Perception
What a language model can perceive of a web page when nothing is rendered as an image: general knowledge, useful even if you never install Dinoer.
A language model does not see a web page. Given a screenshot, it gets pixels: it can say that the page looks wrong, but it cannot act on anything, because nothing in an image can be addressed. Given the raw HTML, it gets everything at once, including hundreds of lines of layout markup.
Dinoer gives it neither. No image at all, in any mode. It gives the structure that the browser already builds, as text.
Two views, both text
The accessibility tree is the structure the browser builds for screen readers: headings, roles, links, states. The extracted text is the content of the page with its markup and noise tags removed, for when what matters is what the page says rather than how it is organised.
Neither replaces the other. The tree tells you what a page is; the text tells you what it says. A page can have a rich tree and little content, or the reverse.
Reproduce it
The outputs in this section were captured against https://example.com on
14 August 2026:
/opt/dinoer/venv/bin/python3 /opt/dinoer/shot.py \
--url https://example.com --a11y --guide-version 1.6