diff --git a/capiscio_sdk/executor.py b/capiscio_sdk/executor.py
index ecb7075..30b697c 100644
--- a/capiscio_sdk/executor.py
+++ b/capiscio_sdk/executor.py
@@ -8,6 +8,13 @@
 except ImportError:
     RequestContext = Any  # type: ignore[misc,assignment]
 
+try:
+    from google.protobuf.json_format import MessageToDict
+    from google.protobuf.message import Message as ProtobufMessage
+except ImportError:
+    MessageToDict = None  # type: ignore[assignment,misc]
+    ProtobufMessage = None  # type: ignore[assignment,misc]
+
 from .config import SecurityConfig
 from .validators import MessageValidator, ProtocolValidator
 from .infrastructure import ValidationCache, RateLimiter
@@ -86,7 +93,12 @@ async def execute(self, context: RequestContext, event_queue: Any) -> None:
             return
         
         # Convert message to dict for validation (our validators expect dict format)
-        message_dict = message.model_dump() if hasattr(message, 'model_dump') else {}
+        if hasattr(message, 'model_dump'):
+            message_dict = message.model_dump()
+        elif ProtobufMessage is not None and isinstance(message, ProtobufMessage):
+            message_dict = MessageToDict(message, preserving_proto_field_name=True)
+        else:
+            message_dict = {}
         
         # Extract identifier for rate limiting
         identifier = message_dict.get("message_id") or message.message_id
diff --git a/examples/simple_agent/agent_executor.py b/examples/simple_agent/agent_executor.py
index d108782..dd0e07a 100644
--- a/examples/simple_agent/agent_executor.py
+++ b/examples/simple_agent/agent_executor.py
@@ -8,7 +8,7 @@
 
 from a2a.server.agent_execution import AgentExecutor, RequestContext
 from a2a.server.events import EventQueue
-from a2a.utils import new_agent_text_message
+from a2a.types import Message, Part, Role
 
 
 class SimpleAgent:
@@ -69,7 +69,12 @@ async def execute(
         result = await self.agent.invoke(text or "hello")
         
         # Send response back to client
-        await event_queue.enqueue_event(new_agent_text_message(result))
+        response_msg = Message(
+            message_id="response",
+            role=Role.ROLE_AGENT,
+            parts=[Part(text=result)],
+        )
+        await event_queue.enqueue_event(response_msg)
     
     async def cancel(
         self,
@@ -84,6 +89,9 @@ async def cancel(
             event_queue: Queue for sending events
         """
         # This simple agent doesn't support cancellation
-        await event_queue.enqueue_event(
-            new_agent_text_message("Task cancellation not supported by this simple agent.")
+        cancel_msg = Message(
+            message_id="cancel-response",
+            role=Role.ROLE_AGENT,
+            parts=[Part(text="Task cancellation not supported by this simple agent.")],
         )
+        await event_queue.enqueue_event(cancel_msg)
diff --git a/examples/simple_agent/test_client.py b/examples/simple_agent/test_client.py
index 1ecf8d5..c2b7069 100644
--- a/examples/simple_agent/test_client.py
+++ b/examples/simple_agent/test_client.py
@@ -17,7 +17,8 @@
 import httpx
 import asyncio
 from datetime import datetime
-from a2a.types import Message, TextPart, Role, MessageSendParams
+from a2a.types import Message, Part, Role, SendMessageRequest
+from google.protobuf.json_format import MessageToDict
 
 
 async def send_message(client: httpx.AsyncClient, text: str, message_id: str = None):
@@ -28,16 +29,16 @@ async def send_message(client: httpx.AsyncClient, text: str, message_id: str = N
     # Create proper A2A message
     message = Message(
         message_id=message_id,
-        role=Role.user,
-        parts=[TextPart(text=text)]
+        role=Role.ROLE_USER,
+        parts=[Part(text=text)]
     )
     
