chore: re-trigger CI checks

Co-Authored-By: João <joao@crewai.com>

docs/en/concepts/flows.mdx

@@ -574,10 +574,6 @@ When you run this Flow, the output will change based on the random boolean value
 
 ### Human in the Loop (human feedback)
 
-<Note>
-The `@human_feedback` decorator requires **CrewAI version 1.8.0 or higher**.
-</Note>
-
 The `@human_feedback` decorator enables human-in-the-loop workflows by pausing flow execution to collect feedback from a human. This is useful for approval gates, quality review, and decision points that require human judgment.
 
 ```python Code

docs/en/learn/human-feedback-in-flows.mdx

@@ -7,10 +7,6 @@ mode: "wide"
 
 ## Overview
 
-<Note>
-The `@human_feedback` decorator requires **CrewAI version 1.8.0 or higher**. Make sure to update your installation before using this feature.
-</Note>
-
 The `@human_feedback` decorator enables human-in-the-loop (HITL) workflows directly within CrewAI Flows. It allows you to pause flow execution, present output to a human for review, collect their feedback, and optionally route to different listeners based on the feedback outcome.
 
 This is particularly valuable for:

docs/en/learn/human-in-the-loop.mdx

@@ -11,10 +11,10 @@ Human-in-the-Loop (HITL) is a powerful approach that combines artificial intelli
 
 CrewAI offers two main approaches for implementing human-in-the-loop workflows:
 
-| Approach | Best For | Integration | Version |
-|----------|----------|-------------|---------|
-| **Flow-based** (`@human_feedback` decorator) | Local development, console-based review, synchronous workflows | [Human Feedback in Flows](/en/learn/human-feedback-in-flows) | **1.8.0+** |
-| **Webhook-based** (Enterprise) | Production deployments, async workflows, external integrations (Slack, Teams, etc.) | This guide | - |
+| Approach | Best For | Integration |
+|----------|----------|-------------|
+| **Flow-based** (`@human_feedback` decorator) | Local development, console-based review, synchronous workflows | [Human Feedback in Flows](/en/learn/human-feedback-in-flows) |
+| **Webhook-based** (Enterprise) | Production deployments, async workflows, external integrations (Slack, Teams, etc.) | This guide |
 
 <Tip>
 If you're building flows and want to add human review steps with routing based on feedback, check out the [Human Feedback in Flows](/en/learn/human-feedback-in-flows) guide for the `@human_feedback` decorator.

docs/ko/concepts/flows.mdx

@@ -567,10 +567,6 @@ Fourth method running
 
 ### Human in the Loop (인간 피드백)
 
-<Note>
-`@human_feedback` 데코레이터는 **CrewAI 버전 1.8.0 이상**이 필요합니다.
-</Note>
-
 `@human_feedback` 데코레이터는 인간의 피드백을 수집하기 위해 플로우 실행을 일시 중지하는 human-in-the-loop 워크플로우를 가능하게 합니다. 이는 승인 게이트, 품질 검토, 인간의 판단이 필요한 결정 지점에 유용합니다.
 
 ```python Code

docs/ko/learn/human-feedback-in-flows.mdx

@@ -7,10 +7,6 @@ mode: "wide"
 
 ## 개요
 
-<Note>
-`@human_feedback` 데코레이터는 **CrewAI 버전 1.8.0 이상**이 필요합니다. 이 기능을 사용하기 전에 설치를 업데이트하세요.
-</Note>
-
 `@human_feedback` 데코레이터는 CrewAI Flow 내에서 직접 human-in-the-loop(HITL) 워크플로우를 가능하게 합니다. Flow 실행을 일시 중지하고, 인간에게 검토를 위해 출력을 제시하고, 피드백을 수집하고, 선택적으로 피드백 결과에 따라 다른 리스너로 라우팅할 수 있습니다.
 
 이는 특히 다음과 같은 경우에 유용합니다:

docs/ko/learn/human-in-the-loop.mdx

@@ -5,22 +5,9 @@ icon: "user-check"
 mode: "wide"
 ---
 
-휴먼 인 더 루프(HITL, Human-in-the-Loop)는 인공지능과 인간의 전문 지식을 결합하여 의사결정을 강화하고 작업 결과를 향상시키는 강력한 접근 방식입니다. CrewAI는 필요에 따라 HITL을 구현하는 여러 가지 방법을 제공합니다.
+휴먼 인 더 루프(HITL, Human-in-the-Loop)는 인공지능과 인간의 전문 지식을 결합하여 의사결정을 강화하고 작업 결과를 향상시키는 강력한 접근 방식입니다. 이 가이드에서는 CrewAI 내에서 HITL을 구현하는 방법을 안내합니다.
 
