Start a project

Tool 03 of 194 · AI visibility

llms-full.txt Builder

Paste, drop or fetch your key pages and get one clean Markdown file of their real content, without menus, footers, cookie banners or repeated promos.

llms-full.txt Builder at a glanceIllustration: 1, Your pages; 2, processed in your browser; 3, llms-full.txt.llms-full.txt Builder</>1Your pages3llms-full.txt2In yourbrowser
1 Your pages 2 processed on your device 3 llms-full.txt

llms-full.txt Builder

Pages

Or add .html files, or fetch up to 25 pages below.

Fetch pages from a live site (up to 25, one site)

    Each page is read once by LeejTools/1.0, one at a time; leej.uk keeps only a count.

    About the site

    Each page keeps its main content (main, article, or the densest block of text). Navigation, headers, footers, cookie banners and any paragraph repeated on most pages are left out.

    Result

    TitleURLMain content fromWords keptWords left outRepeats removedCheck
    llms-full.txt
    llms.txt (index)

    01

    What it does

    Extracts each page's main content, converts it to Markdown, removes blocks repeated across pages and bundles llms-full.txt and llms.txt.

    • Main content from main, article or text-density scoring
    • Headings, lists, tables (colspan), code and absolute links
    • Repeated blocks across pages removed
    • Words kept and left out for every page

    02

    Private by design

    Everything you paste or upload stays in your browser. Only when you press Fetch does leej.uk read the address you give, once, as LeejTools/1.0, keeping nothing but a daily count.

    03

    What you get

    llms-full.txt and a matching llms.txt, with a Copy and a Download button on every result.

    • llms-full.txt
    • Markdown
    • AI visibility
    • Content

    Why it matters

    llms-full.txt puts the text of a site's important pages into one Markdown file, so an AI assistant or a developer's tool can read the lot in one go. Copying pages by hand drags in menus, cookie notices, footers and the same promo paragraph on every page, which wastes the reader's context window and buries the content.

    This builder finds each page's main content the way reader modes do (the main or article element, or the block with the most text and fewest links), converts it to clean Markdown with absolute links, then compares pages and removes paragraphs repeated on most of them. You get the full file, an llms.txt index that points to it, and a count of what was kept on each page.

    How it works

    1. InputPage HTML, pasted or fetched
    2. In your browserMain content kept, menus and repeats dropped
    3. OutputOne Markdown file with token counts

    Where it fits in a designer's workflow

    1. Plan
    2. Design
    3. Build
    4. Test
    5. Launch
    Giving AI tools a clean copy of your content.

    By the numbers

    Words kept and left out per sample pageHartley Oak: solid oak furniture m: kept: 48 words; Hartley Oak: solid oak furniture m: left out: 26 words; How to care for an oak table: kept: 80 words; How to care for an oak table: left out: 38 words; Delivery and returns: kept: 40 words; Delivery and returns: left out: 40 wordsHartley Oak: solid oak furniture m: kept48 wordsHartley Oak: solid oak furniture m: left out26 wordsHow to care for an oak table: kept80 wordsHow to care for an oak table: left out38 wordsDelivery and returns: kept40 wordsDelivery and returns: left out40 words
    Counted by this tool on its three sample pages: 168 words of main content kept, 104 left out as menus, footers and repeated promos (38% of the text). The main block is found from <main>, <article> or, failing those, text density. Source: llms.txt proposal.
    Show the numbers as a table
    PageMain content fromWords keptWords left out
    Hartley Oak: solid oak furniture made in Leeds<main>4826
    How to care for an oak tabletext density8038
    Delivery and returns<article>4040

    How to use it

    1. Paste each page's HTML after a '=== address' line, add .html files, or fetch up to 25 pages.
    2. Set the site name and summary.
    3. Check the words kept for each page; a page with almost none is probably built by JavaScript.
    4. Download llms-full.txt and llms.txt and put both at your site's root.

    FAQ

    Why does a page have almost no words?

    Its content is built by JavaScript after the page loads, so the HTML has little text. Paste the rendered HTML from your browser's developer tools instead.

    What counts as a repeated block?

    A paragraph or list that appears in at least 60% of the pages you add (and at least two). Headings are always kept.

    How big should llms-full.txt be?

    There's no fixed limit, but smaller is more useful. Include the pages that explain what you do, not every blog post.

    Are my pages uploaded?

    Pasted HTML and files stay on your device. Pages you fetch are read once by leej.uk's fetcher, which keeps no content.

    Work with me

    Want a site that gets these right?

    I build fast, accessible websites and web apps that pass these checks from day one. Tell me about your project and I'll reply personally with honest next steps.

    All developer tools · Back to leej.uk