diff --git a/scripts/surgical_fix.py b/scripts/surgical_fix.py
new file mode 100644
index 0000000..faab74b
--- /dev/null
+++ b/scripts/surgical_fix.py
@@ -0,0 +1,114 @@
+#!/usr/bin/env python3
+"""
+Surgical fix using exact string patterns found in the file.
+"""
+
+import sys
+
+def fix_align_patterns(content):
+    """Fix known problematic align patterns."""
+
+    transformations = 0
+
+    # List of exact patterns to replace
+    patterns = [
+        # Pattern 1: Simple case with line break in middle
+        (
+            '<align pos="right">gehorsamsten Diener\n<line type="break" />Jacob Michael Reinhold Lenz</align>',
+            '<align pos="right">gehorsamsten Diener</align>\n<line type="break" /><align pos="right">Jacob Michael Reinhold Lenz</align>'
+        ),
+
+        # Pattern 2: Multi-line signature block
+        (
+            '<align pos="right">Hoch Edelgeborner Hochgelahrter Herr <aq>Secretair</aq>\n<line type="break" tab="7" />Verehrungswürdigster Gönner\n<line type="break" tab="7" />Ew. HochEdelgebh:</align>',
+            '<align pos="right">Hoch Edelgeborner Hochgelahrter Herr <aq>Secretair</aq></align>\n<line type="break" tab="7" /><align pos="right">Verehrungswürdigster Gönner</align>\n<line type="break" tab="7" /><align pos="right">Ew. HochEdelgebh:</align>'
+        ),
+
+        # Pattern 3: With <aq> tag containing line break
+        (
+            '<align pos="center">\n  <aq>Interfusa nitentes\n<line type="break"/>Vites aequora Cycladas.</aq>\n</align>',
+            '<align pos="center">\n  <aq>Interfusa nitentes</aq></align>\n<line type="break"/><align pos="center"><aq>Vites aequora Cycladas.</aq>\n</align>'
+        ),
+
+        # Pattern 4: Signature with aq
+        (
+            '<align pos="right">Sie ewig liebender <aq>Alcibiades</aq>\n<line type="break"/>J. M. R. L.</align>',
+            '<align pos="right">Sie ewig liebender <aq>Alcibiades</aq></align>\n<line type="break"/><align pos="right">J. M. R. L.</align>'
+        ),
+    ]
+
+    for old_pattern, new_pattern in patterns:
+        if old_pattern in content:
+            content = content.replace(old_pattern, new_pattern)
+            transformations += 1
+            print(f"Applied transformation {transformations}")
+
+    # Now handle the more complex multi-line patterns using targeted replacements
+    # Let's find and fix the complex ones one by one
+
+    # Find patterns like "text<line/>text" within align tags
+    import re
+
+    # Pattern for align elements containing line breaks
+    align_pattern = r'<align pos="([^"]+)"[^>]*>(.*?)</align>'
+
+    def fix_align_content(match):
+        pos = match.group(1)
+        content_part = match.group(2)
+
+        # Check if this content contains line elements
+        if '<line' not in content_part:
+            return match.group(0)  # No change needed
+
+        # Split on line elements, preserving the line elements
+        parts = re.split(r'(<line[^>]*(?:/>|></line>))', content_part)
+
+        result = []
+        current_text = ""
+
+        for part in parts:
+            if part.startswith('<line'):
+                # This is a line element
+                if current_text.strip():
+                    result.append(f'<align pos="{pos}">{current_text}</align>')
+                    current_text = ""
+                result.append(part)
+            else:
+                current_text += part
+
+        # Add remaining text
+        if current_text.strip():
+            result.append(f'<align pos="{pos}">{current_text}</align>')
+
+        return ''.join(result)
+
+    # Apply the pattern replacement
+    new_content = re.sub(align_pattern, fix_align_content, content, flags=re.DOTALL)
+
+    if new_content != content:
+        additional_transforms = len(re.findall(r'</align>\s*<line', new_content)) - len(re.findall(r'</align>\s*<line', content))
+        transformations += additional_transforms
+        print(f"Applied {additional_transforms} additional regex transformations")
+
+    print(f"Total transformations: {transformations}")
+    return new_content
+
+def main():
+    if len(sys.argv) != 2:
+        print("Usage: python surgical_fix.py FILE.xml")
+        sys.exit(1)
+
+    file_path = sys.argv[1]
+
+    with open(file_path, 'r', encoding='utf-8') as f:
+        content = f.read()
+
+    transformed_content = fix_align_patterns(content)
+
+    with open(file_path, 'w', encoding='utf-8') as f:
+        f.write(transformed_content)
+
+    print("Surgical fix complete!")
+
+if __name__ == "__main__":
+    main()
\ No newline at end of file