-## HITL 접근 방식 선택
-
-CrewAI는 human-in-the-loop 워크플로우를 구현하기 위한 두 가지 주요 접근 방식을 제공합니다:
-
-| 접근 방식 | 적합한 용도 | 통합 | 버전 |
-|----------|----------|-------------|---------|
-| **Flow 기반** (`@human_feedback` 데코레이터) | 로컬 개발, 콘솔 기반 검토, 동기식 워크플로우 | [Flow에서 인간 피드백](/ko/learn/human-feedback-in-flows) | **1.8.0+** |
-| **Webhook 기반** (Enterprise) | 프로덕션 배포, 비동기 워크플로우, 외부 통합 (Slack, Teams 등) | 이 가이드 | - |
-
-<Tip>
-Flow를 구축하면서 피드백을 기반으로 라우팅하는 인간 검토 단계를 추가하려면 `@human_feedback` 데코레이터에 대한 [Flow에서 인간 피드백](/ko/learn/human-feedback-in-flows) 가이드를 참조하세요.
-</Tip>
-
-## Webhook 기반 HITL 워크플로우 설정
+## HITL 워크플로우 설정
 
 <Steps>
     <Step title="작업 구성">

docs/pt-BR/concepts/flows.mdx

@@ -309,10 +309,6 @@ Ao executar esse Flow, a saída será diferente dependendo do valor booleano ale
 
