diff --git a/src/codegen/git/repo_operator/repo_operator.py b/src/codegen/git/repo_operator/repo_operator.py
index 9f1b66ca9..dd00d7e42 100644
--- a/src/codegen/git/repo_operator/repo_operator.py
+++ b/src/codegen/git/repo_operator/repo_operator.py
@@ -1,3 +1,4 @@
+import codecs
 import fnmatch
 import glob
 import logging
@@ -577,6 +578,25 @@ def get_filepaths_for_repo(self, ignore_list):
         if ignore_list:
             filepaths = [f for f in filepaths if not any(fnmatch.fnmatch(f, pattern) or f.startswith(pattern) for pattern in ignore_list)]
 
+        # Fix bug where unicode characters are not handled correctly
+        for i, filepath in enumerate(filepaths):
+            # Check if it is one of the broken cases
+            if filepath.startswith('"'):
+                # Step 1: Strip the quotes
+                filepath = filepath.strip('"').strip("'")
+
+                # Step 2: Convert the Python string to raw ASCII bytes (so \\ stays as two 0x5C).
+                raw_filepath = filepath.encode("ascii")
+
+                # Step 3: Use escape_decode to process backslash escapes like \346 -> 0xE6
+                decoded_filepath, _ = codecs.escape_decode(raw_filepath)
+
+                # Step 4: Decode those bytes as UTF-8 to get the actual Unicode text
+                filepath = decoded_filepath.decode("utf-8")
+
+                # Step 5: Replace the original filepath with the decoded filepath
+                filepaths[i] = filepath
+
         return filepaths
 
     # TODO: unify param naming i.e. subdirectories vs subdirs probably use subdirectories since that's in the DB
diff --git a/tests/unit/codegen/sdk/core/test_directory.py b/tests/unit/codegen/sdk/core/test_directory.py
index 8e8c5fc85..1af8480df 100644
--- a/tests/unit/codegen/sdk/core/test_directory.py
+++ b/tests/unit/codegen/sdk/core/test_directory.py
@@ -7,8 +7,10 @@
 
 from codegen.sdk.codebase.codebase_context import CodebaseContext
 from codegen.sdk.codebase.config import CodebaseConfig
+from codegen.sdk.codebase.factory.get_session import get_codebase_session
 from codegen.sdk.core.directory import Directory
 from codegen.sdk.core.file import File
+from codegen.shared.enums.programming_language import ProgrammingLanguage
 
 
 @pytest.fixture
@@ -220,3 +222,15 @@ def test_get_set_delete_item(mock_directory):
 
     with pytest.raises(KeyError, match="subdir_2"):
         del mock_directory["subdir_2"]
+
+
+def test_unicode_in_filename(tmpdir) -> None:
+    with get_codebase_session(
+        tmpdir=tmpdir,
+        files={"ascii.py": "print('Hello, world!')", "test/我很喜欢冰激淋/test-file 12'3_🍦.py": "print('Hello, world!')"},
+        programming_language=ProgrammingLanguage.PYTHON,
+        verify_output=True,
+    ) as codebase:
+        file = codebase.get_file("test/我很喜欢冰激淋/test-file 12'3_🍦.py")
+        assert file is not None
+        assert file.content == "print('Hello, world!')"