Download the screenshots and diagrams from a docs site
Paste a documentation URL and the scanner lists what the page carries: screenshots, exported diagrams, and the inline SVG most docs sites use for icons and architecture drawings. Diagrams that a JavaScript library draws after the page loads are the exception, and they are the one thing to expect missing.
What this does
- Measured on a live framework site, where the scanner found 100% of the images a real browser loads, checked against headless-Chromium ground truth rather than estimated.
- The grid arrives holding exactly the two things this page names — the page's own images and its inline diagrams — with the site's favicons and social preview tags left out of the way.
- Diagrams written directly into the page are included rather than skipped. They carry no URL of their own, so they are rebuilt as files; the SVG page covers that mechanism in detail.
- Docs screenshots are often far larger than they appear on the page. One 4 KB range request per image reports its real pixel size and byte size without downloading the file.
What a docs page actually holds
Documentation is an unusual mix. Screenshots are raster files, usually PNG and usually much larger than the width they display at. Diagrams are either SVG files or SVG written straight into the page, and the second kind has no URL at all — it is markup, not a link to a picture.
Both are read here, which is why the format filter is the useful control on a docs page: PNG isolates the screenshots, SVG isolates the drawings.
The gap worth knowing about
The one class this cannot reach is a diagram drawn by JavaScript at read time. Mermaid is the common case: the page ships a text description of the diagram and a library turns it into a picture in your browser. Nothing in the served HTML is an image, so nothing is found. That is a real limitation rather than a temporary one — reaching it would mean running the page in a browser, which is the architecture this tool deliberately does not have.
What it won't do
- Diagrams rendered by JavaScript are not there. Mermaid, chart libraries and anything drawn into the page after it loads leave nothing in the HTML for a static reader to find. On a documentation site this is the most likely gap by far.
- One page per scan, no crawl, and 30 scans an hour. A documentation site is many pages and there is no mode that walks them — you paste the ones you want.
- Versioned documentation gives you the version in the URL you paste. Nothing here resolves "latest" or follows a version switcher.
- An inline diagram coloured by the page's stylesheet arrives as geometry without that styling. The shape is right; the palette may not be, because the colour lives outside the element being extracted.
- One site was measured, and it is a framework site rather than a pure documentation host. Treat the number as one solid data point, not a class result.
- If the site's robots.txt asks automated tools not to read the page, the scan stops and says so. There is no override.
Questions
Will my Mermaid diagrams come through?
No. Mermaid and similar libraries ship a text description in the HTML and draw the picture in your browser afterwards, so there is no image in the page for a static reader to find. What you can extract is whatever the site exported as a file — many docs sites do both.
How do I separate the screenshots from the diagrams?
Use the format filter in the sidebar. Screenshots are almost always PNG and diagrams are almost always SVG, so ticking one or the other splits a docs page cleanly in a single click.
Can I get every page of the documentation at once?
No. This scans one page per request, with no crawl and no queue, and the allowance is 30 scans an hour. For a whole site that means pasting URLs one at a time, which is honest work for a handful of pages and the wrong tool for hundreds.
Why did an icon lose its colour?
Because the colour was never in the icon. Docs sites commonly leave an inline SVG's fill to the surrounding stylesheet so it can adapt to a dark theme, and that styling is outside the element extracted. You get the shape, which is the reusable part.
Related
Images belong to their creators. Only download what you have the right to use. See all tools or read what our requests do.