 ### Human in the Loop (feedback humano)
 
-<Note>
-O decorador `@human_feedback` requer **CrewAI versão 1.8.0 ou superior**.
-</Note>
-
 O decorador `@human_feedback` permite fluxos de trabalho human-in-the-loop, pausando a execução do flow para coletar feedback de um humano. Isso é útil para portões de aprovação, revisão de qualidade e pontos de decisão que requerem julgamento humano.
 
 ```python Code

docs/pt-BR/learn/human-feedback-in-flows.mdx

@@ -7,10 +7,6 @@ mode: "wide"
 
 ## Visão Geral
 
-<Note>
-O decorador `@human_feedback` requer **CrewAI versão 1.8.0 ou superior**. Certifique-se de atualizar sua instalação antes de usar este recurso.
-</Note>
-
 O decorador `@human_feedback` permite fluxos de trabalho human-in-the-loop (HITL) diretamente nos CrewAI Flows. Ele permite pausar a execução do flow, apresentar a saída para um humano revisar, coletar seu feedback e, opcionalmente, rotear para diferentes listeners com base no resultado do feedback.
 
 Isso é particularmente valioso para:

docs/pt-BR/learn/human-in-the-loop.mdx

@@ -5,22 +5,9 @@ icon: "user-check"
 mode: "wide"
 ---
 
-Human-in-the-Loop (HITL) é uma abordagem poderosa que combina a inteligência artificial com a experiência humana para aprimorar a tomada de decisões e melhorar os resultados das tarefas. CrewAI oferece várias maneiras de implementar HITL dependendo das suas necessidades.
+Human-in-the-Loop (HITL) é uma abordagem poderosa que combina a inteligência artificial com a experiência humana para aprimorar a tomada de decisões e melhorar os resultados das tarefas. Este guia mostra como implementar HITL dentro da CrewAI.
 
-## Escolhendo Sua Abordagem HITL
-
-CrewAI oferece duas abordagens principais para implementar workflows human-in-the-loop:
-
-| Abordagem | Melhor Para | Integração | Versão |
-|----------|----------|-------------|---------|
-| **Baseada em Flow** (decorador `@human_feedback`) | Desenvolvimento local, revisão via console, workflows síncronos | [Feedback Humano em Flows](/pt-BR/learn/human-feedback-in-flows) | **1.8.0+** |
-| **Baseada em Webhook** (Enterprise) | Deployments em produção, workflows assíncronos, integrações externas (Slack, Teams, etc.) | Este guia | - |
-
-<Tip>
-Se você está construindo flows e deseja adicionar etapas de revisão humana com roteamento baseado em feedback, confira o guia [Feedback Humano em Flows](/pt-BR/learn/human-feedback-in-flows) para o decorador `@human_feedback`.
-</Tip>
-
-## Configurando Workflows HITL Baseados em Webhook
+## Configurando Workflows HITL
 
 <Steps>
     <Step title="Configure sua Tarefa">

lib/crewai/src/crewai/events/event_listener.py

@@ -209,9 +209,10 @@ class EventListener(BaseEventListener):
         @crewai_event_bus.on(TaskCompletedEvent)
         def on_task_completed(source: Any, event: TaskCompletedEvent) -> None:
             # Handle telemetry
-            span = self.execution_spans.pop(source, None)
+            span = self.execution_spans.get(source)
             if span:
                 self._telemetry.task_ended(span, source, source.agent.crew)
+            self.execution_spans[source] = None
 
             # Pass task name if it exists
             task_name = get_task_name(source)
@@ -221,10 +222,11 @@ class EventListener(BaseEventListener):
 
         @crewai_event_bus.on(TaskFailedEvent)
         def on_task_failed(source: Any, event: TaskFailedEvent) -> None:
-            span = self.execution_spans.pop(source, None)
+            span = self.execution_spans.get(source)
             if span:
                 if source.agent and source.agent.crew:
                     self._telemetry.task_ended(span, source, source.agent.crew)
+                self.execution_spans[source] = None
 
             # Pass task name if it exists
             task_name = get_task_name(source)

lib/crewai/src/crewai/utilities/file_handler.py

@@ -2,8 +2,11 @@ from datetime import datetime
 import json
 import os
 import pickle
+import tempfile
+import threading
 from typing import Any, TypedDict
 
+import portalocker
 from typing_extensions import Unpack
 
 
@@ -123,10 +126,15 @@ class FileHandler:
 
 
 class PickleHandler:
-    """Handler for saving and loading data using pickle.
+    """Thread-safe handler for saving and loading data using pickle.
+
+    This class provides thread-safe file operations using portalocker for
+    cross-process file locking and atomic write operations to prevent
+    data corruption during concurrent access.
 
     Attributes:
         file_path: The path to the pickle file.
+        _lock: Threading lock for thread-safe operations within the same process.
     """
 
     def __init__(self, file_name: str) -> None:
@@ -141,34 +149,62 @@ class PickleHandler:
             file_name += ".pkl"
 
         self.file_path = os.path.join(os.getcwd(), file_name)
+        self._lock = threading.Lock()
 
     def initialize_file(self) -> None:
         """Initialize the file with an empty dictionary and overwrite any existing data."""
         self.save({})
 
     def save(self, data: Any) -> None:
-        """
-        Save the data to the specified file using pickle.
+        """Save the data to the specified file using pickle with thread-safe atomic writes.
+
+        This method uses a two-phase approach for thread safety:
