@@ -65,20 +65,28 @@ specify init my-project --integration copilot --preset compliance
 ## Naming Features with the Helper Scripts
 
 When calling the bundled `create-new-feature` helper scripts directly, generated
-names retain only ASCII letters and digits. A description entirely in a non-Latin
-script, or made only of punctuation, can therefore produce an empty suffix such
-as `001-`. The scripts warn on stderr when this happens, including during a dry
-run; JSON output remains parseable.
+names retain Unicode letters and decimal digits in UTF-8, so a description such as
+`添加用户` produces `001-添加用户`. Descriptions made only of punctuation can still
+produce an empty suffix such as `001-`; the scripts warn on stderr when this
+happens, including during a dry run. JSON output remains parseable.
 
-Keep the original description and supply a readable ASCII short name:
+To choose a different name, keep the original description and supply a short name:
 
 ```bash
-bash .specify/scripts/bash/create-new-feature.sh --json --short-name user-auth "添加用户"
+bash .specify/scripts/bash/create-new-feature.sh --json --short-name 用户管理 "添加用户"
 ```
 
 The Python helper also accepts `--short-name`; the PowerShell helper uses
 `-ShortName`. A supplied short name is cleaned by the same rules, so it must
-contain at least one ASCII letter or digit.
+contain at least one letter or digit. For non-ASCII names, the Bash helper needs
+an installed UTF-8 locale and a Python 3 interpreter for Unicode classification.
+ASCII input, including tabs and newlines, is sanitized without either requirement.
+If `LC_ALL` is non-empty, Bash uses that locale rather than selecting another:
+Unicode names fail with an error if the selected locale is not usable for UTF-8
+names. With `LC_ALL` unset or empty, Bash selects an installed UTF-8 locale even
+when `LANG` or `LC_CTYPE` names a non-UTF-8 locale.
+ASCII capitals are lowercased; non-ASCII letter casing is preserved across the
+script variants.
 
 ## Check Installed Tools
 
 
@@ -112,7 +112,7 @@ read_feature_json_feature_directory() {
     fi
     if [[ -z "$_fd" ]] && command -v python3 >/dev/null 2>&1; then
         # Use Python so pretty-printed/multi-line JSON still parses correctly.
-        if ! _fd=$(python3 -c "import json,sys; d=json.load(open(sys.argv[1])); v=d.get('feature_directory'); print(v if v else '')" "$fj" 2>/dev/null); then
+        if ! _fd=$(python3 -c "import json,sys; d=json.load(open(sys.argv[1], encoding='utf-8')); v=d.get('feature_directory'); print(v if v else '')" "$fj" 2>/dev/null); then
             _fd=''
         fi
     fi
 
