Taggart
Accessible PDF viewer: one script, honest about the file
Every PDF on your site opens in a reading view built from the document's own tags — outline, read-aloud, dark mode. Free, no account, nothing uploaded.
One line, and every PDF on the site opens in a reading view
<script src="https://taggartpdf.com/viewer.js" defer></script>
That is the whole installation. Every link ending in .pdf gets a button beside it — Open in accessible reader — and the link itself goes on working.
What opens is not a page image with a text layer behind it. It is the document, read out of its own tag tree and rendered as HTML: headings are headings, lists are lists, tables have header cells, figures announce their alternative text. A screen reader in that view gets what a screen reader is supposed to get, and so does a keyboard, and so does anyone who needs the text at 24 points on a dark background.
What it gives the reader
- An outline from the document’s real headings — not from font sizes.
- Read aloud, following the tagged reading order, highlighting as it goes.
- Dark mode and text size, because a PDF page is a fixed rectangle and a reading view does not have to be.
- A keyboard-complete dialog: Escape closes it, focus is trapped while it is open and returned where it came from.
- The original, one click away, always.
What it will not do
It will not tell you the document is accessible.
This is the line that separates a reading view from a compliance product, and most viewers in this market blur it. A viewer sits between the file and one reader, in one browser, with JavaScript running. It cannot help the person who downloads the PDF and opens it in Acrobat, or Preview, or a Kindle, or the reader their employer standardised on in 2014. It cannot help a regulator reading the file itself. The document is what ships; the viewer is a courtesy on top of it.
So when the file has no tags, the viewer says exactly that, in the banner, above the text:
This document has no tags. There is no structure to follow, so the text below is in the order the glyphs are painted, which is what a screen reader gets too. An outline could be guessed from type size; it is not, because a guessed outline is indistinguishable from a real one to the person who needs it.
That sentence is the product. A viewer that quietly invents an outline makes an inaccessible document feel handled, which is worse than leaving it obviously broken — it removes the reason to fix it.
The honest order of operations
- Check the file. Free, with no account up to 60 pages and no page cap with a free one. Check a PDF →
- Repair what a machine can repair, and decide the rest yourself, with the decisions recorded.
- Then drop the viewer in, so the structure you just put into the file is what readers get in the browser too.
Doing it in that order means the viewer is showing something real. Doing it the other way round means it is showing a well-lit picture of a problem you still have.
Options
<script src="https://taggartpdf.com/viewer.js" defer
data-selector='a[href$=".pdf"]'
data-label="Read this accessibly"
data-accent="#0f766e"></script>
data-selector narrows which links get the button. data-label is the button’s text. data-accent matches it to your site. There is nothing else to configure, and no key to obtain.
You can also open it yourself:
TaggartViewer.open('/reports/annual-2026.pdf', 'Annual report 2026')
Browser support
Anywhere <dialog> and ES modules work, which is every current browser. Read-aloud uses the browser’s own speech synthesis, so it speaks in whatever voices the reader already has installed — including the ones they have configured for Hebrew or Arabic. Where the document declares a language, that is the voice it asks for.
It is running on this page. The link below is a real tagged PDF — Taggart's own remediation report, on this same origin, read by the same script the snippet above installs. Open it in the reader, then download it and compare.
Frequently asked questions
Does the viewer make my PDFs accessible?
No, and any viewer that says it does is selling you something. A reading view can only present what the file contains. If the document is tagged, the outline and the reading order come from its tag tree and the result is genuinely good. If it is not tagged, there is no structure to present — the viewer says so in a banner and shows the text in the order the glyphs are painted, which is what a screen reader gets. The repair happens in the file, not in the viewer.
Do my documents get uploaded anywhere?
No. The script runs in your page and reads the PDF from wherever it already lives on your own site, same-origin. Nothing is sent to Taggart, nothing is proxied through a CDN, and there is no account. Your server logs are the only record that anyone opened anything.
Why does it not guess an outline from the font sizes?
Because a guessed outline is indistinguishable from a real one to the person who needs it. Somebody navigating by heading has no way to tell a structure the author wrote from one a script inferred from type size — and acting on a wrong one costs them more than having none. Where the tags exist, they are used. Where they do not, the viewer says so.
What does it cost?
Nothing. It is one static file on this domain with no key, no quota and no callback. If it makes your documents worth reading you will eventually want them fixed properly, and that is what Taggart sells.
Does it replace the download?
No. The link keeps working exactly as it did — right-click, middle-click, save-as, and every assistive shortcut still reach the file. The reader is offered as a separate control beside it, because hijacking a download is a surprise, and a surprise is an accessibility problem of its own.
Related guides
- How to handle PDFs on a website you are making accessibleA site audit that ignores its documents has audited the wrapper. Which PDFs are in scope, how to find them all, and the order to work through them.
- How to check if a PDF is accessible: four free checkersCheck PDF accessibility free, online or on your own machine: four checkers against PDF/UA and WCAG, what each catches, and the failures none of them see.
Run your own file through it. The validator is free and unlimited — no page cap, no watermark, and no account at all up to 60 pages — so you can see exactly what you are dealing with before deciding anything.
Last updated . Competitor prices are as published by each vendor.