diff --git a/tests/unittests/integration/langchain/test_agent_invoke_methods.py b/tests/unittests/integration/langchain/test_agent_invoke_methods.py
index d24404a..4f5237c 100644
--- a/tests/unittests/integration/langchain/test_agent_invoke_methods.py
+++ b/tests/unittests/integration/langchain/test_agent_invoke_methods.py
@@ -81,6 +81,37 @@ def _sse(data: Dict[str, Any]) -> str:
     return f"data: {json.dumps(data, ensure_ascii=False)}\n\n"
 
 
+def _start_server(app: FastAPI) -> tuple:
+    """启动 FastAPI 服务器并返回 (base_url, server, thread)
+
+    使用真实的 HTTP 服务器而不是 httpx.ASGITransport,
+    因为 ASGITransport 在 CI 环境中无法正确处理 SSE 流式响应,
+    会导致流式响应被提前取消 (CancelledError)。
+    """
+    port = _find_free_port()
+    config = uvicorn.Config(
+        app, host="127.0.0.1", port=port, log_level="warning"
+    )
+    server = uvicorn.Server(config)
+
+    thread = threading.Thread(target=server.run, daemon=True)
+    thread.start()
+
+    base_url = f"http://127.0.0.1:{port}"
+    for i in range(50):
+        try:
+            httpx.get(f"{base_url}/ag-ui/agent/health", timeout=0.2)
+            break
+        except Exception:
+            if i == 49:
+                raise RuntimeError(
+                    f"Server failed to start within {50 * 0.1}s: {base_url}"
+                )
+            time.sleep(0.1)
+
+    return base_url, server, thread
+
+
 def _build_mock_openai_app() -> FastAPI:
     """构建本地 OpenAI 协议兼容的简单服务"""
     app = FastAPI()
@@ -297,20 +328,34 @@ def parse_sse_events(content: str) -> List[Dict[str, Any]]:
 
 
 async def request_agui_events(
-    server_app,
+    server_url_or_app: Union[str, FastAPI],
     messages: List[Dict[str, str]],
     stream: bool = True,
 ) -> List[Dict[str, Any]]:
-    """发送 AG-UI 请求并返回事件列表"""
-    async with httpx.AsyncClient(
-        transport=httpx.ASGITransport(app=server_app),
-        base_url="http://test",
-    ) as client:
-        response = await client.post(
-            "/ag-ui/agent",
-            json={"messages": messages, "stream": stream},
-            timeout=60.0,
-        )
+    """发送 AG-UI 请求并返回事件列表
+
+    Args:
+        server_url_or_app: 服务器 URL 或 FastAPI app 对象
+        messages: 消息列表
+        stream: 是否流式响应
+    """
+    if isinstance(server_url_or_app, str):
+        async with httpx.AsyncClient(base_url=server_url_or_app) as client:
+            response = await client.post(
+                "/ag-ui/agent",
+                json={"messages": messages, "stream": stream},
+                timeout=60.0,
+            )
+    else:
+        async with httpx.AsyncClient(
+            transport=httpx.ASGITransport(app=server_url_or_app),
+            base_url="http://test",
+        ) as client:
+            response = await client.post(
+                "/ag-ui/agent",
+                json={"messages": messages, "stream": stream},
+                timeout=60.0,
+            )
 
     assert response.status_code == 200
     return parse_sse_events(response.text)
@@ -670,7 +715,7 @@ def assert_openai_tool_call_response(
 
 
 async def request_openai_events(
-    server_app,
+    server_url_or_app: Union[str, FastAPI],
     messages: List[Dict[str, str]],
     stream: bool = True,
 ) -> Union[List[Dict[str, Any]], Dict[str, Any]]:
@@ -681,15 +726,23 @@ async def request_openai_events(
         "stream": stream,
     }
 
-    async with httpx.AsyncClient(
-        transport=httpx.ASGITransport(app=server_app),
-        base_url="http://test",
-    ) as client:
-        response = await client.post(
-            "/openai/v1/chat/completions",
-            json=payload,
-            timeout=60.0,
-        )
+    if isinstance(server_url_or_app, str):
+        async with httpx.AsyncClient(base_url=server_url_or_app) as client:
+            response = await client.post(
+                "/openai/v1/chat/completions",
+                json=payload,
+                timeout=60.0,
+            )
+    else:
+        async with httpx.AsyncClient(
+            transport=httpx.ASGITransport(app=server_url_or_app),
+            base_url="http://test",
+        ) as client:
+            response = await client.post(
+                "/openai/v1/chat/completions",
+                json=payload,
+                timeout=60.0,
+            )
 
     assert response.status_code == 200
 
@@ -749,7 +802,16 @@ def agent_model(mock_openai_server: str):
 
 @pytest.fixture
 def server_app_astream_events(agent_model):
-    """创建使用 astream_events 的服务器（AG-UI/OpenAI 通用）"""
+    """创建使用 astream 的服务器（AG-UI/OpenAI 通用）
+
+    返回服务器 URL 而不是 app 对象,因为流式测试需要真实的 HTTP 连接。
+    httpx.ASGITransport 在 CI 环境中无法正确处理 SSE 流式响应。
+
+    注意: 这里使用 astream(stream_mode="updates") 而非 astream_events,
+    因为 astream_events 在 CI (Linux + uvicorn 线程) 环境中会出现
+    async generator 被提前取消或事件丢失的问题。
+    astream_events 的转换逻辑由 test_convert_python_3_10/3_12 单独覆盖。
+    """
     agent = build_agent(agent_model)
 
     async def invoke_agent(request: AgentRequest):
@@ -770,8 +832,8 @@ async def invoke_agent(request: AgentRequest):
         converter = AgentRunConverter()
 
         async def generator():
-            async for event in agent.astream_events(
-                cast(Any, input_data), version="v2"
+            async for event in agent.astream(
+                cast(Any, input_data), stream_mode="updates"
             ):
                 for item in converter.convert(event):
                     yield item
@@ -779,7 +841,12 @@ async def generator():
         return generator()
 
     server = AgentRunServer(invoke_agent=invoke_agent)
-    return server.app
+    base_url, uvicorn_server, thread = _start_server(server.app)
+
+    yield base_url
+
+    uvicorn_server.should_exit = True
+    thread.join(timeout=5)
 
