diff --git a/nodes/text.py b/nodes/text.py
index 47a60c5..42e8a52 100644
--- a/nodes/text.py
+++ b/nodes/text.py
@@ -1,3 +1,7 @@
+import re
+import random
+id_pattern = re.compile(r'^[a-zA-Z0-9 ]+$')
+
 class TextMultiline:
     @classmethod
     def INPUT_TYPES(s):
@@ -18,7 +22,6 @@ class TextMultiline:
             text = text.replace("\n", "")
         return (text,)
 
-
 class TextConcatMulti:
     @classmethod
     def INPUT_TYPES(s):
@@ -47,3 +50,143 @@ class TextConcatMulti:
                 text = text.replace("\n", "")
             texts.append(str(text))
         return (separator.join(texts),)
+
+class PromptReplacer:
+    """
+    A node that replaces search_word variants in input_text with random picks from input_list.
+    Automatically detects numbered variants (wordx, wordx2, wordx3...) in the text.
+    In the input list, * items go to wordx, ** items go to wordx2, *** items go to wordx3, etc.
+    Unprefixed items are the shared pool for all variants.
+    Supports block comments (//) and single-line comments (#).
+    """
+    @classmethod
+    def INPUT_TYPES(cls):
+        return {
+            "required": {
+                "input_text": ("STRING", {"default": '', "multiline": False}),
+                "input_list": ("STRING", {"default": '', "multiline": True}),
+                "search_word": ("STRING", {"default": 'wordx'}),
+                "seed": ("INT", {"default": 1337, "min": 0, "max": 0xffffffffffffffff}),
+            }
+        }
+
+    RETURN_TYPES = ("STRING", "STRING")
+    RETURN_NAMES = ("prompt", "selected_item")
+    FUNCTION = "execute"
+    OUTPUT_NODE = True
+    CATEGORY = "sa_nodes/text"
+
+    def parse_input(self, input_list):
+        entries = input_list.strip().split('\n') if input_list.strip() else ['']
+        # Keys: 'shared' for no-star items, and 1/2/3... for * / ** / *** items
+        variant_pools = {}
+        in_comment_block = False
+
+        for entry in entries:
+            stripped_entry = entry.strip()
+
+            if stripped_entry == '//':
+                if in_comment_block:
+                    in_comment_block = False
+                else:
+                    in_comment_block = True
+                continue
+
+            if in_comment_block or stripped_entry == '' or stripped_entry.startswith('#'):
+                continue
+
+            # Count leading stars to determine variant assignment
+            star_count = 0
+            content = stripped_entry
+            while content.startswith('*'):
+                star_count += 1
+                content = content[1:].strip()
+            if star_count == 0:
+                pool_key = 'shared'
+            else:
+                # 1 star = wordx (variant 0), 2 stars = wordx2 (variant 2), etc.
+                pool_key = 0 if star_count == 1 else star_count
+
+            # Parse id@description
+            parts = content.split('@', 1)
+            if len(parts) == 1:
+                item_id = None
+                description = parts[0].strip() or None
+            else:
+                item_id = parts[0].strip()
+                assert id_pattern.match(item_id), f'IDs can only contain alphanumeric characters (a-z, A-Z, 0-9). Offending id is {item_id}'
+                description = parts[1].strip() or None
+
+            if not description:
+                continue
+
+            variant_pools.setdefault(pool_key, []).append((item_id, description))
+
+        return variant_pools
+
+    def find_variants_in_text(self, text, base_word):
+        """Find all numbered variants of base_word present in text (e.g. wordx, wordx2, wordx3)."""
+        variant_nums = []
+        # Match numbered variants first (wordx2, wordx3, etc.)
+        for match in re.finditer(re.escape(base_word) + r'(\d+)', text):
+            variant_nums.append(int(match.group(1)))
+        # Check for standalone base_word (wordx) — not part of a numbered variant
+        num_pattern = re.escape(base_word) + r'\d'
+        if base_word in text and not re.search(num_pattern, text):
+            variant_nums.append(0)
+        # If text has both wordx and wordx2, check for wordx not followed by a digit
+        if re.search(num_pattern, text):
+            # Also check for standalone base_word using word boundary
+            if re.search(re.escape(base_word) + r'(?!\d)', text):
+                variant_nums.append(0)
+        return sorted(set(variant_nums))
+
+    def execute(self, input_list, input_text, search_word, seed):
+        random.seed(seed)
+
+        variant_pools = self.parse_input(input_list)
+        shared_pool = variant_pools.get('shared', [])
+        all_dedicated = any(k != 'shared' for k in variant_pools)
+        if not shared_pool and not all_dedicated:
+            raise ValueError("Input list contains no valid entries to pick from.")
+
+        # Detect which numbered variants exist in the text
+        variant_nums = self.find_variants_in_text(input_text, search_word)
+        if not variant_nums:
+            return (input_text, '')
+
+        # For each variant, pick a random item: prefer dedicated pool, fall back to shared
+        replacements = {}  # variant_num -> (description, id)
+        shared_available = list(shared_pool)
+        for num in variant_nums:
+            dedicated = variant_pools.get(num, [])
+            if dedicated:
+                pick = random.choice(dedicated)
+                replacements[num] = pick
+            elif shared_available:
+                pick = random.choice(shared_available)
+                shared_available.remove(pick)
+                replacements[num] = pick
+            else:
+                raise ValueError(f"No items available for variant {search_word}{num if num else ''}.")
+
+        # Build selected_item string and filename
+        selected_parts = []
+        filename_parts = []
+        for num in sorted(replacements):
+            item_id, desc = replacements[num]
+            selected_parts.append(desc)
+            filename_parts.append(item_id or f'no_id_{num}')
+        selected_item = ", ".join(selected_parts)
+        filename = '-'.join(filename_parts)
+
+        # Replace in text: process longer variants first to avoid overlap (wordx2 before wordx)
+        result_text = input_text
+        for num in sorted(variant_nums, reverse=True):
+            _, desc = replacements[num]
+            if num == 0:
+                result_text = result_text.replace(search_word, desc)
+            else:
+                result_text = result_text.replace(search_word + str(num), desc)
+
+        return (result_text, selected_item)