From 54724ae21e34c0b7042c4a65b840ba46fdf781e4 Mon Sep 17 00:00:00 2001
From: SengokuCola <1026294844@qq.com>
Date: Fri, 30 May 2025 11:04:29 +0800
Subject: [PATCH] =?UTF-8?q?feat=EF=BC=9A=E5=A4=84=E7=90=86=E5=99=A8?=
 =?UTF-8?q?=E5=A4=84=E7=90=86=E6=97=B6=E9=97=B4=E4=B8=8A=E9=99=90=EF=BC=8C?=
 =?UTF-8?q?=E5=A4=84=E7=90=86=E5=B9=B6=E8=A1=8C=E6=A8=A1=E5=BC=8F=EF=BC=8C?=
 =?UTF-8?q?planner=E5=92=8C=E6=80=9D=E8=80=83prompt=EF=BC=8C=E5=BE=AA?=
 =?UTF-8?q?=E7=8E=AF=E8=A7=82=E5=AF=9F=E5=99=A8=E5=A4=A7=E4=BC=98=E5=8C=96?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

feat：增加了处理器处理时间上限，记忆处理并行模式，优化了planner和思考prompt，优化了循环观察器
---
 scripts/070configexe.py                       |  6 +-
 scripts/configexe.toml                        |  9 ++
 .../expressors/exprssion_learner.py           | 23 +++--
 src/chat/focus_chat/heartFC_chat.py           | 94 ++++++++++---------
 .../info_processors/mind_processor.py         | 26 ++---
 .../info_processors/self_processor.py         |  4 +-
 .../info_processors/tool_processor.py         |  8 +-
 .../planners/actions/no_reply_action.py       |  4 +-
 .../focus_chat/planners/modify_actions.py     |  3 -
 src/chat/focus_chat/planners/planner.py       | 25 +++--
 .../observation/actions_observation.py        | 10 ++
 .../observation/chatting_observation.py       | 18 ++++
 .../observation/hfcloop_observation.py        | 65 +++++++++----
 .../heart_flow/observation/observation.py     |  8 ++
 .../observation/structure_observation.py      | 10 ++
 .../observation/working_observation.py        | 10 ++
 src/chat/heart_flow/sub_heartflow.py          |  4 -
 src/config/official_configs.py                | 13 ++-
 src/individuality/expression_style.py         |  4 +-
 .../test_plugin/actions/mute_action.py        |  2 +-
 template/bot_config_template.toml             |  7 +-
 21 files changed, 233 insertions(+), 120 deletions(-)

diff --git a/scripts/070configexe.py b/scripts/070configexe.py
index eeba7d63..be1f56e4 100644
--- a/scripts/070configexe.py
+++ b/scripts/070configexe.py
@@ -403,15 +403,15 @@ class ConfigEditor:
                     # 创建模型名称标签（大字体）
                     model_name = var.get() if var.get() else providers[0]
                     section_translations = {
-                        "model.utils": "工具模型",
-                        "model.utils_small": "小型工具模型",
+                        "model.utils": "麦麦组件模型",
+                        "model.utils_small": "小型麦麦组件模型",
                         "model.memory_summary": "记忆概括模型",
                         "model.vlm": "图像识别模型",
                         "model.embedding": "嵌入模型",
                         "model.normal_chat_1": "普通聊天：主要聊天模型",
                         "model.normal_chat_2": "普通聊天：次要聊天模型",
                         "model.focus_working_memory": "专注模式：工作记忆模型",
-                        "model.focus_chat_mind": "专注模式：聊天规划模型",
+                        "model.focus_chat_mind": "专注模式：聊天思考模型",
                         "model.focus_tool_use": "专注模式：工具调用模型",
                         "model.focus_planner": "专注模式：决策模型",
                         "model.focus_expressor": "专注模式：表达器模型",
diff --git a/scripts/configexe.toml b/scripts/configexe.toml
index 603393f7..a322025f 100644
--- a/scripts/configexe.toml
+++ b/scripts/configexe.toml
@@ -302,6 +302,15 @@ description = "思考的时间间隔（秒），可以有效减少消耗"
 name = "连续回复能力"
 description = "连续回复能力，值越高，麦麦连续回复的概率越高"
 
