fix(cloud): start API before runtime reconcile

src/langbot/pkg/core/app.py

@@ -301,6 +301,22 @@ class Application:
     async def initialize(self):
         pass
 
+    async def _initialize_plugin_runtime(self) -> None:
+        try:
+            await self.plugin_connector.initialize()
+        except asyncio.CancelledError:
+            raise
+        except Exception as exc:
+            self.logger.warning(f'Plugin runtime unavailable during startup; reconnecting in background: {exc}')
+            self.plugin_connector.schedule_reconnect()
+
+    def _start_plugin_runtime_initialization(self):
+        return self.task_mgr.create_task(
+            self._initialize_plugin_runtime(),
+            name='plugin-runtime-initialization',
+            scopes=[core_entities.LifecycleControlScope.APPLICATION],
+        )
+
     async def run(self):
         self.event_loop_monitor.start()
         try:
@@ -322,8 +338,6 @@ class Application:
                     name='cloud-manifest-refresh',
                     scopes=[core_entities.LifecycleControlScope.APPLICATION],
                 )
-            await self.plugin_connector.initialize_plugins()
-
             # 后续可能会允许动态重启其他任务
             # 故为了防止程序在非 Ctrl-C 情况下退出，这里创建一个不会结束的协程
             async def never_ending():
@@ -348,6 +362,7 @@ class Application:
                 name='http-api-controller',
                 scopes=[core_entities.LifecycleControlScope.APPLICATION],
             )
+            self._start_plugin_runtime_initialization()
 
             # Telemetry instance heartbeat (startup + daily); respects
             # space.disable_telemetry via TelemetryManager.send().

src/langbot/pkg/core/stages/build_app.py

@@ -303,13 +303,6 @@ class BuildAppStage(stage.BootingStage):
             )
 
         plugin_connector_inst = plugin_connector.PluginRuntimeConnector(ap, runtime_disconnect_callback)
-        try:
-            await plugin_connector_inst.initialize()
-        except Exception as exc:
-            # Keep the API/UI available while an external or managed runtime is
-            # starting, then recover in the background with bounded backoff.
-            ap.logger.warning(f'Plugin runtime unavailable during startup; reconnecting in background: {exc}')
-            plugin_connector_inst.schedule_reconnect()
         ap.plugin_connector = plugin_connector_inst
         workspace_service_inst.release_startup_execution_bindings()
 

tests/unit_tests/core/test_app_shutdown.py

@@ -144,3 +144,18 @@ async def test_runtime_resource_stats_are_aggregate_and_constant_time() -> None:
     assert stats['models']['providers'] == 1
     assert stats['runtimes']['plugin_installations'] == 1
     assert stats['runtimes']['plugin_runtime_connected'] is True
+
+
+@pytest.mark.asyncio
+async def test_start_plugin_runtime_initialization_is_scheduled() -> None:
+    app = Application()
+    app.plugin_connector = SimpleNamespace(initialize=AsyncMock())
+    captured = {}
+    app.task_mgr = SimpleNamespace(create_task=lambda coro, **kwargs: captured.update(coro=coro, kwargs=kwargs))
+
+    app._start_plugin_runtime_initialization()
+
+    assert captured['kwargs']['name'] == 'plugin-runtime-initialization'
+    assert captured['kwargs']['scopes']
+    await captured['coro']
+    app.plugin_connector.initialize.assert_awaited_once_with()