-    # Wrap in MessageSendParams and serialize
-    params = MessageSendParams(message=message)
+    # Wrap in SendMessageRequest and serialize
+    params = SendMessageRequest(message=message)
     
     response = await client.post(
         "http://localhost:8080/v1/tasks",
-        json=params.model_dump(mode="json")
+        json=MessageToDict(params)
     )
     return response
 
diff --git a/pyproject.toml b/pyproject.toml
index 22709c6..f3bdddc 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -27,7 +27,7 @@ classifiers = [
 ]
 
 dependencies = [
-    "a2a-sdk>=0.1.0",
+    "a2a-sdk>=1.0.0",
     "httpx>=0.27.0",
     "pydantic>=2.0.0",
     "cryptography>=42.0.0",
diff --git a/tests/integration/test_real_executor.py b/tests/integration/test_real_executor.py
index ab73229..c1f4c6f 100644
--- a/tests/integration/test_real_executor.py
+++ b/tests/integration/test_real_executor.py
@@ -13,9 +13,11 @@
 import pytest
 import base64
 from a2a.server.agent_execution import AgentExecutor, RequestContext
+from a2a.server.context import ServerCallContext
 from a2a.server.events import EventQueue
-from a2a.types import Message, TextPart, FilePart, DataPart, Role, MessageSendParams, FileWithBytes, FileWithUri
-from a2a.utils import new_agent_text_message
+from a2a.types import Message, Part, Role, SendMessageRequest
+from google.protobuf.json_format import ParseDict
+from google.protobuf.struct_pb2 import Struct, Value
 from capiscio_sdk import secure, SecurityConfig
 from capiscio_sdk.errors import CapiscioValidationError, CapiscioRateLimitError
 
@@ -52,27 +54,37 @@ async def execute(self, context: RequestContext, event_queue: EventQueue) -> Non
         
         # Process and respond
         result = await self.agent.invoke(text)
-        await event_queue.enqueue_event(new_agent_text_message(result))
+        response_msg = Message(
+            message_id="response-1",
+            role=Role.ROLE_AGENT,
+            parts=[Part(text=result)],
+        )
+        await event_queue.enqueue_event(response_msg)
     
     async def cancel(self, context: RequestContext, event_queue: EventQueue) -> None:
         """Handle cancellation."""
-        await event_queue.enqueue_event(new_agent_text_message("Cancelled"))
+        cancel_msg = Message(
+            message_id="cancel-1",
+            role=Role.ROLE_AGENT,
+            parts=[Part(text="Cancelled")],
+        )
+        await event_queue.enqueue_event(cancel_msg)
 
 
 def create_valid_message(text="test", message_id="msg-1"):
     """Helper to create a valid A2A message."""
     return Message(
         message_id=message_id,
-        role=Role.user,
-        parts=[TextPart(text=text)]
+        role=Role.ROLE_USER,
+        parts=[Part(text=text)]
     )
 
 
 def create_request_context(message):
     """Helper to create a RequestContext."""
-    # RequestContext expects MessageSendParams, not Message directly
-    params = MessageSendParams(message=message)
-    return RequestContext(request=params)
+    # RequestContext expects SendMessageRequest, not Message directly
+    params = SendMessageRequest(message=message)
+    return RequestContext(call_context=ServerCallContext(), request=params)
 
 
 class SimpleEventQueue:
@@ -119,7 +131,7 @@ async def test_integration_invalid_message_blocked():
     # Create invalid message (empty message_id)
     invalid_message = Message(
         message_id="",  # Empty!
-        role=Role.user,
+        role=Role.ROLE_USER,
         parts=[]  # Empty!
     )
     
@@ -174,7 +186,7 @@ async def test_integration_monitor_mode_allows_invalid():
     # Create invalid message
     invalid_message = Message(
         message_id="",  # Empty!
-        role=Role.user,
+        role=Role.ROLE_USER,
         parts=[]
     )
     
