probe-rtl.js is Arabic-specific. Hebrew shares the direction but almost none of the language rules, so it needs its own checks rather than an extension of the Arabic ones.
What is already shared — everything in the RTL layout, bidi, mirroring and truncation sections applies unchanged.
What differs and would need writing:
- Plural agreement: Hebrew has singular / dual / plural with grammatical gender, so a
{count} {label} template breaks differently than in Arabic.
- Numerals: Hebrew uses Western digits, so the mixed-numeral check does not apply — but the numeral-gender agreement rule does.
- Gendered UI copy: Hebrew marks gender in verbs and adjectives, so the gendered-imperative check has a direct analogue, with a different neutral-form strategy.
- Niqqud is normally absent in UI text; its presence or absence affects line-height needs.
- Final letter forms (ך ם ן ף ץ) — worth a check for text that has been programmatically truncated mid-word.
Suggested shape: a probe-he.js alongside probe-rtl.js, with SKILL.md selecting by page language. Reuse the element-level text collection and the layout sections rather than reimplementing them.
Requires someone who reads Hebrew. Open to discussing the scope before any code.
probe-rtl.jsis Arabic-specific. Hebrew shares the direction but almost none of the language rules, so it needs its own checks rather than an extension of the Arabic ones.What is already shared — everything in the RTL layout, bidi, mirroring and truncation sections applies unchanged.
What differs and would need writing:
{count} {label}template breaks differently than in Arabic.Suggested shape: a
probe-he.jsalongsideprobe-rtl.js, withSKILL.mdselecting by page language. Reuse the element-level text collection and the layout sections rather than reimplementing them.Requires someone who reads Hebrew. Open to discussing the scope before any code.