+[translations.items.parallel_processing]
+name = "并行处理"
+description = "是否并行处理回忆和处理器阶段，可以节省时间"
+
+[translations.items.processor_max_time]
+name = "处理器最大时间"
+description = "处理器最大时间，单位秒，如果超过这个时间，处理器会自动停止"
+
+
 [translations.items.observation_context_size]
 name = "观察上下文大小"
 description = "观察到的最长上下文大小，建议15，太短太长都会导致脑袋尖尖"
diff --git a/src/chat/focus_chat/expressors/exprssion_learner.py b/src/chat/focus_chat/expressors/exprssion_learner.py
index 817a188f..afee74af 100644
--- a/src/chat/focus_chat/expressors/exprssion_learner.py
+++ b/src/chat/focus_chat/expressors/exprssion_learner.py
@@ -19,11 +19,13 @@ def init_prompt() -> None:
     learn_style_prompt = """
 {chat_str}
 
-请从上面这段群聊中概括除了人名为"SELF"之外的人的语言风格，只考虑文字，不要考虑表情包和图片
-不要涉及具体的人名，只考虑语言风格
-语言风格包含特殊内容和情感
-思考有没有特殊的梗，一并总结成语言风格
-总结成如下格式的规律，总结的内容要详细，但具有概括性：
+请从上面这段群聊中概括除了人名为"SELF"之外的人的语言风格
+1. 只考虑文字，不要考虑表情包和图片
+2. 不要涉及具体的人名，只考虑语言风格
+3. 语言风格包含特殊内容和情感
+4. 思考有没有特殊的梗，一并总结成语言风格
+5. 例子仅供参考，请严格根据群聊内容总结!!!
+注意：总结成如下格式的规律，总结的内容要详细，但具有概括性：
 当"xxx"时，可以"xxx", xxx不超过10个字
 
 例如：
@@ -31,7 +33,7 @@ def init_prompt() -> None:
 当"表示讽刺的赞同，不想讲道理"时，使用"对对对"
 当"想说明某个观点，但懒得明说"，使用"懂的都懂"
 
-注意不要总结你自己的发言
+注意不要总结你自己（SELF）的发言
 现在请你概括
 """
     Prompt(learn_style_prompt, "learn_style_prompt")
@@ -40,9 +42,10 @@ def init_prompt() -> None:
 {chat_str}
 
 请从上面这段群聊中概括除了人名为"SELF"之外的人的语法和句法特点，只考虑纯文字，不要考虑表情包和图片
-不要总结【图片】，【动画表情】，[图片]，[动画表情]，不总结 表情符号 at @ 回复 和[回复]
-不要涉及具体的人名，只考虑语法和句法特点,
-语法和句法特点要包括，句子长短（具体字数），有何种语病，如何拆分句子。
+1.不要总结【图片】，【动画表情】，[图片]，[动画表情]，不总结 表情符号 at @ 回复 和[回复]
+2.不要涉及具体的人名，只考虑语法和句法特点,
+3.语法和句法特点要包括，句子长短（具体字数），有何种语病，如何拆分句子。
+4. 例子仅供参考，请严格根据群聊内容总结!!!
 总结成如下格式的规律，总结的内容要简洁，不浮夸：
 当"xxx"时，可以"xxx"
 
@@ -51,7 +54,7 @@ def init_prompt() -> None:
 当"不用详细说明的一般表达"时，使用"非常简洁的句子"的句法
 当"需要单纯简单的确认"时，使用"单字或几个字的肯定(1-2个字)"的句法
 
-注意不要总结你自己的发言
+注意不要总结你自己（SELF）的发言
 现在请你概括
 """
     Prompt(learn_grammar_prompt, "learn_grammar_prompt")
diff --git a/src/chat/focus_chat/heartFC_chat.py b/src/chat/focus_chat/heartFC_chat.py
index 3835b2da..c69dea6b 100644
--- a/src/chat/focus_chat/heartFC_chat.py
+++ b/src/chat/focus_chat/heartFC_chat.py
@@ -44,21 +44,10 @@ PROCESSOR_CLASSES = {
     "ToolProcessor": (ToolProcessor, "tool_use_processor"),
     "WorkingMemoryProcessor": (WorkingMemoryProcessor, "working_memory_processor"),
     "SelfProcessor": (SelfProcessor, "self_identify_processor"),
-    # "ActionProcessor": (ActionProcessor, "action_processor"),  # 这个处理器不需要配置键名，默认启用
 }
 