+        1. Threading lock for same-process thread safety
+        2. Atomic write (write to temp file, then rename) for cross-process safety
+           and data integrity
 
         Args:
-          data: The data to be saved to the file.
+            data: The data to be saved to the file.
         """
-        with open(self.file_path, "wb") as f:
-            pickle.dump(obj=data, file=f)
+        with self._lock:
+            dir_name = os.path.dirname(self.file_path) or os.getcwd()
+            fd, temp_path = tempfile.mkstemp(
+                suffix=".pkl.tmp", prefix="pickle_", dir=dir_name
+            )
+            try:
+                with os.fdopen(fd, "wb") as f:
+                    pickle.dump(obj=data, file=f)
+                    f.flush()
+                    os.fsync(f.fileno())
+                os.replace(temp_path, self.file_path)
+            except Exception:
+                if os.path.exists(temp_path):
+                    os.unlink(temp_path)
+                raise
 
     def load(self) -> Any:
-        """Load the data from the specified file using pickle.
+        """Load the data from the specified file using pickle with thread-safe locking.
+
+        This method uses portalocker for cross-process read locking to ensure
+        data consistency when multiple processes may be accessing the file.
 
         Returns:
-            The data loaded from the file.
+            The data loaded from the file, or an empty dictionary if the file
+            does not exist or is empty.
         """
-        if not os.path.exists(self.file_path) or os.path.getsize(self.file_path) == 0:
-            return {}  # Return an empty dictionary if the file does not exist or is empty
+        with self._lock:
+            if (
+                not os.path.exists(self.file_path)
+                or os.path.getsize(self.file_path) == 0
+            ):
+                return {}
 
-        with open(self.file_path, "rb") as file:
-            try:
-                return pickle.load(file)  # noqa: S301
-            except EOFError:
-                return {}  # Return an empty dictionary if the file is empty or corrupted
-            except Exception:
-                raise  # Raise any other exceptions that occur during loading
+            with portalocker.Lock(
+                self.file_path, "rb", flags=portalocker.LOCK_SH
+            ) as file:
+                try:
+                    return pickle.load(file)  # noqa: S301
+                except EOFError:
+                    return {}
+                except Exception:
+                    raise

lib/crewai/tests/events/test_event_listener_execution_spans.py

@@ -1,243 +0,0 @@
-"""Tests for EventListener execution_spans cleanup to prevent memory leaks."""
-
-import asyncio
-from unittest.mock import MagicMock, patch
-
-import pytest
-
-from crewai.events.event_bus import crewai_event_bus
-from crewai.events.event_listener import EventListener
-from crewai.events.types.task_events import (
-    TaskCompletedEvent,
-    TaskFailedEvent,
-    TaskStartedEvent,
-)
-from crewai.tasks.task_output import TaskOutput
-
-
-class MockAgent:
-    """Mock agent for testing."""
-
-    def __init__(self, role: str = "test_role"):
-        self.role = role
-        self.crew = MagicMock()
-
-
-class MockTask:
-    """Mock task for testing."""
-
-    def __init__(self, task_id: str = "test_task"):
-        self.id = task_id
-        self.name = "Test Task"
-        self.description = "A test task description"
-        self.agent = MockAgent()
-
-
-@pytest.fixture
-def event_listener():
-    """Create a fresh EventListener instance for testing."""
-    EventListener._instance = None
-    EventListener._initialized = False
-    listener = EventListener()
-    listener.setup_listeners(crewai_event_bus)
-    return listener
-
-
-@pytest.fixture
-def mock_task():
-    """Create a mock task for testing."""
-    return MockTask()
-
-
-@pytest.fixture
-def mock_task_output():
-    """Create a mock task output for testing."""
-    return TaskOutput(
-        description="Test task description",
-        raw="Test output",
-        agent="test_agent",
-    )
-
-
-@pytest.mark.asyncio
-async def test_execution_spans_removed_on_task_completed(
-    event_listener, mock_task, mock_task_output
-):
-    """Test that execution_spans entries are properly removed when a task completes.
-
-    This test verifies the fix for the memory leak where completed tasks were
-    setting execution_spans[source] = None instead of removing the key entirely.
-    """
-    with patch.object(event_listener._telemetry, "task_started") as mock_task_started:
-        with patch.object(event_listener._telemetry, "task_ended"):
-            mock_span = MagicMock()
-            mock_task_started.return_value = mock_span
-
-            start_event = TaskStartedEvent(context="test context", task=mock_task)
-            future = crewai_event_bus.emit(mock_task, start_event)
-            if future:
-                await asyncio.wrap_future(future)
-
-            assert mock_task in event_listener.execution_spans
-            assert event_listener.execution_spans[mock_task] == mock_span
-
-            completed_event = TaskCompletedEvent(output=mock_task_output, task=mock_task)
