An hour of Acrobat tag‑tree cleanup, in under a minute.

No. Accessibility is a judgment about the whole document: correct semantic tags, meaningful alt text, a logical reading order, sufficient colour contrast, and plain language. Those are decisions only you can make, and most of them have to happen in InDesign before you export. What we do is automate the repetitive structural cleanup that would otherwise eat hours of your time in Acrobat’s Tags tree.

Effective prompting needs spec vocabulary you’d never need otherwise — Matterhorn codes, MCID chains, /THead scoping. You bring a cryptic PAC report and a deadline; we already know the fixes.

It may work, but it is not tested, supported, or guaranteed. This tool is built around the specific tag structures InDesign produces. If your upload’s /Producer field does not include “Adobe InDesign” we show a warning on the upload page, but we do not refuse the file.

No. Uploaded PDFs live in /tmp/<uuid>/ only long enough to audit, process, and deliver. As soon as you choose Download, Email, or Download and Email on the result page, we record a tombstone row for the delivery, stream the file to your chosen destination, and delete the working directory. No file content ever touches our database or long-term disk.

Sign in and click Manage subscription in the top nav. You will land in Stripe’s Customer Portal where you can cancel immediately or at the end of the current period. No email exchange required.

The “What this tool does” section above groups every capability into six areas — tag-tree hygiene, links & TOCs, lists / headings / navigation, figures & media, tables, and catalog metadata — each expandable for the specific passes underneath. In total the pipeline runs roughly two dozen discrete structural transformations per upload, but the count isn’t the selling point; the hour of Acrobat Tags-panel labour they replace is.

Our pre-flight audit reports on 22 distinct Matterhorn-protocol categories — the machine-checkable subset where the fix can be applied without re-authoring the source document. The pipeline auto-fixes 18 of those 22; the remaining four are findings only an author can resolve (alt text, heading hierarchy, table header-row style, and complex content-stream shapes our auto-fix declines on for safety).

Measured on a real 29-page Adobe InDesign export: our audit flagged 584 individual items before processing; 578 cleared after the pipeline ran — 99% of items, 7 of 9 flagged categories. The remaining six items are source-side fixes the tool flagged for manual attention, not pipeline regressions. The full per-category report lives in docs/measurements/ in the repo, and the methodology is documented there too.

Two honest caveats: the percentage is measured by our own audit (which checks the same Matterhorn checkpoints PAC 2024 reports on, but groups findings differently — PAC’s row count on the same file will typically be higher). And document shape matters enormously: a clean InDesign export will see a different number than a Word document with deeply nested tables. The 99% on this fixture is documented ground truth, not extrapolated marketing.

Sometimes, but we make no guarantees. The tool is designed around Adobe InDesign export shapes — <Span> consolidation, Form XObject artifact handling, the /Tag___TOC* naming convention — and those don’t always apply to Word output. The passes that do transfer well include tag-tree cleanup, TOC entry cleanup, link-destination validation and repair, bookmark generation from headings, and /Placement /Block on inline tags. The ones that are InDesign-specific (Span consolidation, Form XObject artifact-wrap, Table-in-P lift) no-op on Word PDFs rather than misbehave. Upload a sample and check the pre-flight audit — it tells you what’s applicable before you process. Always validate the output with PAC 2024.

Yes. The configure screen renders an “Apply this fix” checkbox next to each automatic pass — default checked. Uncheck any you’d prefer to handle yourself or whose output you want to preserve as-is: redundant <Span> consolidation, multi-line MCID merging, tag-tree cleanup, list-label splitting, lifting standalone figures out of paragraph wrappers, TOC structural wrapping, TOC entry cleanup, retagging TOC links as references, populating link tooltips, repairing broken link destinations, bookmark generation, or structure-based /Tabs /S. The rest of the pipeline still runs, and the result page lists anything you opted out of so you know what PAC will continue to flag. The mandatory PDF/UA baseline (language, XMP metadata, parent-tree integrity) always runs.

ACHECKS — the same team that runs the ACHECKS accessibility evaluation platform. This tool is a standalone product, separate from the ACHECKS account system and billed independently.