-
-WAITING_TIME_THRESHOLD = 300  # 等待新消息时间阈值，单位秒
-
-EMOJI_SEND_PRO = 0.3  # 设置一个概率，比如 30% 才真的发
-
-CONSECUTIVE_NO_REPLY_THRESHOLD = 3  # 连续不回复的阈值
-
 logger = get_logger("hfc")  # Logger Name Changed
 
-# 设定处理器超时时间（秒）
-PROCESSOR_TIMEOUT = 30
-
 
 async def _handle_cycle_delay(action_taken_this_cycle: bool, cycle_start_time: float, log_prefix: str):
     """处理循环延迟"""
@@ -150,7 +139,7 @@ class HeartFChatting:
         # 添加循环信息管理相关的属性
         self._cycle_counter = 0
         self._cycle_history: Deque[CycleDetail] = deque(maxlen=10)  # 保留最近10个循环的信息
-        self._current_cycle: Optional[CycleDetail] = None
+        self._current_cycle_detail: Optional[CycleDetail] = None
         self._shutting_down: bool = False  # 关闭标志位
 
         # 存储回调函数
@@ -262,12 +251,12 @@ class HeartFChatting:
         try:
             exception = task.exception()
             if exception:
-                logger.error(f"{self.log_prefix} HeartFChatting: 麦麦脱离了聊天(异常): {exception}")
+                logger.error(f"{self.log_prefix} HeartFChatting: 脱离了聊天(异常): {exception}")
                 logger.error(traceback.format_exc())  # Log full traceback for exceptions
             else:
-                logger.info(f"{self.log_prefix} HeartFChatting: 麦麦脱离了聊天 (外部停止)")
+                logger.info(f"{self.log_prefix} HeartFChatting: 脱离了聊天 (外部停止)")
         except asyncio.CancelledError:
-            logger.info(f"{self.log_prefix} HeartFChatting: 麦麦脱离了聊天(任务取消)")
+            logger.info(f"{self.log_prefix} HeartFChatting: 脱离了聊天(任务取消)")
         finally:
             self._loop_active = False
             self._loop_task = None
@@ -286,7 +275,8 @@ class HeartFChatting:
 
                 # 创建新的循环信息
                 self._cycle_counter += 1
-                self._current_cycle = CycleDetail(self._cycle_counter)
+                self._current_cycle_detail = CycleDetail(self._cycle_counter)
+                self._current_cycle_detail.prefix = self.log_prefix
 
                 # 初始化周期状态
                 cycle_timers = {}
@@ -295,13 +285,12 @@ class HeartFChatting:
                 # 执行规划和处理阶段
                 async with self._get_cycle_context():
                     thinking_id = "tid" + str(round(time.time(), 2))
-                    self._current_cycle.set_thinking_id(thinking_id)
+                    self._current_cycle_detail.set_thinking_id(thinking_id)
                     # 主循环：思考->决策->执行
                     async with global_prompt_manager.async_message_scope(self.chat_stream.context.get_template_name()):
                         logger.debug(f"模板 {self.chat_stream.context.get_template_name()}")
                         loop_info = await self._observe_process_plan_action_loop(cycle_timers, thinking_id)
 
-                        print(loop_info["loop_action_info"]["command"])
                         if loop_info["loop_action_info"]["command"] == "stop_focus_chat":
                             logger.info(f"{self.log_prefix} 麦麦决定停止专注聊天")
                             # 如果设置了回调函数，则调用它
@@ -314,10 +303,10 @@ class HeartFChatting:
                                     logger.error(traceback.format_exc())
                             break
 
-                    self._current_cycle.set_loop_info(loop_info)
+                    self._current_cycle_detail.set_loop_info(loop_info)
 
-                    self.hfcloop_observation.add_loop_info(self._current_cycle)
-                    self._current_cycle.timers = cycle_timers
+                    self.hfcloop_observation.add_loop_info(self._current_cycle_detail)
+                    self._current_cycle_detail.timers = cycle_timers
 
                     # 防止循环过快消耗资源
                     await _handle_cycle_delay(
@@ -325,8 +314,8 @@ class HeartFChatting:
                     )
 
                 # 完成当前循环并保存历史
-                self._current_cycle.complete_cycle()
-                self._cycle_history.append(self._current_cycle)
+                self._current_cycle_detail.complete_cycle()
+                self._cycle_history.append(self._current_cycle_detail)
 
                 # 记录循环信息和计时器结果
                 timer_strings = []
@@ -335,7 +324,7 @@ class HeartFChatting:
                     timer_strings.append(f"{name}: {formatted_time}")
 
