fix: use utf-8 for harness codex prompts

scripts/execute.py

@@ -375,7 +375,8 @@ class StepExecutor:
         prompt = preamble + step_file.read_text(encoding="utf-8")
         result = subprocess.run(
             [self._codex_command(), "exec", "--dangerously-bypass-approvals-and-sandbox", "--json", "-"],
-            cwd=self._root, capture_output=True, text=True, input=prompt, timeout=1800,
+            cwd=self._root, capture_output=True, text=True, input=prompt,
+            encoding="utf-8", errors="replace", timeout=1800,
         )
 
         if result.returncode != 0:

scripts/test_execute.py

@@ -305,6 +305,8 @@ class ExecuteRunnerSafetyTests(unittest.TestCase):
             )
             self.assertEqual(kwargs["input"], long_preamble + "# Step 1\n")
             self.assertEqual(kwargs["cwd"], str(root))
+            self.assertEqual(kwargs["encoding"], "utf-8")
+            self.assertEqual(kwargs["errors"], "replace")
 
     def test_codex_command_prefers_windows_cmd_or_exe_shim(self):
         execute = load_execute()