@@ -146,17 +146,82 @@ spec_prefix_exists() {
 
 # Function to clean and format a branch name
 #
-# Three details keep this byte-identical to the Python and PowerShell twins:
-#   * LC_ALL=C -- in a UTF-8 locale glibc resolves the a-z *range* through
-#     collation, so [^a-z0-9] keeps accented lowercase letters that
-#     re.sub(r"[^a-z0-9]", ...) and .NET's -replace both strip.
+# Three details keep this consistent with the Python and PowerShell twins:
+#   * Unicode classification uses Python: POSIX [:alnum:] differs by platform.
 #   * `--*` instead of the GNU-only `\+`, which POSIX/BSD sed reads as a literal
 #     '+', leaving repeated separators uncollapsed on macOS.
 #   * printf instead of echo, so a name of "-n"/"-e"/"-E" is text, not options.
+contains_non_ascii() {
+    LC_ALL=C grep -q '[^[:print:][:cntrl:]]'
+}
+
+UNICODE_LOCALE=""
+locale_candidates=(C.UTF-8 C.utf8 en_US.UTF-8 en_US.utf8 "${LC_CTYPE:-${LANG:-}}")
+if [ -n "${LC_ALL:-}" ]; then
+    locale_candidates=("$LC_ALL")
+fi
+for candidate in "${locale_candidates[@]}"; do
+    if [ -n "$candidate" ] && [ "$(printf 'é。' | LC_ALL="$candidate" sed 's/[^[:alnum:]]/-/g' 2>/dev/null)" = 'é-' ]; then
+        UNICODE_LOCALE="$candidate"
+        break
+    fi
+done
+
+if [ -z "$UNICODE_LOCALE" ]; then
+    UNICODE_LOCALE=C
+    if printf '%s' "${SHORT_NAME:-$FEATURE_DESCRIPTION}" | contains_non_ascii; then
+        if [ -n "${LC_ALL:-}" ]; then
+            echo "Error: A UTF-8 locale is required to create a Unicode feature name; LC_ALL=$LC_ALL is not usable" >&2
+        else
+            echo "Error: A UTF-8 locale is required to create a Unicode feature name" >&2
+        fi
+        exit 1
+    fi
+fi
+
+unicode_words() {
+    local name="${1//$'\n'/ }"
+    local separator="$2"
+    if printf '%s' "$name" | contains_non_ascii; then
+        local -a python_cmd=()
+        local override="${SPECKIT_PYTHON_EXECUTABLE:-${SPECKIT_PYTHON:-}}"
+        if [ -n "$override" ] && command -v "$override" >/dev/null 2>&1 &&
+            "$override" -c 'import sys; raise SystemExit(sys.version_info.major != 3)' >/dev/null 2>&1; then
+            python_cmd=("$override")
+        else
+            local python_line
+            while IFS= read -r python_line; do
+                python_cmd+=("$python_line")
+            done < <(_python3_command)
+        fi
+        if [ "${#python_cmd[@]}" -eq 0 ]; then
+            echo "Error: Python 3 is required to create a Unicode feature name" >&2
+            return 1
+        fi
+        printf '%s' "$name" | "${python_cmd[@]}" -c '
+import sys
+value = sys.stdin.buffer.read().decode("utf-8")
+lower = str.maketrans("ABCDEFGHIJKLMNOPQRSTUVWXYZ", "abcdefghijklmnopqrstuvwxyz")
+result = "".join(
+    char if char.isalpha() or char.isdecimal() else sys.argv[1]
+    for char in value.translate(lower)
+)
+sys.stdout.buffer.write(result.encode("utf-8"))
+' "$separator"
+    else
+        printf '%s' "$name" | LC_ALL=C tr '[:upper:]' '[:lower:]' | LC_ALL=C sed "s/[^a-z0-9]/$separator/g"
+    fi
+}
+
 clean_branch_name() {
     local name="$1"
-    local -x LC_ALL=C
-    printf '%s\n' "$name" | tr '[:upper:]' '[:lower:]' | sed 's/[^a-z0-9]/-/g' | sed 's/--*/-/g' | sed 's/^-//' | sed 's/-$//'
+    local cleaned
+    cleaned=$(unicode_words "$name" '-') || return 1
+    printf '%s\n' "$cleaned" | sed 's/--*/-/g' | sed 's/^-//' | sed 's/-$//'
+}
+
+branch_byte_count() {
+    printf '%s' "$1" | wc -c | tr -d '[:space:]'
 }
 
 # Fit a feature prefix and suffix within GitHub's branch-name limit.
@@ -165,11 +230,25 @@ fit_branch_name() {
     local branch_suffix="$2"
     local branch_name="${feature_num}-${branch_suffix}"
 
-    if [ ${#branch_name} -gt $MAX_BRANCH_LENGTH ]; then
+    if [ "$(branch_byte_count "$branch_name")" -gt "$MAX_BRANCH_LENGTH" ]; then
         local prefix_length=$(( ${#feature_num} + 1 ))
         local max_suffix_length=$((MAX_BRANCH_LENGTH - prefix_length))
         local truncated_suffix
-        truncated_suffix=$(printf '%s' "$branch_suffix" | cut -c "1-$max_suffix_length" | sed 's/-$//')
+        local -x LC_ALL="$UNICODE_LOCALE"
+        local low=0 high=${#branch_suffix} mid
+        if (( high > max_suffix_length )); then
+            high=$max_suffix_length
+        fi
+        while (( low < high )); do
+            mid=$(((low + high + 1) / 2))
+            if [ "$(branch_byte_count "${branch_suffix:0:$mid}")" -le "$max_suffix_length" ]; then
+                low=$mid
+            else
+                high=$((mid - 1))
+            fi
+        done
+        truncated_suffix="${branch_suffix:0:$low}"
+        truncated_suffix="${truncated_suffix%-}"
         branch_name="${feature_num}-${truncated_suffix}"
     fi
 
@@ -209,27 +288,25 @@ generate_branch_name() {
     # Common stop words to filter out
     local stop_words="^(i|a|an|the|to|for|of|in|on|at|by|with|from|is|are|was|were|be|been|being|have|has|had|do|does|did|will|would|should|could|can|may|might|must|shall|this|that|these|those|my|your|our|their|want|need|add|get|set)$"
 
-    # Convert to lowercase and split into words. LC_ALL=C for the same
-    # collation reason documented on clean_branch_name, and so the `grep -qw`
-    # acronym probe below uses ASCII word boundaries like the Python twin's
-    # (?<![0-9A-Za-z_]) lookarounds.
-    local -x LC_ALL=C
-    local clean_name=$(printf '%s' "$description" | tr '[:upper:]' '[:lower:]' | sed 's/[^a-z0-9]/ /g')
+    # Use a UTF-8 locale for character-safe length checks and split words.
+    local -x LC_ALL="$UNICODE_LOCALE"
+    local clean_name
+    clean_name=$(unicode_words "$description" ' ') || return 1
 