                 # 新增：输出每个处理器的耗时
-                processor_time_costs = self._current_cycle.loop_processor_info.get("processor_time_costs", {})
+                processor_time_costs = self._current_cycle_detail.loop_processor_info.get("processor_time_costs", {})
                 processor_time_strings = []
                 for pname, ptime in processor_time_costs.items():
                     formatted_ptime = f"{ptime * 1000:.2f}毫秒" if ptime < 1 else f"{ptime:.2f}秒"
@@ -345,9 +334,9 @@ class HeartFChatting:
                 )
 
                 logger.info(
-                    f"{self.log_prefix} 第{self._current_cycle.cycle_id}次思考,"
-                    f"耗时: {self._current_cycle.end_time - self._current_cycle.start_time:.1f}秒, "
-                    f"动作: {self._current_cycle.loop_plan_info['action_result']['action_type']}"
+                    f"{self.log_prefix} 第{self._current_cycle_detail.cycle_id}次思考,"
+                    f"耗时: {self._current_cycle_detail.end_time - self._current_cycle_detail.start_time:.1f}秒, "
+                    f"动作: {self._current_cycle_detail.loop_plan_info['action_result']['action_type']}"
                     + (f"\n详情: {'; '.join(timer_strings)}" if timer_strings else "")
                     + processor_time_log
                 )
@@ -384,7 +373,7 @@ class HeartFChatting:
                 self._processing_lock.release()
 
     async def _process_processors(
-        self, observations: List[Observation], running_memorys: List[Dict[str, Any]], cycle_timers: dict
+        self, observations: List[Observation], running_memorys: List[Dict[str, Any]]
     ) -> tuple[List[InfoBase], Dict[str, float]]:
         # 记录并行任务开始时间
         parallel_start_time = time.time()
@@ -400,7 +389,7 @@ class HeartFChatting:
             async def run_with_timeout(proc=processor):
                 return await asyncio.wait_for(
                     proc.process_info(observations=observations, running_memorys=running_memorys),
-                    timeout=PROCESSOR_TIMEOUT,
+                    timeout=global_config.focus_chat.processor_max_time,
                 )
 
             task = asyncio.create_task(run_with_timeout())
@@ -429,8 +418,8 @@ class HeartFChatting:
                     # 记录耗时
                     processor_time_costs[processor_name] = duration_since_parallel_start
                 except asyncio.TimeoutError:
-                    logger.info(f"{self.log_prefix} 处理器 {processor_name} 超时（>{PROCESSOR_TIMEOUT}s），已跳过")
-                    processor_time_costs[processor_name] = PROCESSOR_TIMEOUT
+                    logger.info(f"{self.log_prefix} 处理器 {processor_name} 超时（>{global_config.focus_chat.processor_max_time}s），已跳过")
+                    processor_time_costs[processor_name] = global_config.focus_chat.processor_max_time
                 except Exception as e:
                     logger.error(
                         f"{self.log_prefix} 处理器 {processor_name} 执行失败，耗时 (自并行开始): {duration_since_parallel_start:.2f}秒. 错误: {e}",
@@ -473,28 +462,42 @@ class HeartFChatting:
                 }
 
                 self.all_observations = observations
-
-            with Timer("回忆", cycle_timers):
-                running_memorys = await self.memory_activator.activate_memory(observations)
-
+                
             with Timer("调整动作", cycle_timers):
                 # 处理特殊的观察
-                await self.action_modifier.modify_actions(observations=observations, running_memorys=running_memorys)
+                await self.action_modifier.modify_actions(observations=observations)
                 await self.action_observation.observe()
                 observations.append(self.action_observation)
 
-            with Timer("执行 信息处理器", cycle_timers):
-                all_plan_info, processor_time_costs = await self._process_processors(
-                    observations, running_memorys, cycle_timers
-                )
+            # 根据配置决定是否并行执行回忆和处理器阶段
+            # print(global_config.focus_chat.parallel_processing)
+            if global_config.focus_chat.parallel_processing:
+                # 并行执行回忆和处理器阶段
+                with Timer("并行回忆和处理", cycle_timers):
+                    memory_task = asyncio.create_task(self.memory_activator.activate_memory(observations))
+                    processor_task = asyncio.create_task(self._process_processors(observations, []))
+                    
+                    # 等待两个任务完成
+                    running_memorys, (all_plan_info, processor_time_costs) = await asyncio.gather(memory_task, processor_task)
+            else:
+                # 串行执行
+                with Timer("回忆", cycle_timers):
+                    running_memorys = await self.memory_activator.activate_memory(observations)
+
+                with Timer("执行 信息处理器", cycle_timers):
+                    all_plan_info, processor_time_costs = await self._process_processors(
+                        observations, running_memorys
+                    )
+
+            loop_processor_info = {
+                "all_plan_info": all_plan_info,
+                "processor_time_costs": processor_time_costs,
+            }
+
 
