Deprecated and format-only controls

How the text is hidden

Three ranges, all Cf, all default-ignorable, all rendering as nothing. U+206A to U+206F are the six controls Unicode itself formally deprecated: inhibit and activate symmetric swapping, inhibit and activate Arabic form shaping, national and nominal digit shapes. U+1D173 to U+1D17A are the musical symbol begin/end beam, tie, slur and phrase controls, and U+1BCA0 to U+1BCA3 are shorthand format controls. Any of them can be typed into a document with Alt+X in Word, a hex editor, or a numeric character reference in HTML.

Why a model still reads it

They are text, so an extractor returns them and a model reads whatever they do to tokenisation. Before this class existed they were already folded out for instruction matching — so the verdict survived — and then handed straight back inside text.canonical with no finding, which is silence rather than a downgrade.

What we do about it

The pack class deprecated-format matches the three ranges, is declared highConfidence: true, and is graded medium by the confidence-band mapping. The characters are stripped from text.canonical, and for DOCX from the cleaned file's text elements. `` pins which classes carry the high-confidence band, so a change to the grading lands in one reviewed place.

How often it fires

0% of 1,244 real documents — 400 web pages (distinct hosts), 630 PDFs (contracts and web), 150 government Word documents, 65 SEC filings, measured 2026-08-21.

This is an alert-volume number and nothing else. It says how often the alarm sounds on documents as found — not how often it is right, and not whether what it found was harmless. Documents as found may themselves carry concealment. Read it against the population named above rather than as a property of documents in general.