-            future = crewai_event_bus.emit(mock_task, completed_event)
-            if future:
-                await asyncio.wrap_future(future)
-
-            assert mock_task not in event_listener.execution_spans
-
-
-@pytest.mark.asyncio
-async def test_execution_spans_removed_on_task_failed(event_listener, mock_task):
-    """Test that execution_spans entries are properly removed when a task fails.
-
-    This test verifies the fix for the memory leak where failed tasks were
-    setting execution_spans[source] = None instead of removing the key entirely.
-    """
-    with patch.object(event_listener._telemetry, "task_started") as mock_task_started:
-        with patch.object(event_listener._telemetry, "task_ended"):
-            mock_span = MagicMock()
-            mock_task_started.return_value = mock_span
-
-            start_event = TaskStartedEvent(context="test context", task=mock_task)
-            future = crewai_event_bus.emit(mock_task, start_event)
-            if future:
-                await asyncio.wrap_future(future)
-
-            assert mock_task in event_listener.execution_spans
-            assert event_listener.execution_spans[mock_task] == mock_span
-
-            failed_event = TaskFailedEvent(error="Test error", task=mock_task)
-            future = crewai_event_bus.emit(mock_task, failed_event)
-            if future:
-                await asyncio.wrap_future(future)
-
-            assert mock_task not in event_listener.execution_spans
-
-
-@pytest.mark.asyncio
-async def test_execution_spans_dict_size_does_not_grow_unbounded(
-    event_listener, mock_task_output
-):
-    """Test that execution_spans dictionary size remains bounded after many tasks.
-
-    This test simulates the memory leak scenario where many tasks complete/fail
-    and verifies that the dictionary doesn't grow unboundedly.
-    """
-    num_tasks = 100
-
-    with patch.object(event_listener._telemetry, "task_started") as mock_task_started:
-        with patch.object(event_listener._telemetry, "task_ended"):
-            mock_task_started.return_value = MagicMock()
-
-            for i in range(num_tasks):
-                task = MockTask(task_id=f"task_{i}")
-
-                start_event = TaskStartedEvent(context="test context", task=task)
-                future = crewai_event_bus.emit(task, start_event)
-                if future:
-                    await asyncio.wrap_future(future)
-
-                if i % 2 == 0:
-                    completed_event = TaskCompletedEvent(
-                        output=mock_task_output, task=task
-                    )
-                    future = crewai_event_bus.emit(task, completed_event)
-                else:
-                    failed_event = TaskFailedEvent(error="Test error", task=task)
-                    future = crewai_event_bus.emit(task, failed_event)
-
-                if future:
-                    await asyncio.wrap_future(future)
-
-            assert len(event_listener.execution_spans) == 0
-
-
-@pytest.mark.asyncio
-async def test_execution_spans_handles_missing_task_gracefully(
-    event_listener, mock_task, mock_task_output
-):
-    """Test that completing/failing a task not in execution_spans doesn't cause errors.
-
-    This ensures the fix using pop(source, None) handles edge cases gracefully.
-    """
-    with patch.object(event_listener._telemetry, "task_ended"):
-        assert mock_task not in event_listener.execution_spans
-
-        completed_event = TaskCompletedEvent(output=mock_task_output, task=mock_task)
-        future = crewai_event_bus.emit(mock_task, completed_event)
-        if future:
-            await asyncio.wrap_future(future)
-
-        assert mock_task not in event_listener.execution_spans
-
-
-@pytest.mark.asyncio
-async def test_execution_spans_handles_missing_task_on_failure_gracefully(
-    event_listener, mock_task
-):
-    """Test that failing a task not in execution_spans doesn't cause errors.
-
-    This ensures the fix using pop(source, None) handles edge cases gracefully.
-    """
-    with patch.object(event_listener._telemetry, "task_ended"):
-        assert mock_task not in event_listener.execution_spans
-
-        failed_event = TaskFailedEvent(error="Test error", task=mock_task)
-        future = crewai_event_bus.emit(mock_task, failed_event)
-        if future:
-            await asyncio.wrap_future(future)
-
-        assert mock_task not in event_listener.execution_spans
-
-
-@pytest.mark.asyncio
-async def test_telemetry_task_ended_called_with_span_on_completion(
-    event_listener, mock_task, mock_task_output
-):
-    """Test that telemetry.task_ended is called with the correct span on completion."""
-    with patch.object(event_listener._telemetry, "task_started") as mock_task_started:
-        with patch.object(event_listener._telemetry, "task_ended") as mock_task_ended:
-            mock_span = MagicMock()
-            mock_task_started.return_value = mock_span
-
-            start_event = TaskStartedEvent(context="test context", task=mock_task)
-            future = crewai_event_bus.emit(mock_task, start_event)
-            if future:
-                await asyncio.wrap_future(future)
-