-                loop_processor_info = {
-                    "all_plan_info": all_plan_info,
-                    "processor_time_costs": processor_time_costs,
-                }
 
             with Timer("规划器", cycle_timers):
-                plan_result = await self.action_planner.plan(all_plan_info, cycle_timers)
+                plan_result = await self.action_planner.plan(all_plan_info, running_memorys)
 
                 loop_plan_info = {
                     "action_result": plan_result.get("action_result", {}),
@@ -526,6 +529,7 @@ class HeartFChatting:
                     "action_taken": success,
                     "reply_text": reply_text,
                     "command": command,
+                    "taken_time": time.time(),
                 }
 
             loop_info = {
diff --git a/src/chat/focus_chat/info_processors/mind_processor.py b/src/chat/focus_chat/info_processors/mind_processor.py
index 12671169..a865f769 100644
--- a/src/chat/focus_chat/info_processors/mind_processor.py
+++ b/src/chat/focus_chat/info_processors/mind_processor.py
@@ -24,9 +24,7 @@ logger = get_logger("processor")
 def init_prompt():
     group_prompt = """
 你的名字是{bot_name}
-{memory_str}
-{extra_info}
-{relation_prompt}
+{memory_str}{extra_info}{relation_prompt}
 {cycle_info_block}
 现在是{time_now}，你正在上网，和qq群里的网友们聊天，以下是正在进行的聊天内容：
 {chat_observe_info}
@@ -46,15 +44,11 @@ def init_prompt():
 
     private_prompt = """
 你的名字是{bot_name}
-{memory_str}
-{extra_info}
-{relation_prompt}
+{memory_str}{extra_info}{relation_prompt}
 {cycle_info_block}
 现在是{time_now}，你正在上网，和qq群里的网友们聊天，以下是正在进行的聊天内容：
 {chat_observe_info}
-
 {action_observe_info}
-
 以下是你之前对聊天的观察和规划，你的名字是{bot_name}：
 {last_mind}
 
@@ -155,14 +149,14 @@ class MindProcessor(BaseProcessor):
 
         # ---------- 0. 更新和清理 structured_info ----------
         if self.structured_info:
-            updated_info = []
-            for item in self.structured_info:
-                item["ttl"] -= 1
-                if item["ttl"] > 0:
-                    updated_info.append(item)
-                else:
-                    logger.debug(f"{self.log_prefix} 移除过期的 structured_info 项: {item['id']}")
-            self.structured_info = updated_info
+            # updated_info = []
+            # for item in self.structured_info:
+            #     item["ttl"] -= 1
+            #     if item["ttl"] > 0:
+            #         updated_info.append(item)
+            #     else:
+            #         logger.debug(f"{self.log_prefix} 移除过期的 structured_info 项: {item['id']}")
+            # self.structured_info = updated_info
             self._update_structured_info_str()
         logger.debug(
             f"{self.log_prefix} 当前完整的 structured_info: {safe_json_dumps(self.structured_info, ensure_ascii=False)}"
diff --git a/src/chat/focus_chat/info_processors/self_processor.py b/src/chat/focus_chat/info_processors/self_processor.py
index d489efa2..72562ad5 100644
--- a/src/chat/focus_chat/info_processors/self_processor.py
+++ b/src/chat/focus_chat/info_processors/self_processor.py
@@ -25,9 +25,7 @@ def init_prompt():
 你是一个AI，但是你伪装成了一个人类，你的人格是，{prompt_personality}
 {indentify_block}
 
-{relation_prompt}
-{memory_str}
-
+{relation_prompt}{memory_str}
 现在是{time_now}，你正在上网，和qq群里的网友们聊天，以下是正在进行的聊天内容：
 {chat_observe_info}
 
diff --git a/src/chat/focus_chat/info_processors/tool_processor.py b/src/chat/focus_chat/info_processors/tool_processor.py
index 3c0dc116..46c2657d 100644
--- a/src/chat/focus_chat/info_processors/tool_processor.py
+++ b/src/chat/focus_chat/info_processors/tool_processor.py
@@ -23,10 +23,7 @@ def init_prompt():
     # 添加工具执行器提示词
     tool_executor_prompt = """
 你是一个专门执行工具的助手。你的名字是{bot_name}。现在是{time_now}。
-
-你当前的额外信息：
 {memory_str}
-
 群里正在进行的聊天内容：
 {chat_observe_info}
 
