diff --git a/Cargo.lock b/Cargo.lock
index 50dfbb5135..82d92156ae 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -1064,7 +1064,7 @@ dependencies = [
 [[package]]
 name = "craft-proof"
 version = "0.1.0"
-source = "git+https://github.com/storytold/craft-libs?rev=e6e948f7d5cd14ebefae829a67aafdec33f1e993#e6e948f7d5cd14ebefae829a67aafdec33f1e993"
+source = "git+https://github.com/storytold/craft-libs?rev=0675d583417c1f340749dc5a37d0f00dec554502#0675d583417c1f340749dc5a37d0f00dec554502"
 dependencies = [
  "log",
  "miniz_oxide 0.9.1",
diff --git a/Cargo.toml b/Cargo.toml
index e120d82d82..8657ec71b4 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -27,7 +27,7 @@ wordcraft-pdf = { path = "crates/pdf" }
 # Pinned like the other craft-libs crates (each has its own rev). This rev is the head of
 # storytold/craft-libs#483 (SCOWL English, stacked on #482, which adds craft-proof); move it
 # to the merge commit once both are merged.
-wordcraft-proof = { package = "craft-proof", git = "https://github.com/storytold/craft-libs", rev = "e6e948f7d5cd14ebefae829a67aafdec33f1e993", version = "=0.1.0" }
+wordcraft-proof = { package = "craft-proof", git = "https://github.com/storytold/craft-libs", rev = "0675d583417c1f340749dc5a37d0f00dec554502", version = "=0.1.0" }
 wordcraft-engine = { path = "crates/engine" }
 wordcraft-mcp = { path = "crates/mcp" }
 wordcraft-zotero = { path = "crates/zotero" }
diff --git a/craft-i18n.rev b/craft-i18n.rev
index f0674ea67c..60aae6a4c3 100644
--- a/craft-i18n.rev
+++ b/craft-i18n.rev
@@ -1 +1 @@
-efb1127ac65b238ccfe7df1893f33e91b924b96e
+059cbf95c2a3eb95d637ca8584faca1129d17f48
diff --git a/docs/localization-parity.md b/docs/localization-parity.md
index 4ddc782f52..545a4bc507 100644
--- a/docs/localization-parity.md
+++ b/docs/localization-parity.md
@@ -128,7 +128,7 @@ Set Proofing Language › Set As Default keeps the chosen language in user prefe
 
 | Language | Spelling | Suggestions | Grammar | Hyphenation |
 |---|---|---|---|---|
-| English: American (`en-US`, other regions, text with no language), British spelling (`en-GB`, `en-AU`, `en-NZ`, `en-IE`, `en-ZA`, `en-IN`…), Canadian (`en-CA`) | SCOWL 2026 (`rel-2026.02.25`, size 60) Hunspell dictionaries, one per English: en-US, en-GB `-ise` (also the other British-spelling Englishes) and en-CA, embedded from craft-i18n `language-data/` and read by the affix engine (ordinals such as `21st` and `112th` by `COMPOUNDRULE`). A word is right when any of the three or our supplement (`en-extra.txt`) knows it, unless our list of ~280 American/British pairs (craft-i18n [`language-data/en-us/en-variants.txt`](https://github.com/storytold/craft-i18n/blob/main/language-data/en-us/en-variants.txt)) marks it as the other dialect's: each English flags the other dialect's spellings (`colour` in en-US; `color`, `center`, `analyze` in en-GB; `organise` in en-CA), British English accepts `-ize` and `-ise` (Oxford spelling, as Word does; SCOWL's en-GB `-ise` alone would flag `organize`), and a dialect's form of a known word is right even if the list lacks it (`realisation`). An English Hunspell dictionary in File › Options › Proofing › Dictionaries folder comes first (the document's own English); without the SCOWL files, an installed Hunspell English, then the Moby word list (public domain, ~165k words) built into craft-proof. No system dictionary needed | the dialect's own spelling first (`colour` → `color`), then the dialect's SCOWL dictionary's suggestions (its TRY/REP tables, one typo, a stem scan), in the dialect's spelling | repeated word, `a`/`an`, sentence capital, space before punctuation, double space | Moby dictionary, our Liang patterns, heuristic |
+| English: American (`en-US`, other regions, text with no language), British spelling (`en-GB`, `en-AU`, `en-NZ`, `en-IE`, `en-ZA`, `en-IN`…), Canadian (`en-CA`) | SCOWL 2026 (`rel-2026.02.25`, size 60) Hunspell dictionaries, one per English: en-US, en-GB `-ise` (also the other British-spelling Englishes) and en-CA, embedded from craft-i18n `language-data/` and read by the affix engine (ordinals such as `21st` and `112th` by `COMPOUNDRULE`). A word is right when the English's own dictionary or our supplement (`en-extra.txt`) knows it (never because another English's does: `colourblind` is wrong in en-US), unless our list of ~280 American/British pairs (craft-i18n [`language-data/en-us/en-variants.txt`](https://github.com/storytold/craft-i18n/blob/main/language-data/en-us/en-variants.txt)) marks it as the other dialect's: each English flags the other dialect's spellings (`colour` in en-US; `color`, `center`, `analyze` in en-GB; `organise` in en-CA), British English accepts `-ize` and `-ise` (Oxford spelling, as Word does; SCOWL's en-GB `-ise` alone would flag `organize`), and a dialect's form of a known word is right even if the list lacks it (`realisation`). Text with no language takes any English's word. An English Hunspell dictionary in File › Options › Proofing › Dictionaries folder comes first (the document's own English); without the SCOWL files, an installed Hunspell English, then the Moby word list (public domain, ~165k words) built into craft-proof. No system dictionary needed | the dialect's own spelling first (`colour` → `color`), then the dialect's SCOWL dictionary's suggestions (its TRY/REP tables, one typo, a stem scan; a missing space only between two words of 3+ letters, last), in the dialect's spelling | repeated word, `a`/`an`, sentence capital, space before punctuation, double space | Moby dictionary, our Liang patterns, heuristic |