     # Filter words: remove stop words and words shorter than 3 chars (unless they're uppercase acronyms in original)
     local meaningful_words=()
     for word in $clean_name; do
         # Skip empty words
         [ -z "$word" ] && continue
 
-        # Keep words that are NOT stop words AND (length >= 3 OR are potential acronyms)
-        if ! echo "$word" | grep -qiE "$stop_words"; then
-            if [ ${#word} -ge 3 ]; then
+        # Retain non-ASCII words even when shorter than three characters.
+        if ! printf '%s\n' "$word" | LC_ALL=C grep -qE "$stop_words"; then
+            if [ ${#word} -ge 3 ] || printf '%s' "$word" | contains_non_ascii; then
                 meaningful_words+=("$word")
             # Keep short words that appear as an uppercase acronym in the original.
             # Uppercase via tr and match with grep -w (both portable) rather than
             # bash's 4+ "^^" case expansion (breaks on macOS bash 3.2) and \b (non-POSIX).
-            elif printf '%s' "$description" | grep -qw -- "$(printf '%s' "$word" | tr '[:lower:]' '[:upper:]')"; then
+            elif printf '%s' "$description" | LC_ALL=C grep -qw -- "$(printf '%s' "$word" | LC_ALL=C tr '[:lower:]' '[:upper:]')"; then
                 meaningful_words+=("$word")
             fi
         fi
@@ -266,7 +343,7 @@ else
 fi
 
 if [ -z "$BRANCH_SUFFIX" ]; then
-    echo "[specify] Warning: Feature name is empty after removing unsupported characters. Use --short-name with ASCII letters or digits (for example, user-auth)." >&2
+    echo "[specify] Warning: Feature name is empty after removing unsupported characters. Use --short-name with letters or digits (for example, user-auth)." >&2
 fi
 
 # Warn if --number and --timestamp are both specified
@@ -339,8 +416,8 @@ ORIGINAL_BRANCH_NAME="${FEATURE_NUM}-${BRANCH_SUFFIX}"
 BRANCH_NAME=$(fit_branch_name "$FEATURE_NUM" "$BRANCH_SUFFIX")
 if [ "$BRANCH_NAME" != "$ORIGINAL_BRANCH_NAME" ]; then
     >&2 echo "[specify] Warning: Branch name exceeded GitHub's 244-byte limit"
-    >&2 echo "[specify] Original: $ORIGINAL_BRANCH_NAME (${#ORIGINAL_BRANCH_NAME} bytes)"
-    >&2 echo "[specify] Truncated to: $BRANCH_NAME (${#BRANCH_NAME} bytes)"
+    >&2 echo "[specify] Original: $ORIGINAL_BRANCH_NAME ($(branch_byte_count "$ORIGINAL_BRANCH_NAME") bytes)"
+    >&2 echo "[specify] Truncated to: $BRANCH_NAME ($(branch_byte_count "$BRANCH_NAME") bytes)"
 fi
 
 FEATURE_DIR="$SPECS_DIR/$BRANCH_NAME"
 
@@ -83,10 +83,32 @@ function Test-SpecPrefixInUse {
         Select-Object -First 1)
 }
 
+function ConvertTo-AsciiLower {
+    param([string]$Name)
+
+    return [regex]::Replace($Name, '[A-Z]', { param($match) $match.Value.ToLowerInvariant() })
+}
+
+function ConvertTo-UnicodeWords {
+    param([string]$Name, [string]$Separator)
+
+    $lowerName = ConvertTo-AsciiLower -Name $Name
+    return [regex]::Replace($lowerName, '[\uD800-\uDBFF][\uDC00-\uDFFF]|[^\p{L}\p{Nd}]', {
+        param($match)
+        if ($match.Length -eq 2) {
+            $category = [System.Globalization.CharUnicodeInfo]::GetUnicodeCategory($match.Value, 0)
+            if ($category.ToString() -match '(Letter|DecimalDigitNumber)$') {
+                return $match.Value
+            }
+        }
+        return $Separator
+    })
+}
+
 function ConvertTo-CleanBranchName {
     param([string]$Name)
 
-    return $Name.ToLower() -replace '[^a-z0-9]', '-' -replace '-{2,}', '-' -replace '^-', '' -replace '-$', ''
+    return (ConvertTo-UnicodeWords -Name $Name -Separator '-') -replace '-{2,}', '-' -replace '^-', '' -replace '-$', ''
 }
 
 function Get-FittedBranchName {
@@ -96,10 +118,15 @@ function Get-FittedBranchName {
     )
 