-            completed_event = TaskCompletedEvent(output=mock_task_output, task=mock_task)
-            future = crewai_event_bus.emit(mock_task, completed_event)
-            if future:
-                await asyncio.wrap_future(future)
-
-            mock_task_ended.assert_called_once_with(
-                mock_span, mock_task, mock_task.agent.crew
-            )
-
-
-@pytest.mark.asyncio
-async def test_telemetry_task_ended_called_with_span_on_failure(
-    event_listener, mock_task
-):
-    """Test that telemetry.task_ended is called with the correct span on failure."""
-    with patch.object(event_listener._telemetry, "task_started") as mock_task_started:
-        with patch.object(event_listener._telemetry, "task_ended") as mock_task_ended:
-            mock_span = MagicMock()
-            mock_task_started.return_value = mock_span
-
-            start_event = TaskStartedEvent(context="test context", task=mock_task)
-            future = crewai_event_bus.emit(mock_task, start_event)
-            if future:
-                await asyncio.wrap_future(future)
-
-            failed_event = TaskFailedEvent(error="Test error", task=mock_task)
-            future = crewai_event_bus.emit(mock_task, failed_event)
-            if future:
-                await asyncio.wrap_future(future)
-
-            mock_task_ended.assert_called_once_with(
-                mock_span, mock_task, mock_task.agent.crew
-            )

lib/crewai/tests/utilities/test_file_handler.py

@@ -1,6 +1,8 @@
 import os
+import threading
 import unittest
 import uuid
+from concurrent.futures import ThreadPoolExecutor, as_completed
 
 import pytest
 from crewai.utilities.file_handler import PickleHandler
@@ -8,7 +10,6 @@ from crewai.utilities.file_handler import PickleHandler
 
 class TestPickleHandler(unittest.TestCase):
     def setUp(self):
-        # Use a unique file name for each test to avoid race conditions in parallel test execution
         unique_id = str(uuid.uuid4())
         self.file_name = f"test_data_{unique_id}.pkl"
         self.file_path = os.path.join(os.getcwd(), self.file_name)
@@ -47,3 +48,234 @@ class TestPickleHandler(unittest.TestCase):
 
         assert str(exc.value) == "pickle data was truncated"
         assert "<class '_pickle.UnpicklingError'>" == str(exc.type)
+
+
+class TestPickleHandlerThreadSafety(unittest.TestCase):
+    """Tests for thread-safety of PickleHandler operations."""
+
+    def setUp(self):
+        unique_id = str(uuid.uuid4())
+        self.file_name = f"test_thread_safe_{unique_id}.pkl"
+        self.file_path = os.path.join(os.getcwd(), self.file_name)
+        self.handler = PickleHandler(self.file_name)
+
+    def tearDown(self):
+        if os.path.exists(self.file_path):
+            os.remove(self.file_path)
+
+    def test_concurrent_writes_same_handler(self):
+        """Test that concurrent writes from multiple threads using the same handler don't corrupt data."""
+        num_threads = 10
+        num_writes_per_thread = 20
+        errors: list[Exception] = []
+        write_count = 0
+        count_lock = threading.Lock()
+
+        def write_data(thread_id: int) -> None:
+            nonlocal write_count
+            for i in range(num_writes_per_thread):
+                try:
+                    data = {"thread": thread_id, "iteration": i, "data": f"value_{thread_id}_{i}"}
+                    self.handler.save(data)
+                    with count_lock:
+                        write_count += 1
+                except Exception as e:
+                    errors.append(e)
+
+        threads = []
+        for i in range(num_threads):
+            t = threading.Thread(target=write_data, args=(i,))
+            threads.append(t)
+            t.start()
+
+        for t in threads:
+            t.join()
+
+        assert len(errors) == 0, f"Errors occurred during concurrent writes: {errors}"
+        assert write_count == num_threads * num_writes_per_thread
+        loaded_data = self.handler.load()
+        assert isinstance(loaded_data, dict)
+        assert "thread" in loaded_data
+        assert "iteration" in loaded_data
+
+    def test_concurrent_reads_same_handler(self):
+        """Test that concurrent reads from multiple threads don't cause issues."""
+        test_data = {"key": "value", "nested": {"a": 1, "b": 2}}
+        self.handler.save(test_data)
+
+        num_threads = 20
+        results: list[dict] = []
+        errors: list[Exception] = []
+        results_lock = threading.Lock()
+
+        def read_data() -> None:
+            try:
+                data = self.handler.load()
+                with results_lock:
+                    results.append(data)
+            except Exception as e:
+                errors.append(e)
+
+        threads = []
+        for _ in range(num_threads):
+            t = threading.Thread(target=read_data)
+            threads.append(t)
+            t.start()
+
+        for t in threads:
+            t.join()
+
+        assert len(errors) == 0, f"Errors occurred during concurrent reads: {errors}"
+        assert len(results) == num_threads
+        for result in results:
+            assert result == test_data
+
+    def test_concurrent_read_write_same_handler(self):
+        """Test that concurrent reads and writes don't corrupt data or cause errors."""
