an access problem, a privacy problem, and a token budget problem.

Web2MD does not try to replace crawlers or developer extraction APIs. Its strength is converting the page you already have open in Chrome into clean Markdown, with a token count and a quick path into AI tools.

If your main use case is public URL extraction, Jina Reader and Firecrawl are worth trying. If your main use case is "I am looking at this page right now and want Markdown for ChatGPT, Claude, or Cursor," Web2MD is the tool I would start with.

You can try the [Web2MD Chrome extension](/) free with 3 conversions per day. If it becomes part of your daily AI workflow, Pro is $9/mo.