diff --git a/_lldbtest/main.py b/_lldbtest/main.py
new file mode 100644
index 00000000..64fe68ed
--- /dev/null
+++ b/_lldbtest/main.py
@@ -0,0 +1,363 @@
+import lldb
+import io
+import os
+import sys
+import argparse
+from dataclasses import dataclass, field
+from typing import List
+import cmd
+
+sys.stdout = io.TextIOWrapper(sys.stdout.buffer, line_buffering=False)
+
+def print_to_lldb(*args, **kwargs):
+    print(*args, **kwargs, flush=True)
+
+@dataclass
+class Test:
+    source_file: str
+    line_number: int
+    variable: str
+    expected_value: str
+
+@dataclass
+class TestResult:
+    test: Test
+    status: str
+    actual: str = None
+    message: str = None
+    missing: set = None
+    extra: set = None
+
+@dataclass
+class TestCase:
+    source_file: str
+    start_line: int
+    end_line: int
+    tests: List[Test]
+
+@dataclass
+class CaseResult:
+    test_case: TestCase
+    function: str
+    results: List[TestResult]
+
+@dataclass
+class TestResults:
+    total: int = 0
+    passed: int = 0
+    failed: int = 0
+    case_results: List[CaseResult] = field(default_factory=list)
+
+class LLDBDebugger:
+    def __init__(self, executable_path, plugin_path=None):
+        self.executable_path = executable_path
+        self.plugin_path = plugin_path
+        self.debugger = lldb.SBDebugger.Create()
+        self.debugger.SetAsync(False)
+        self.target = None
+        self.process = None
+
+    def setup(self):
+        if self.plugin_path:
+            self.debugger.HandleCommand(f'command script import "{self.plugin_path}"')
+
+        self.target = self.debugger.CreateTarget(self.executable_path)
+        if not self.target:
+            raise Exception(f"Failed to create target for {self.executable_path}")
+
+    def set_breakpoint(self, file_spec, line_number):
+        breakpoint = self.target.BreakpointCreateByLocation(file_spec, line_number)
+        if not breakpoint.IsValid():
+            raise Exception(f"Failed to set breakpoint at {file_spec}:{line_number}")
+        return breakpoint
+
+    def run_to_breakpoint(self):
+        if not self.process:
+            self.process = self.target.LaunchSimple(None, None, os.getcwd())
+        else:
+            self.process.Continue()
+
+        if self.process.GetState() != lldb.eStateStopped:
+            raise Exception("Process didn't stop at breakpoint")
+
+    def get_variable_value(self, var_name):
+        frame = self.process.GetSelectedThread().GetFrameAtIndex(0)
+        var = frame.FindVariable(var_name)
+        if var.IsValid():
+            type_name = var.GetTypeName()
+            if var.GetNumChildren() > 0:
+                children = []
+                for i in range(var.GetNumChildren()):
+                    child = var.GetChildAtIndex(i)
+                    child_name = child.GetName()
+                    child_value = child.GetValue()
+                    children.append(f"{child_name} = {child_value}")
+                return f"{type_name}({', '.join(children)})"
+            else:
+                value = var.GetValue() or var.GetSummary() or str(var)
+                return str(value) if value is not None else "None"
+        return None
+
+    def get_all_variable_names(self):
+        frame = self.process.GetSelectedThread().GetFrameAtIndex(0)
+        return set(var.GetName() for var in frame.GetVariables(True, True, True, False))
+
+    def get_current_function_name(self):
+        frame = self.process.GetSelectedThread().GetFrameAtIndex(0)
+        return frame.GetFunctionName()
+
+    def cleanup(self):
+        if self.process and self.process.IsValid():
+            self.process.Kill()
+        lldb.SBDebugger.Destroy(self.debugger)
+
+    def run_console(self):
+        print_to_lldb("\nEntering LLDB interactive mode. Type 'quit' to exit and continue with the next test case.")
+
+        old_stdin, old_stdout, old_stderr = sys.stdin, sys.stdout, sys.stderr
+        sys.stdin, sys.stdout, sys.stderr = sys.__stdin__, sys.__stdout__, sys.__stderr__
+
+        self.debugger.SetAsync(True)
+        self.debugger.HandleCommand("settings set auto-confirm true")
+        self.debugger.HandleCommand("command script import lldb")
+
+        interpreter = self.debugger.GetCommandInterpreter()
+        while True:
+            print_to_lldb("\n(lldb) ", end="")
+            command = input().strip()
+            if command.lower() == 'quit':
+                break
+
+            result = lldb.SBCommandReturnObject()
+            interpreter.HandleCommand(command, result)
+            if result.Succeeded():
+                print_to_lldb(result.GetOutput().rstrip())
+            else:
+                print_to_lldb(result.GetError().rstrip())
+
+        sys.stdin, sys.stdout, sys.stderr = old_stdin, old_stdout, old_stderr
+        print_to_lldb("\nExiting LLDB interactive mode.")