+        initial_data = {"counter": 0}
+        self.handler.save(initial_data)
+
+        num_writers = 5
+        num_readers = 10
+        writes_per_thread = 10
+        reads_per_thread = 20
+        write_errors: list[Exception] = []
+        read_errors: list[Exception] = []
+        read_results: list[dict] = []
+        results_lock = threading.Lock()
+
+        def writer(thread_id: int) -> None:
+            for i in range(writes_per_thread):
+                try:
+                    data = {"writer": thread_id, "write_num": i}
+                    self.handler.save(data)
+                except Exception as e:
+                    write_errors.append(e)
+
+        def reader() -> None:
+            for _ in range(reads_per_thread):
+                try:
+                    data = self.handler.load()
+                    with results_lock:
+                        read_results.append(data)
+                except Exception as e:
+                    read_errors.append(e)
+
+        threads = []
+        for i in range(num_writers):
+            t = threading.Thread(target=writer, args=(i,))
+            threads.append(t)
+
+        for _ in range(num_readers):
+            t = threading.Thread(target=reader)
+            threads.append(t)
+
+        for t in threads:
+            t.start()
+
+        for t in threads:
+            t.join()
+
+        assert len(write_errors) == 0, f"Write errors: {write_errors}"
+        assert len(read_errors) == 0, f"Read errors: {read_errors}"
+        for result in read_results:
+            assert isinstance(result, dict)
+
+    def test_atomic_write_no_partial_data(self):
+        """Test that atomic writes prevent partial/corrupted data from being read."""
+        large_data = {"key": "x" * 100000, "numbers": list(range(10000))}
+        num_iterations = 50
+        errors: list[Exception] = []
+        corruption_detected = False
+        corruption_lock = threading.Lock()
+
+        def writer() -> None:
+            for _ in range(num_iterations):
+                try:
+                    self.handler.save(large_data)
+                except Exception as e:
+                    errors.append(e)
+
+        def reader() -> None:
+            nonlocal corruption_detected
+            for _ in range(num_iterations * 2):
+                try:
+                    data = self.handler.load()
+                    if data and data != {} and data != large_data:
+                        with corruption_lock:
+                            corruption_detected = True
+                except Exception as e:
+                    errors.append(e)
+
+        writer_thread = threading.Thread(target=writer)
+        reader_thread = threading.Thread(target=reader)
+
+        writer_thread.start()
+        reader_thread.start()
+
+        writer_thread.join()
+        reader_thread.join()
+
+        assert len(errors) == 0, f"Errors occurred: {errors}"
+        assert not corruption_detected, "Partial/corrupted data was read"
+
+    def test_thread_pool_concurrent_operations(self):
+        """Test thread safety using ThreadPoolExecutor for more realistic concurrent access."""
+        num_operations = 100
+        errors: list[Exception] = []
+
+        def operation(op_id: int) -> str:
+            try:
+                if op_id % 3 == 0:
+                    self.handler.save({"op_id": op_id, "type": "write"})
+                    return f"write_{op_id}"
+                else:
+                    data = self.handler.load()
+                    return f"read_{op_id}_{type(data).__name__}"
+            except Exception as e:
+                errors.append(e)
+                return f"error_{op_id}"
+
+        with ThreadPoolExecutor(max_workers=20) as executor:
+            futures = [executor.submit(operation, i) for i in range(num_operations)]
+            results = [f.result() for f in as_completed(futures)]
+
+        assert len(errors) == 0, f"Errors occurred: {errors}"
+        assert len(results) == num_operations
+
+    def test_multiple_handlers_same_file(self):
+        """Test that multiple PickleHandler instances for the same file work correctly."""
+        handler1 = PickleHandler(self.file_name)
+        handler2 = PickleHandler(self.file_name)
+
+        num_operations = 50
+        errors: list[Exception] = []
+
+        def use_handler1() -> None:
+            for i in range(num_operations):
+                try:
+                    handler1.save({"handler": 1, "iteration": i})
+                except Exception as e:
+                    errors.append(e)
+
+        def use_handler2() -> None:
+            for i in range(num_operations):
+                try:
+                    handler2.save({"handler": 2, "iteration": i})
+                except Exception as e:
+                    errors.append(e)
+
+        t1 = threading.Thread(target=use_handler1)
+        t2 = threading.Thread(target=use_handler2)
+
+        t1.start()
+        t2.start()
+
+        t1.join()
+        t2.join()
+
+        assert len(errors) == 0, f"Errors occurred: {errors}"
+        final_data = self.handler.load()
+        assert isinstance(final_data, dict)
+        assert "handler" in final_data
+        assert "iteration" in final_data