 | Russian (`ru`) | A Russian Hunspell dictionary installed on the system (`ru_RU` from the OS, a package manager, LibreOffice's `dict-ru` or the `WORDCRAFT_DICTIONARIES` folder; UTF-8, KOI8-R or Windows-1251; not bundled, so no Russian spelling marks without it), read by the same affix engine; `ё` optional (`елка`/`ёлка`, `еще`/`ещё`); hyphenated words whole, then by parts (`кто-то`, `по-русски`, `кое-где`, `северо-западный`, `ИИ-консультант`); solid compounds (`высокоэластичный`), `не` with adjectives, participles and verbal nouns (not verbs), productive prefixes (`ультра-`, `эко-`) and prefixed verbs; abbreviations with endings (`ГИПа`) and camel case left alone; our supplement of ~200 modern words with their declensions (craft-i18n [`language-data/ru/ru-extra.txt`](https://github.com/storytold/craft-i18n/blob/main/language-data/ru/ru-extra.txt): `маркетплейс`, `промокод`, `интеграция`, `колеровать`) | `ё`/`е` first, then slips within letter groups (unstressed vowels, voiced and voiceless pairs, `ь`/`ъ`), the common confusions (`жы`, `чя`, `тся`/`ться`, doubled consonants, unpronounced consonants: `здраствуй` → `здравствуй`), a missing space (`незнаю` → `не знаю`; not a preposition before a verb), one typo, then a stem scan | Russian messages: repeated word, space before punctuation, double space, sentence capital (knows `т. е.`, `г.`, `руб.`, `ул.`, initials, numbers, dialogue dashes), possibly missing comma before `что`/`чтобы`/`если`/`хотя`/`поскольку`/`который` and before `потому что` (not after conjunctions and particles such as `и`, `только`, `даже`, `так`, `для того`; not after a preposition; not in `хотя бы`, `что-то`), doubled punctuation | The rules of Russian word division as code (no pattern file): a vowel on each side, two letters at least, `ь`/`ъ`/`й` kept with the letter before, a break after `ъ`, a consonant cluster divided after its first consonant |
 | Polish (`pl`) | The Hunspell `pl_PL` dictionary installed on the system (OS package, Homebrew, LibreOffice, the folder set in File › Options › Proofing › Dictionaries folder, or `WORDCRAFT_DICTIONARIES`; not bundled, so no Polish spelling marks without it; the same pane shows which file each language found), read at runtime by a Hunspell-style affix engine (craft-proof [`src/affix.rs`](https://github.com/storytold/craft-libs/blob/main/crates/proof/src/affix.rs) in craft-libs); names capitalized, compounds part by part, `Kennedy'ego` | missing diacritics first (`zolw` → `żółw`), the dictionary's confusion table (`żeka` → `rzeka`, `chuśtawka` → `huśtawka`), missing space (`napewno` → `na pewno`), one typo, then a stem scan within two typos | Polish messages: repeated word, space before punctuation, double space, sentence capital (knows `np.`, `r.`, initials, ordinals, dialogue dashes), missing comma before `że`/`który`/`ponieważ`/`gdyż`/`aby`/`żeby`/`jeśli`/`jeżeli`/`gdyby`/`lecz` (not after `mimo`, `chyba`, `tak`, prepositions…), `we`/`ze` (`we wtorek`, `ze szkoły`), doubled punctuation | TeX Polish patterns (Kołodziejska, Jackowski, Ryćko; MIT option) with their 20 exception words; a Polish syllable heuristic for letters outside the alphabet |
 | Portuguese (`pt`, any region) | The Hunspell `pt_BR` dictionary installed on the system — VERO and its derivatives (same places as Polish; not bundled, LGPL/MPL), read at runtime by the same affix engine, which reads its `FLAG UTF-8`, 106 flags, continuation classes and `FORBIDDENWORD` spellings; names capitalized, compounds part by part (`guarda-roupa`, `ex-marido`) | missing accents first (`coraçao` → `coração`, `saude` → `saúde`), the dictionary's confusion table (`seje` → `seja`, `excessão` → `exceção`), missing space (`nãoé` → `não é`), one typo, then an accent-folding stem scan | Portuguese messages: repeated word (the emphatic doubles are allowed), space before punctuation, double space, sentence capital (knows `etc.`, `pág.`, `Dr.`, initials, ordinals, dialogue dashes), doubled punctuation; no comma rule — `que` opens a clause only sometimes | TeX Portuguese patterns (Rezende, Almeida, Araujo, Benevides; BSD 3-clause) with their 2 exception words |
diff --git a/packaging/linux/flatpak/ai.storyteller.wordcraft.yml b/packaging/linux/flatpak/ai.storyteller.wordcraft.yml
index 73d1a6bf64..0e6b54c47e 100644
--- a/packaging/linux/flatpak/ai.storyteller.wordcraft.yml
+++ b/packaging/linux/flatpak/ai.storyteller.wordcraft.yml
@@ -126,6 +126,6 @@ modules:
       # the two equal).
       - type: git
         url: https://github.com/storytold/craft-i18n
-        commit: efb1127ac65b238ccfe7df1893f33e91b924b96e
+        commit: 059cbf95c2a3eb95d637ca8584faca1129d17f48
         dest: craft-i18n
       - cargo-sources.json