Any page becomes clean Markdown.
Copying and pasting a page into a model brings along the menu, banner, footer and script, and you pay tokens for all of it. Here, you get only the content, in Markdown, with the source in the header.
- Module
- Collection
- Input
- A URL
- Runs on
- Rovemark server
- Cost
- Free, no signup
The text that matters, without the rest of the page attached
A web page is made for a browser: it has a menu, banner, newsletter form, footer and a stack of scripts. When you copy it into a model, it all comes along, takes up context, costs tokens and gets in the way of reading what matters.
The tool fetches the page, drops what is not content, picks the main block and converts it to Markdown: headings become #, lists become -, tables become tables, and links are made absolute. At the top sits a header with the source URL, so the model knows where it came from.
- 01
You paste the page address
One page, no signup. The tool fetches straight from the origin, the way a crawler would.
- 02
Everything that is not content is dropped
Script, style, menu, footer and sidebar are discarded. What remains is the block with the most continuous text, where the content lives.
- 03
Out comes Markdown, ready to paste
With a word count and a token estimate, so you know how much of it will fit in the model’s context.
- 01
Why Markdown and not plain text?
- Because Markdown preserves hierarchy at almost no token cost. The model sees what is a heading, what is a list item and what is a table, and responds better to structured text than to a continuous block where everything has the same weight.
- 02
Does it work on any site?
- It works where the text is in the served HTML. Sites that only build content after JavaScript return an empty shell, and the tool warns you when that is the case, instead of delivering an empty file while pretending it succeeded.
- 03
Does the tool store the page I converted?
- No. The page is fetched, converted and returned within the same request. Nothing is saved on the server side.
- 04
Does it work behind a login or paywall?
- No. The request is made without a session, exactly like an anonymous visitor, so you receive what anyone who is not logged in would see. Content behind login requires the Collection module, which works with an authenticated session.
- 05
Why is the token count an estimate?
- Because each model has its own tokenizer and the exact number changes from one to another. The estimate uses the average ratio per character, adjusted for writing, enough to size context and cost, not to settle an invoice.
- 06
Can I convert a whole site at once?
- Not here: the free tool handles one page at a time. A whole site, with scheduling and API delivery, is what the Collection module handles, the same engine, running at scale.
One page is easy. What about a thousand?
This tool does one page, right now. Crawling a whole site, repeating it daily and delivering the result over an API is what the Collection module does, on the same engine.