feat：支持模型缓存和相关配置

src/common/data_models/llm_service_data_models.py

@@ -68,6 +68,8 @@ class LLMResponseResult(BaseDataModel):
     prompt_tokens: int = 0
     completion_tokens: int = 0
     total_tokens: int = 0
+    prompt_cache_hit_tokens: int = 0
+    prompt_cache_miss_tokens: int = 0
 
 
 @dataclass(slots=True)
@@ -125,6 +127,8 @@ class LLMServiceResult(BaseDataModel):
             "prompt_tokens": self.completion.prompt_tokens,
             "completion_tokens": self.completion.completion_tokens,
             "total_tokens": self.completion.total_tokens,
+            "prompt_cache_hit_tokens": self.completion.prompt_cache_hit_tokens,
+            "prompt_cache_miss_tokens": self.completion.prompt_cache_miss_tokens,
         }
         if self.completion.tool_calls is not None:
             payload["tool_calls"] = [

src/common/logger_color_and_mapping.py

@@ -32,6 +32,7 @@ MODULE_COLORS: Dict[str, Tuple[str, Optional[str], bool]] = {
     "remote": ("#6c6c6c", None, False),  # 深灰色，更不显眼
     "planner": ("#008080", None, False),
     "maisaka_reasoning_engine": ("#008080", None, False),
+    "maisaka_chat_loop": ("#0087ff", None, False),
     "maisaka_runtime": ("#ff5fff", None, False),
     "relation": ("#af87af", None, False),  # 柔和的紫色，不刺眼
     # 聊天相关模块