+
+
+def parse_expected_values(source_files):
+    test_cases = []
+    for source_file in source_files:
+        with open(source_file, 'r') as f:
+            content = f.readlines()
+            i = 0
+            while i < len(content):
+                line = content[i].strip()
+                if line.startswith('// Expected:'):
+                    start_line = i + 1
+                    tests = []
+                    i += 1
+                    while i < len(content):
+                        line = content[i].strip()
+                        if not line.startswith('//'):
+                            break
+                        parts = line.lstrip('//').strip().split(':', 1)
+                        if len(parts) == 2:
+                            var, value = map(str.strip, parts)
+                            tests.append(Test(source_file, i + 1, var, value))
+                        i += 1
+                    end_line = i
+                    test_cases.append(
+                        TestCase(source_file, start_line, end_line, tests))
+                else:
+                    i += 1
+    return test_cases
+
+
+def run_tests(executable_path, source_files, verbose, interactive, plugin_path):
+    debugger = LLDBDebugger(executable_path, plugin_path)
+    test_cases = parse_expected_values(source_files)
+    if verbose:
+        print_to_lldb(
+            f"Running tests for {', '.join(source_files)} with {executable_path}")
+        print_to_lldb(f"Found {len(test_cases)} test cases")
+
+    try:
+        debugger.setup()
+        results = execute_tests(debugger, test_cases, interactive)
+        print_test_results(results, verbose)
+
+        if results.total != results.passed:
+            os._exit(1)
+
+    except Exception as e:
+        print_to_lldb(f"Error: {str(e)}")
+
+    finally:
+        debugger.cleanup()
+
+
+def execute_tests(debugger, test_cases, interactive):
+    results = TestResults()
+
+    for test_case in test_cases:
+        breakpoint = debugger.set_breakpoint(
+            test_case.source_file, test_case.end_line)
+        debugger.run_to_breakpoint()
+
+        function_name = debugger.get_current_function_name()
+        all_variable_names = debugger.get_all_variable_names()
+
+        case_result = execute_test_case(
+            debugger, test_case, all_variable_names)
+
+        results.total += len(case_result.results)
+        results.passed += sum(1 for r in case_result.results if r.status == 'pass')
+        results.failed += sum(1 for r in case_result.results if r.status != 'pass')
+        results.case_results.append(case_result)
+
+        # Print the current case test results before entering interactive mode
+        print_to_lldb(f"\nTest case: {case_result.test_case.source_file}:{case_result.test_case.start_line}-{case_result.test_case.end_line} in function '{case_result.function}'")
+        for result in case_result.results:
+            # Always print verbose results here
+            print_test_result(result, True)
+
+        if interactive and any(r.status != 'pass' for r in case_result.results):
+            print_to_lldb("\nTest case failed. Entering LLDB interactive mode.")
+            debugger.run_console()
+            # After exiting the console, we need to ensure the process is in a valid state
+            if debugger.process.GetState() == lldb.eStateRunning:
+                debugger.process.Stop()
+            elif debugger.process.GetState() == lldb.eStateExited:
+                # If the process has exited, we need to re-launch it
+                debugger.process = debugger.target.LaunchSimple(
+                    None, None, os.getcwd())
+
+        debugger.target.BreakpointDelete(breakpoint.GetID())
+
+    return results
+
+
+def execute_test_case(debugger, test_case, all_variable_names):
+    results = []
+
+    for test in test_case.tests:
+        if test.variable == "all variables":
+            result = execute_all_variables_test(test, all_variable_names)
+        else:
+            result = execute_single_variable_test(debugger, test)
+        results.append(result)
+
+    return CaseResult(test_case, debugger.get_current_function_name(), results)
+
+
+def execute_all_variables_test(test, all_variable_names):
+    expected_vars = set(test.expected_value.split())
+    if expected_vars == all_variable_names:
+        return TestResult(
+            test=test,
+            status='pass',
+            actual=all_variable_names
+        )
+    else:
+        return TestResult(
+            test=test,
+            status='fail',
+            actual=all_variable_names,
+            missing=expected_vars - all_variable_names,
+            extra=all_variable_names - expected_vars
+        )
+
+
+def execute_single_variable_test(debugger, test):
+    actual_value = debugger.get_variable_value(test.variable)
+    if actual_value is None:
+        print_to_lldb(f"Unable to fetch value for {test.variable}")
+        return TestResult(
+            test=test,
+            status='error',
+            message='Unable to fetch value'
+        )
+
+    # 移除可能的空格，但保留括号
+    actual_value = actual_value.strip()
+    expected_value = test.expected_value.strip()
+
+    # 比较处理后的值
+    if actual_value == expected_value:
+        return TestResult(
+            test=test,
+            status='pass',
+            actual=actual_value
+        )
+    else:
+        return TestResult(
+            test=test,
+            status='fail',
+            actual=actual_value
+        )
+
+
+def print_test_results(results: TestResults, verbose):
+    for case_result in results.case_results:
+        print_to_lldb(f"\nTest case: {case_result.test_case.source_file}:{case_result.test_case.start_line}-{case_result.test_case.end_line} in function '{case_result.function}'")
+        for result in case_result.results:
+            print_test_result(result, verbose)
+
+    print_to_lldb("\nTest results:")
+    print_to_lldb(f"  Total tests: {results.total}")
+    print_to_lldb(f"  Passed tests: {results.passed}")
+    print_to_lldb(f"  Failed tests: {results.failed}")
+    if results.total == results.passed:
+        print_to_lldb("All tests passed!")
