unify plugin llm clients on shared utils layer

plugins/ai_auto_response/main.py

@@ -8,6 +8,7 @@ from loguru import logger
 
 from base.plugin_common.message_plugin_interface import MessagePluginInterface
 from base.plugin_common.plugin_interface import PluginStatus
+from utils.ai.unified_llm import UnifiedLLMClient
 from utils.robot_cmd.robot_command import GroupBotManager, PermissionStatus
 from utils.wechat.contact_manager import ContactManager
 from wechat_ipad import WechatAPIClient
@@ -21,7 +22,6 @@ from .context.image_context import (
     prepare_quote_image_inputs,
 )
 from .context.quote_context import parse_quote_context
-from .core.llm_client import LLMClient
 from .memory.memory_store import MemoryStore
 from .memory.vector_memory import VectorMemoryStore
 from .profile.persona_engine import PersonaEngine
@@ -113,7 +113,7 @@ class AIAutoResponsePlugin(MessagePluginInterface):
         self.vector_memory = VectorMemoryStore(self._config.get("memory", {}) or {})
         self.context_builder = ContextBuilder(int((self._config.get("mode", {}) or {}).get("recent_context_size", 30)))
         self.decision_flow = DecisionFlow()
-        self.llm_client = LLMClient(self._config.get("api", {}) or {})
+        self.llm_client = UnifiedLLMClient(self._config.get("api", {}) or {})
         self.social_memory = SocialMemoryService(self.db_manager, self._config.get("memory", {}) or {})
         self.group_facts = GroupFactsService(self._config.get("memory", {}) or {})
         self.memory_ranker = MemoryRanker(self._config.get("memory", {}) or {})