@@ -319,12 +331,12 @@ async def test_integration_file_part_with_bytes():
     file_content = b"Hello, this is file content!"
     message = Message(
         message_id="msg-file-1",
-        role=Role.user,
-        parts=[FilePart(file=FileWithBytes(
-            bytes=base64.b64encode(file_content).decode('utf-8'),
+        role=Role.ROLE_USER,
+        parts=[Part(
+            raw=file_content,
             media_type="text/plain",
-            name="test.txt"
-        ))]
+            filename="test.txt",
+        )]
     )
     
     context = create_request_context(message)
@@ -347,12 +359,12 @@ async def test_integration_file_part_with_uri():
     # Create message with FilePart (URI)
     message = Message(
         message_id="msg-file-2",
-        role=Role.user,
-        parts=[FilePart(file=FileWithUri(
-            uri="https://example.com/document.pdf",
+        role=Role.ROLE_USER,
+        parts=[Part(
+            url="https://example.com/document.pdf",
             media_type="application/pdf",
-            name="document.pdf"
-        ))]
+            filename="document.pdf",
+        )]
     )
     
     context = create_request_context(message)
@@ -373,14 +385,14 @@ async def test_integration_data_part_structured():
     secured = secure(executor, SecurityConfig.production())
     
     # Create message with DataPart
+    data_value = ParseDict(
+        {"query": "SELECT * FROM users", "parameters": {"limit": 10}, "metadata": {"source": "analytics"}},
+        Value(),
+    )
     message = Message(
         message_id="msg-data-1",
-        role=Role.user,
-        parts=[DataPart(data={
-            "query": "SELECT * FROM users",
-            "parameters": {"limit": 10},
-            "metadata": {"source": "analytics"}
-        })]
+        role=Role.ROLE_USER,
+        parts=[Part(data=data_value)]
     )
     
     context = create_request_context(message)
@@ -401,13 +413,14 @@ async def test_integration_multiple_parts_mixed_types():
     secured = secure(executor, SecurityConfig.production())
     
     # Create message with mixed parts
+    data_value = ParseDict({"query": "search term"}, Value())
     message = Message(
         message_id="msg-mixed-1",
-        role=Role.user,
+        role=Role.ROLE_USER,
         parts=[
-            TextPart(text="Here's the query and data:"),
-            DataPart(data={"query": "search term"}),
-            TextPart(text="Please process this.")
+            Part(text="Here's the query and data:"),
+            Part(data=data_value),
+            Part(text="Please process this.")
         ]
     )
     
@@ -431,8 +444,8 @@ async def test_integration_agent_role_message():
     # Create message from agent
     message = Message(
         message_id="msg-agent-1",
-        role=Role.agent,  # Agent sending message
-        parts=[TextPart(text="Response from upstream agent")]
+        role=Role.ROLE_AGENT,  # Agent sending message
+        parts=[Part(text="Response from upstream agent")]
     )
     
     context = create_request_context(message)
@@ -455,8 +468,8 @@ async def test_integration_message_with_context_and_task_ids():
     # Create message with optional fields
     message = Message(
         message_id="msg-ctx-1",
-        role=Role.user,
-        parts=[TextPart(text="Continuing task")],
+        role=Role.ROLE_USER,
+        parts=[Part(text="Continuing task")],
         context_id="ctx-123",
         task_id="task-456"
     )
@@ -479,11 +492,13 @@ async def test_integration_message_with_metadata():
     secured = secure(executor, SecurityConfig.production())
     
     # Create message with metadata
+    meta = Struct()
+    meta.update({"priority": "high", "source": "api"})
     message = Message(
         message_id="msg-meta-1",
-        role=Role.user,
-        parts=[TextPart(text="Test with metadata")],
-        metadata={"priority": "high", "source": "api"}
+        role=Role.ROLE_USER,
+        parts=[Part(text="Test with metadata")],
+        metadata=meta,
     )
     
     context = create_request_context(message)
@@ -506,8 +521,8 @@ async def test_integration_empty_text_part():
     # Create message with empty text
     message = Message(
         message_id="msg-empty-1",
-        role=Role.user,
-        parts=[TextPart(text="")]  # Empty text
+        role=Role.ROLE_USER,
+        parts=[Part(text="")]  # Empty text
     )
     