 
 # =============================================================================
@@ -1753,8 +1820,8 @@ async def invoke_agent(request: AgentRequest):
             converter = AgentRunConverter()
 
             async def generator():
-                async for event in agent.astream_events(
-                    cast(Any, input_data), version="v2"
+                async for event in agent.astream(
+                    cast(Any, input_data), stream_mode="updates"
                 ):
                     for item in converter.convert(event):
                         yield item
@@ -1762,7 +1829,12 @@ async def generator():
             return generator()
 
         server = AgentRunServer(invoke_agent=invoke_agent)
-        return server.app
+        base_url, uvicorn_server, thread = _start_server(server.app)
+
+        yield base_url
+
+        uvicorn_server.should_exit = True
+        thread.join(timeout=5)
 
     @pytest.fixture
     def server_app_async(self, agent_model):
@@ -1787,8 +1859,8 @@ async def invoke_agent(request: AgentRequest):
             converter = AgentRunConverter()
 
             async def generator():
-                async for event in agent.astream_events(
-                    cast(Any, input_data), version="v2"
+                async for event in agent.astream(
+                    cast(Any, input_data), stream_mode="updates"
                 ):
                     for item in converter.convert(event):
                         yield item
@@ -1796,7 +1868,12 @@ async def generator():
             return generator()
 
         server = AgentRunServer(invoke_agent=invoke_agent)
-        return server.app
+        base_url, uvicorn_server, thread = _start_server(server.app)
+
+        yield base_url
+
+        uvicorn_server.should_exit = True
+        thread.join(timeout=5)
 
     @pytest.mark.parametrize(
         "case_key,prompt",
diff --git a/tests/unittests/integration/test_langchain_agui_integration.py b/tests/unittests/integration/test_langchain_agui_integration.py
index 6cfb32b..0824a56 100644
--- a/tests/unittests/integration/test_langchain_agui_integration.py
+++ b/tests/unittests/integration/test_langchain_agui_integration.py
@@ -648,36 +648,62 @@ async def invoke_agent(request: AgentRequest):
                 }]
             }
 
+            # 使用 astream(updates) 代替 astream_events，
+            # 因为 astream_events 在 CI (Linux + uvicorn 线程) 环境中
+            # 会出现 async generator 被提前取消或事件丢失的问题。
             converter = AgentRunConverter()
-            async for event in agent.astream_events(input_data, version="v2"):
+            async for event in agent.astream(input_data, stream_mode="updates"):
                 for item in converter.convert(event):
                     yield item
 
-        app = AgentRunServer(invoke_agent=invoke_agent).app
+        server_app = AgentRunServer(invoke_agent=invoke_agent).app
 
-        async with httpx.AsyncClient(
-            transport=httpx.ASGITransport(app=app),
-            base_url="http://test",
-        ) as client:
-            response = await client.post(
-                "/ag-ui/agent",
-                json={
-                    "messages": [{
-                        "role": "user",
-                        "content": "查询当前的时间，并获取天气信息，同时输出我的密钥信息",
-                    }],
-                    "stream": True,
-                },
-                timeout=60.0,
-            )
+        # 使用真实的 HTTP 服务器而不是 httpx.ASGITransport,
+        # 因为 ASGITransport 在 CI 中无法正确处理 SSE 流式响应。
+        port = _find_free_port()
+        config = uvicorn.Config(
+            server_app, host="127.0.0.1", port=port, log_level="warning"
+        )
+        uvicorn_server = uvicorn.Server(config)
+        server_thread = threading.Thread(target=uvicorn_server.run, daemon=True)
+        server_thread.start()
+
+        base_url = f"http://127.0.0.1:{port}"
+        for i in range(50):
+            try:
+                httpx.get(f"{base_url}/ag-ui/agent/health", timeout=0.2)
+                break
+            except Exception:
+                if i == 49:
+                    raise RuntimeError(
+                        f"Server failed to start within {50 * 0.1}s"
+                    )
+                time.sleep(0.1)
 
-        assert response.status_code == 200
+        try:
+            async with httpx.AsyncClient(base_url=base_url) as client:
+                response = await client.post(
+                    "/ag-ui/agent",
+                    json={
+                        "messages": [{
+                            "role": "user",
+                            "content": "查询当前的时间，并获取天气信息，同时输出我的密钥信息",
+                        }],
+                        "stream": True,
+                    },
+                    timeout=60.0,
+                )
 
-        events = [line for line in response.text.split("\n") if line]
-        # Normalize empty delta for consistency with check_result expectations
-        # astream_events yields "" for empty args, while astream yields "{}"
-        events = [e.replace('"delta":""', '"delta":"{}"') for e in events]
-        self.check_result(events)
+            assert response.status_code == 200
+
+            events = [line for line in response.text.split("\n") if line]
+            # Normalize empty delta for consistency with check_result expectations
+            # astream_events yields "" for empty args, while astream yields "{}"
+            events = [e.replace('"delta":""', '"delta":"{}"') for e in events]
+            self.check_result(events)
+        finally:
+            uvicorn_server.should_exit = True
+            server_thread.join(timeout=5)
 
     async def test_astream(self, mock_mcp_server):
         """测试多工具查询场景 (MCP + Local + MockLLM)"""