+    else:
+        print_to_lldb("Some tests failed")
+
+
+def print_test_result(result: TestResult, verbose):
+    status_symbol = "✓" if result.status == 'pass' else "✗"
+    status_text = "Pass" if result.status == 'pass' else "Fail"
+
+    if result.status == 'pass':
+        if verbose:
+            print_to_lldb(
+                f"{status_symbol} Line {result.test.line_number}, {result.test.variable}: {status_text}")
+            if result.test.variable == 'all variables':
+                print_to_lldb(f"    Variables: {', '.join(sorted(result.actual))}")
+    else:  # fail or error
+        print_to_lldb(
+            f"{status_symbol} Line {result.test.line_number}, {result.test.variable}: {status_text}")
+        if result.test.variable == 'all variables':
+            if result.missing:
+                print_to_lldb(
+                    f"    Missing variables: {', '.join(sorted(result.missing))}")
+            if result.extra:
+                print_to_lldb(
+                    f"    Extra variables: {', '.join(sorted(result.extra))}")
+            print_to_lldb(
+                f"    Expected: {', '.join(sorted(result.test.expected_value.split()))}")
+            print_to_lldb(f"    Actual: {', '.join(sorted(result.actual))}")
+        elif result.status == 'error':
+            print_to_lldb(f"    Error: {result.message}")
+        else:
+            print_to_lldb(f"    Expected: {result.test.expected_value}")
+            print_to_lldb(f"    Actual: {result.actual}")
+
+def main():
+    print_to_lldb(sys.argv)
+    parser = argparse.ArgumentParser(description="LLDB 18 Debug Script with DWARF 5 Support")
+    parser.add_argument("executable", help="Path to the executable")
+    parser.add_argument("sources", nargs='+', help="Paths to the source files")
+    parser.add_argument("-v", "--verbose", action="store_true", help="Enable verbose output")
+    parser.add_argument("-i", "--interactive", action="store_true", help="Enable interactive mode on test failure")
+    parser.add_argument("--plugin", help="Path to the LLDB plugin")
+    args = parser.parse_args()
+
+    plugin_path = args.plugin or os.path.join(os.path.dirname(os.path.realpath(__file__)), "go_lldb_plugin.py")
+    run_tests(args.executable, args.sources, args.verbose, args.interactive, plugin_path)
+
+if __name__ == "__main__":
+    main()
+
+def run_commands(debugger, command, result, internal_dict):
+    print_to_lldb(sys.argv)
+    main()
+    debugger.HandleCommand("quit")
+
+def __lldb_init_module(debugger, internal_dict):
+    # debugger.HandleCommand('command script add -f main.run_commands run_tests')
+    pass
diff --git a/_lldbtest/runmain.lldb b/_lldbtest/runmain.lldb
new file mode 100644
index 00000000..a75456e3
--- /dev/null
+++ b/_lldbtest/runmain.lldb
@@ -0,0 +1,9 @@
+# LLDB can't be load from Python 3.12, should load Python script by lldb command
+# See https://github.com/llvm/llvm-project/issues/70453
+
+# go run ./cmd/llgo build -o cl/_testdata/debug/out -dbg ./cl/_testdata/debug
+# lldb -S _lldbtest/runmain.lldb
+
+command script import _lldbtest/main.py
+script main.run_tests("cl/_testdata/debug/out", ["cl/_testdata/debug/in.go"], True, True, None)
+quit