     context = create_request_context(message)
@@ -531,8 +546,8 @@ async def test_integration_very_long_text():
     long_text = "A" * 10000
     message = Message(
         message_id="msg-long-1",
-        role=Role.user,
-        parts=[TextPart(text=long_text)]
+        role=Role.ROLE_USER,
+        parts=[Part(text=long_text)]
     )
     
     context = create_request_context(message)
@@ -555,8 +570,8 @@ async def test_integration_special_characters_unicode():
     # Create message with various Unicode characters
     message = Message(
         message_id="msg-unicode-1",
-        role=Role.user,
-        parts=[TextPart(text="Hello 世界! 🚀 Émojis & Spëcial çhars: <>&\"'")]
+        role=Role.ROLE_USER,
+        parts=[Part(text="Hello 世界! 🚀 Émojis & Spëcial çhars: <>&\"'")]
     )
     
     context = create_request_context(message)
@@ -584,8 +599,8 @@ async def test_integration_xss_attempt_in_text():
     # Create message with XSS pattern (should be allowed to pass - content validation is app responsibility)
     message = Message(
         message_id="msg-xss-1",
-        role=Role.user,
-        parts=[TextPart(text="<script>alert('XSS')</script>")]
+        role=Role.ROLE_USER,
+        parts=[Part(text="<script>alert('XSS')</script>")]
     )
     
     context = create_request_context(message)
@@ -609,8 +624,8 @@ async def test_integration_sql_injection_pattern():
     # Create message with SQL injection pattern
     message = Message(
         message_id="msg-sql-1",
-        role=Role.user,
-        parts=[TextPart(text="'; DROP TABLE users; --")]
+        role=Role.ROLE_USER,
+        parts=[Part(text="'; DROP TABLE users; --")]
     )
     
     context = create_request_context(message)
@@ -631,10 +646,10 @@ async def test_integration_oversized_message_parts():
     secured = secure(executor, SecurityConfig.production())
     
     # Create message with many parts (100 parts)
-    parts = [TextPart(text=f"Part {i}") for i in range(100)]
+    parts = [Part(text=f"Part {i}") for i in range(100)]
     message = Message(
         message_id="msg-large-1",
-        role=Role.user,
+        role=Role.ROLE_USER,
         parts=parts
     )
     
@@ -658,8 +673,8 @@ async def test_integration_null_bytes_in_text():
     # Create message with null bytes
     message = Message(
         message_id="msg-null-1",
-        role=Role.user,
-        parts=[TextPart(text="Hello\x00World")]
+        role=Role.ROLE_USER,
+        parts=[Part(text="Hello\x00World")]
     )
     
     context = create_request_context(message)
@@ -680,7 +695,7 @@ async def test_integration_null_bytes_in_text():
 @pytest.mark.asyncio
 async def test_integration_invalid_role_value():
     """Test: Invalid enum values in role are caught by SDK type checking."""
-    # Note: The A2A SDK's Pydantic models prevent invalid role values
+    # Note: The A2A SDK's protobuf models prevent invalid role values
     # at construction time, so this test validates that the SDK itself
     # provides type safety. Invalid roles cannot reach our validator.
     
@@ -695,7 +710,7 @@ async def test_integration_invalid_role_value():
         Message(
             message_id="msg-bad-1",
             role="hacker",  # Invalid role - SDK validates this
-            parts=[TextPart(text="Hello")]
+            parts=[Part(text="Hello")]
         )
 
 
@@ -711,8 +726,8 @@ async def test_integration_missing_message_id():
     # Create message with empty messageId
     message = Message(
         message_id="",  # Empty!
-        role=Role.user,
-        parts=[TextPart(text="Test")]
+        role=Role.ROLE_USER,
+        parts=[Part(text="Test")]
     )
     
     context = create_request_context(message)
@@ -740,7 +755,7 @@ async def test_integration_empty_parts_array():
     # Create message with empty parts
     message = Message(
         message_id="msg-empty-parts",
-        role=Role.user,
+        role=Role.ROLE_USER,
         parts=[]  # Empty but valid structure
     )