diff --git a/gen-wcwidth.py b/gen-wcwidth.py
index 1b632b5a5..aeeab712c 100755
--- a/gen-wcwidth.py
+++ b/gen-wcwidth.py
@@ -112,6 +112,10 @@ def parse_ucd() -> None:
             elif category.startswith('S'):
                 all_symbols.add(codepoint)
             elif category == 'Cf':
+                # we add Cf to marks as it contains things like tags and zero
+                # width chars. Not sure if *all* of Cf should be treated as
+                # combining chars, might need to add individual exceptions in
+                # the future.
                 marks.add(codepoint)
 
     with open('nerd-fonts-glyphs.txt') as f:
diff --git a/kitty_tests/screen.py b/kitty_tests/screen.py
index 5ed69d2c5..5a638aff8 100644
--- a/kitty_tests/screen.py
+++ b/kitty_tests/screen.py
@@ -127,6 +127,12 @@ class TestScreen(BaseTest):
         s.draw(q)
         self.ae(q, str(s.line(0)))
         self.ae(s.cursor.x, 8)
+        for x in '\u200b\u200c\u200d':
+            s = self.create_screen()
+            q = f'X{x}Y'
+            s.draw(q)
+            self.ae(q, str(s.line(0)))
+            self.ae(s.cursor.x, 2)
 
     def test_char_manipulation(self):
         s = self.create_screen()