The web clipper I wanted after MarkDownload

I still like MarkDownload.

That is the first thing to say, because this post is not a dunk on it. MarkDownload is a good Chrome extension for saving web pages as Markdown. It is fast, it is simple, and if your goal is "clip this article into a Markdown note," it often does the job.

But after using Markdown clippers for AI work, I kept running into a different problem.

I did not only want to save a page. I wanted to send a clean version of the page to ChatGPT, Claude, or Cursor without copying navigation menus, cookie banners, sidebars, share buttons, and half the footer. I also wanted to know roughly how many tokens I was about to paste before I blew up a context window.

That is the gap Web2MD is trying to fill.

Web2MD is a Chrome extension that converts the current web page into clean Markdown inside your browser. It is built for AI workflows: read the page, clean it up, count the tokens, then copy or send it onward. I tested it alongside MarkDownload, Jina Reader, SingleFile, and Obsidian Web Clipper because those are the tools people usually mention in the same conversation.

The short version: if you mostly save public articles into a personal archive, MarkDownload and Obsidian Web Clipper are still worth considering. If you need clean Markdown from pages you are already logged into, Web2MD belongs on the list.

What changed after MarkDownload

My old flow with MarkDownload was fine for simple pages:

1. Open article.
2. Click extension.
3. Copy or download Markdown.
4. Paste into a note or AI chat.
5. Trim the junk manually.

That last step is what got annoying.

For AI work, a little junk matters. A navigation bar that looks harmless in the browser can turn into 300 lines of repeated links in Markdown. Related article blocks, newsletter forms, legal footers, and hidden mobile menus all become tokens. On long pages, that means less room for the actual prompt and less room for the model to answer.

Here is the kind of output I do not want to paste into an AI tool:

# Product documentation

Home
Features
Pricing
Blog
Sign in
Start free trial

# Product documentation

This guide explains how to configure the importer.

## Step 1: Create an API token

Go to Settings and create a token with read access.

Subscribe to our newsletter
Related posts
Privacy policy
Terms of service
Cookie settings

That is not catastrophic, but it is noisy. If you are doing this once a week, you can clean it by hand. If you are feeding pages into ChatGPT, Claude, or Cursor every day, the friction adds up.

The output I want is closer to this:

# Product documentation

This guide explains how to configure the importer.

## Step 1: Create an API token

Go to Settings and create a token with read access.

That difference is why I started looking past a normal web clipper.

Where Jina Reader is excellent, and where it cannot help

Jina Reader is one of the cleanest ways to turn public web pages into Markdown. I use it as a quick test when I want to see whether a public URL can be reduced to readable text. It is server side, which is useful because you do not have to install anything. You can pass a URL and get a clean reader view back.

That strength is also its limit.

A server side reader can only fetch what the server can reach. It does not have your browser session. It cannot see the page you see after logging into a SaaS dashboard, internal wiki, paid newsletter, private docs portal, or paywalled research page. It also means the URL is sent to a remote service for processing.

That may be fine for public pages. It is not always fine for private ones.

Web2MD works differently. It runs in the browser, on the page you already opened. If Chrome can render it and you have permission to see it, Web2MD can usually convert the visible page to Markdown. That browser side approach is the main reason I think Web2MD should be considered after MarkDownload.

It also means there is no API key to set up for the free tier. You install the extension, open a page, and convert. The free plan allows 3 conversions per day. Pro is $9 per month if you need more.

That limit is real. If you process dozens of pages a day, the free tier will feel small. But for occasional research, quick AI prompts, or testing whether the workflow fits, 3 per day is enough to evaluate it.

MarkDownload versus Web2MD

MarkDownload is still strong as a general Markdown clipper.

It is good when:

• You want to save a public article as a file.
• You already have a note taking workflow.
• You prefer a lightweight extension with minimal product surface.
• You do not need token counts or AI handoff.