     $fittedName = "$FeatureNum-$BranchSuffix"
-    if ($fittedName.Length -gt $maxBranchLength) {
+    if ([System.Text.Encoding]::UTF8.GetByteCount($fittedName) -gt $maxBranchLength) {
         $prefixLength = $FeatureNum.Length + 1
         $maxSuffixLength = $maxBranchLength - $prefixLength
-        $truncatedSuffix = $BranchSuffix.Substring(0, [Math]::Min($BranchSuffix.Length, $maxSuffixLength))
+        $bytes = [System.Text.Encoding]::UTF8.GetBytes($BranchSuffix)
+        $bytesToUse = $maxSuffixLength
+        while ($bytesToUse -gt 0 -and ($bytes[$bytesToUse] -band 0xC0) -eq 0x80) {
+            $bytesToUse--
+        }
+        $truncatedSuffix = [System.Text.Encoding]::UTF8.GetString($bytes, 0, $bytesToUse)
         $truncatedSuffix = $truncatedSuffix -replace '-$', ''
         $fittedName = "$FeatureNum-$truncatedSuffix"
     }
@@ -132,18 +159,19 @@ function Get-BranchName {
         'want', 'need', 'add', 'get', 'set'
     )
 
-    # Convert to lowercase and extract words (alphanumeric only)
-    $cleanName = $Description.ToLower() -replace '[^a-z0-9\s]', ' '
+    # Lowercase ASCII and extract Unicode words, matching the shell variant.
+    $cleanName = ConvertTo-UnicodeWords -Name $Description -Separator ' '
     $words = $cleanName -split '\s+' | Where-Object { $_ }
 
     # Filter words: remove stop words and words shorter than 3 chars (unless they're uppercase acronyms in original)
     $meaningfulWords = @()
     foreach ($word in $words) {
         # Skip stop words
-        if ($stopWords -contains $word) { continue }
+        if ($stopWords -ccontains $word) { continue }
 
-        # Keep words that are length >= 3 OR appear as uppercase in original (likely acronyms)
-        if ($word.Length -ge 3) {
+        # Keep Unicode words even when short; ASCII words still need three
+        # characters or an uppercase acronym in the original.
+        if ($word.Length -ge 3 -or $word -match '[^\x00-\x7F]') {
             $meaningfulWords += $word
         } elseif ($Description -cmatch "(?<![0-9A-Za-z_])$($word.ToUpper())(?![0-9A-Za-z_])") {
             # Keep short words only if they appear as uppercase in original (likely
@@ -164,13 +192,7 @@ function Get-BranchName {
     } else {
         # Fallback to original logic if no meaningful words found
         $result = ConvertTo-CleanBranchName -Name $Description
-        # @() keeps this an array. ConvertTo-CleanBranchName blanks every
-        # non-[a-z0-9] character, so a description written in a non-Latin script
-        # (or made only of punctuation) leaves nothing for the pipeline to
-        # emit -- it yields $null, and [string]::Join on $null throws
-        # ArgumentNullException. With $ErrorActionPreference = 'Stop' that is
-        # terminating, so the script died with a .NET stack trace and exit 1
-        # where the bash and Python twins both return an empty suffix.
+        # @() keeps this an array when the description contains only separators.
         $fallbackWords = @(($result -split '-') | Where-Object { $_ } | Select-Object -First 3)
         return [string]::Join('-', $fallbackWords)
     }
@@ -186,7 +208,7 @@ if ($ShortName) {
 }
 
 if (-not $branchSuffix) {
-    [Console]::Error.WriteLine("[specify] Warning: Feature name is empty after removing unsupported characters. Use -ShortName with ASCII letters or digits (for example, user-auth).")
+    [Console]::Error.WriteLine("[specify] Warning: Feature name is empty after removing unsupported characters. Use -ShortName with letters or digits (for example, user-auth).")
 }
 
 # Treat an explicit empty string as omitted, matching the bash and Python twins.
@@ -258,8 +280,8 @@ $originalBranchName = "$featureNum-$branchSuffix"
 $branchName = Get-FittedBranchName -FeatureNum $featureNum -BranchSuffix $branchSuffix
 if ($branchName -ne $originalBranchName) {
     [Console]::Error.WriteLine("[specify] Warning: Branch name exceeded GitHub's 244-byte limit")
-    [Console]::Error.WriteLine("[specify] Original: $originalBranchName ($($originalBranchName.Length) bytes)")
-    [Console]::Error.WriteLine("[specify] Truncated to: $branchName ($($branchName.Length) bytes)")
+    [Console]::Error.WriteLine("[specify] Original: $originalBranchName ($([System.Text.Encoding]::UTF8.GetByteCount($originalBranchName)) bytes)")
+    [Console]::Error.WriteLine("[specify] Truncated to: $branchName ($([System.Text.Encoding]::UTF8.GetByteCount($branchName)) bytes)")
 }
 
 $featureDir = Join-Path $specsDir $branchName