What this deprecated HTML tags checker looks for
Paste a template, a component or a full page, and this deprecated HTML tags checker walks the markup for two separate things: elements the HTML Living Standard has declared obsolete, and presentational attributes that CSS replaced. Each hit comes back with the line, the element it appeared on, and the declaration or tag to write in its place. Counting matters as much as naming, because forty font tags in one template is a search and replace, while one in a legacy include is an archaeology problem.
Parsing happens in this tab, on the text in the box. Nothing leaves the browser, which also explains why the field asks for source instead of an address: a page cannot read another site's HTML without a server standing in the middle, and that server would have to see your staging URL.
Obsolete is a promise to old pages, not a licence for new ones
There is a gap in HTML between what user agents must handle and what authors may write, and every
element listed here lives in it. Browsers keep drawing <center> and <font> because millions of
documents depend on them and nobody wants a blank screen. The specification is separately explicit
that these must not appear in new markup. Both statements are true at once, which is why "but it
works" never settles the argument.
A second group never worked again. <blink> stopped animating when Firefox 23 dropped it in 2013.
<applet> lost its runtime when browsers removed plugin support. <keygen>, <isindex> and
<menuitem> are gone from every engine. Markup like that is not a style problem — it is dead code
occupying space in a file somebody still has to read. The tool separates the two cases, because they
deserve different urgency: one is tidying, the other is deletion.
The real bill, which is not a ranking
Nobody should promise that removing a <font> tag moves a search position. Google parses this
markup happily and has never suggested otherwise. The costs are elsewhere and they are more
concrete.
Accessibility is the sharpest one. A <marquee> moves text that a reader cannot pause, which fails
WCAG 2.2.2 outright and makes a paragraph unusable for anyone with a vestibular disorder or a
reading difficulty. A layout table carrying valign and cellpadding is announced by a screen
reader as a table with rows and columns that mean nothing. And <acronym> was replaced by <abbr>
partly because assistive technologies had two elements to support where one would do.
Maintenance is the quieter one. A colour written into a hundred <font> tags cannot be changed by a
stylesheet, so a rebrand becomes a find and replace across templates instead of a variable. Every
presentational attribute is a rule that CSS cannot override without !important, and each one makes
the next developer trust the stylesheet a little less.
Where the exceptions are
Two exceptions are worth knowing before you delete anything. Email is the first: Outlook for Windows
has rendered messages with the Word engine since 2007 and ignores most layout CSS, so bgcolor,
valign and nested tables are the working technique there rather than a mistake. Tick the email box
and the tool keeps listing them while dropping the scolding.
The second is width and height on media. On <img>, <video>, <canvas> and <iframe> those
attributes hand the browser an aspect ratio before the file downloads, which is what keeps the page
from jumping. This checker deliberately ignores them there and reports them only on tables, cells,
<hr> and <pre>, where they really are leftovers.
When the markup is clean here, the heading structure checker covers the outline, the image alt checker covers the images, and the mixed content checker covers anything still loaded over plain HTTP.