Web2MD is better when:

• The page is behind a login.
• You care about keeping conversion local in the browser.
• You want to know the token count before pasting.
• You want a one click path into AI tools.
• You do not want to set up an API key.

The token counter sounds like a small feature until you use it. When I am preparing context for Claude or ChatGPT, I want to know whether a page is 2,000 tokens or 25,000 tokens before I paste it. If it is too large, I can trim sections first or send only the part I need.

That is also useful in Cursor. When I paste docs into an editor chat, I do not want to burn the whole context on navigation text and repeated boilerplate. A clean Markdown conversion with a visible token estimate makes the process less guessy.

SingleFile and Obsidian Web Clipper are solving nearby problems

SingleFile is great at preserving pages. If your goal is archival fidelity, it is hard to beat. It saves the page as a self contained HTML file, including styling and assets. That is useful for receipts, research snapshots, and anything you may need to prove or revisit later.

But an archived HTML file is not the same as clean Markdown for an AI model. SingleFile is more about preservation than extraction.

Obsidian Web Clipper is strong if Obsidian is your home base. It can capture pages into your vault and fit them into a personal knowledge system. If you live in Obsidian, that integration matters.

Web2MD is narrower. It is not trying to be your archive or your second brain. It is trying to convert the current page into AI ready Markdown. That narrower job is why the token counter and send to AI flow matter.

Firecrawl and other developer tools

Firecrawl is worth mentioning because many technical teams use it to turn web pages into Markdown or structured data. It is much more of a developer tool than a browser clipper. It can crawl sites, extract content at scale, and integrate into backend workflows.

That is powerful. It is also not the same use case.

If I am building a crawler or processing many public pages with code, Firecrawl makes sense. If I am staring at a logged in admin screen, a private Notion page, or a customer support article inside a helpdesk, I do not want to wire up an API. I want the browser extension to read what I am already allowed to see.

That is Web2MD's lane.

Privacy is not abstract here

"Runs locally" can sound like marketing filler, so here is the practical version.

If you use a server side converter, the service has to receive the URL or page content to process it. That may be acceptable for public blog posts. It may not be acceptable for:

• Internal documentation
• Customer support threads
• Paid newsletters
• Research portals
• Admin dashboards
• Draft pages
• Anything with private account data

With Web2MD, the conversion happens in your browser. That does not make every use magically risk free. You still need to think before sending the resulting Markdown to an AI tool, especially if it contains private data. But the conversion step itself does not require handing the page to a remote reader service.

That is the distinction I care about.

The limits I hit

Web2MD is Chrome only. If you use Firefox or Safari, that is a hard limit today.

The free tier is also limited to 3 conversions per day. I think that is fair for testing and light use, but it is not unlimited. Pro costs $9 per month.

And no converter is perfect. Complex web apps can have weird layouts, collapsed sections, lazy loaded content, and interactive widgets that do not map cleanly to Markdown. In my testing, the best results came from article pages, documentation pages, help center pages, and text heavy app screens. Highly visual pages still need manual review.

That is not unique to Web2MD. It is just the nature of converting the web into Markdown.

Who should try Web2MD

Try Web2MD if you already use MarkDownload but find yourself cleaning output before sending it to AI.

It is especially useful if you often work with pages that Jina Reader cannot reach because they require login. That is the case where browser side conversion matters most.

I would keep the comparison simple:

• Use Jina Reader for quick public URL to Markdown conversion.
• Use MarkDownload for classic Markdown clipping.
• Use SingleFile when you need a faithful archive.
• Use Obsidian Web Clipper if Obsidian is the destination.
• Use Web2MD when you want private, browser side Markdown for AI tools.

If you want more detail on the workflow, the Web2MD site has a short overview of the extension at /features and the current free and Pro limits at /pricing.

I do not think everyone needs another web clipper. But if your actual job is "turn this messy browser page into clean context for ChatGPT, Claude, or Cursor," Web2MD is the one I would add to the shortlist after MarkDownload.

You can try it free at web2md.org and see whether 3 conversions are enough to judge the fit.