Same class as the `omarchy-update-stay-awake` `stat %F` bug (#13320, #13376, #13385, #13892, #14087 — fixes pending in #13352 / #14023), but in three other files. Each parses localized command output against English literals without pinning `LC_ALL=C`.

Verified still present on `master` today (2026-10-08) and in `omarchy-dev 4.0.0.r6720.g8e02fc8-1` / `omarchy 4.0.4-1`.

---

## 1. `bin/omarchy-provision-owner:902` — cryptsetup "Key slot" — **high**

```bash
new_slot=$(cryptsetup open --test-passphrase --verbose --key-file <(printf '%s' "$password") "$device" 2>&1 |
  grep -o 'Key slot [0-9]* unlocked' | grep -o '[0-9]*' | head -1)
```

`cryptsetup` is gettext-localized. `zh_CN` ships:

```
msgid  "Key slot %i unlocked."
msgstr "密钥槽 %i 已解锁。"
```

so the `grep` matches nothing, `new_slot` is empty, and `rekey_luks()` takes the `[[ -z $new_slot ]]` branch:

```
could not identify the user's LUKS slot after re-key; keeping the staged key for retry
```

→ the LUKS re-key in first-boot deferred provisioning / factory reset never completes; it keeps the staged throwaway key and retries forever.

Fix: `LC_ALL=C cryptsetup open --test-passphrase ...`

---

## 2. `bin/omarchy-hibernation-setup:80` — `free` — **medium**

```bash
MEM_TOTAL_HUMAN=$(free --human | awk '/Mem/ {print $2}')
```

`free` (procps-ng) is localized. Under `zh_CN.UTF-8` the rows are `内存：` / `交换：`, so `awk '/Mem/'` matches nothing and the confirmation reads *"Use  on boot drive to make hibernation available?"* (missing the size).

```console
$ LC_ALL=zh_CN.UTF-8 free --human | awk '/Mem/ {print $2}'
                      # empty
$ LC_ALL=C free --human | awk '/Mem/ {print $2}'
31Gi
```

Fix: `MEM_TOTAL_HUMAN=$(LC_ALL=C free --human | awk '/Mem/ {print $2}')`

---

## 3. `install/user/hardware/asus/fix-audio-mixer.sh:9` — `aplay` — **medium**

```bash
card=$(aplay -l 2>/dev/null | grep -i "ALC285" | head -1 | sed 's/card \([0-9]*\).*/\1/')
```

`aplay` is localized; alsa-utils ships translations for `de`, `eu`, `fr`, `ja`, `ka`, `ko`, `sk` (de: `Karte %i: %s ...`, fr: `Carte : ...`). Under those locales the `sed` leaves the whole line in `$card`, so `amixer -c "$card"` fails and the ALC285 Master-unmute silently no-ops.

```console
$ LC_ALL=de_DE.UTF-8 aplay -l | head -1
Karte 0: ...
```

(Not `zh_CN` — there is no `alsa-utils.mo` for Chinese — but it does hit de/fr/ja/ko/eu/ka/sk users.)

Fix: `card=$(LC_ALL=C aplay -l 2>/dev/null | ...)`

---

## Why these are oversights, not intentional

The project already pins the locale whenever it parses command output — 21 call sites, e.g.:

- `omarchy-network-band:54` — `LC_ALL=C nmcli ...` (with a comment explaining why)
- `omarchy-update-pacman:19` — *"LC_ALL passes through so callers that parse ... can pin the locale."*
- `omarchy-update-system-pkgs:26` — `LC_ALL=C omarchy-update-pacman -Syu ...`
- `omarchy-upgrade-to-quattro:968` — `LC_ALL=C pacman -Qi ... | awk '/^Required By/'`
- `omarchy-security-functions:63` — `LC_ALL=C /usr/bin/sudo -h`
- `omarchy-update-stay-awake:143` and `:464` — `local LC_ALL=C`, `LC_ALL=C /usr/bin/od`

The `stay-awake` helper already pins the locale twice; only its `stat` call was missed. These three are the same omission.

## Suggested general fix

A small lint/test that flags locale-sensitive command output parsed against English literals without `LC_ALL=C` (e.g. `stat %F`, `pacman -Qi` field names, `free` row labels, `aplay`/`cryptsetup` messages) would catch this whole class at once.