@@ -165,8 +162,9 @@ class ToolProcessor(BaseProcessor):
         logger.debug(f"开始执行工具调用{prompt}")
         response, _, tool_calls = await self.llm_model.generate_response_tool_async(prompt=prompt, tools=tools)
 
-        logger.debug(f"获取到工具原始输出:\n{tool_calls}")
-        # 处理工具调用和结果收集，类似于SubMind中的逻辑
+        if tool_calls:
+            logger.debug(f"获取到工具原始输出:\n{tool_calls}")
+            # 处理工具调用和结果收集，类似于SubMind中的逻辑
         new_structured_items = []
         used_tools = []  # 记录使用了哪些工具
 
diff --git a/src/chat/focus_chat/planners/actions/no_reply_action.py b/src/chat/focus_chat/planners/actions/no_reply_action.py
index 1b21a8ce..120ebe98 100644
--- a/src/chat/focus_chat/planners/actions/no_reply_action.py
+++ b/src/chat/focus_chat/planners/actions/no_reply_action.py
@@ -26,8 +26,8 @@ class NoReplyAction(BaseAction):
     action_parameters = {}
     action_require = [
         "话题无关/无聊/不感兴趣/不懂",
-        "最后一条消息是你自己发的且无人回应你",
-        "你发送了太多消息，且无人回复",
+        "聊天记录中最新一条消息是你自己发的且无人回应你",
+        "你连续发送了太多消息，且无人回复",
     ]
     default = True
 
diff --git a/src/chat/focus_chat/planners/modify_actions.py b/src/chat/focus_chat/planners/modify_actions.py
index 1648e6cf..731fe5f9 100644
--- a/src/chat/focus_chat/planners/modify_actions.py
+++ b/src/chat/focus_chat/planners/modify_actions.py
@@ -28,11 +28,8 @@ class ActionModifier:
     async def modify_actions(
         self,
         observations: Optional[List[Observation]] = None,
-        running_memorys: Optional[List[Dict]] = None,
         **kwargs: Any,
     ):
-        # print(f"observations: {observations}")
-        # processed_infos = []
 
         # 处理Observation对象
         if observations:
diff --git a/src/chat/focus_chat/planners/planner.py b/src/chat/focus_chat/planners/planner.py
index 528f7a7f..12d8209e 100644
--- a/src/chat/focus_chat/planners/planner.py
+++ b/src/chat/focus_chat/planners/planner.py
@@ -26,9 +26,8 @@ def init_prompt():
         """
 你的自我认知是：
 {self_info_block}
-
 {extra_info_block}
-
+{memory_str}
 你需要基于以下信息决定如何参与对话
 这些信息可能会有冲突，请你整合这些信息，并选择一个最合适的action：
 {chat_content_block}
@@ -49,7 +48,7 @@ def init_prompt():
 请你以下面格式输出你选择的action：
 {{
     "action": "action_name",
-    "reasoning": "你的决策理由",
+    "reasoning": "说明你做出该action的原因",
     "参数1": "参数1的值",
     "参数2": "参数2的值",
     "参数3": "参数3的值",
@@ -84,13 +83,13 @@ class ActionPlanner:
 
         self.action_manager = action_manager
 
-    async def plan(self, all_plan_info: List[InfoBase], cycle_timers: dict) -> Dict[str, Any]:
+    async def plan(self, all_plan_info: List[InfoBase], running_memorys: List[Dict[str, Any]]) -> Dict[str, Any]:
         """
         规划器 (Planner): 使用LLM根据上下文决定做出什么动作。
 
         参数:
             all_plan_info: 所有计划信息
-            cycle_timers: 计时器字典
+            running_memorys: 回忆信息
         """
 
         action = "no_reply"  # 默认动作
@@ -169,6 +168,7 @@ class ActionPlanner:
                 current_available_actions=current_available_actions,  # <-- Pass determined actions
                 cycle_info=cycle_info,  # <-- Pass cycle info
                 extra_info=extra_info,
+                running_memorys=running_memorys,
             )
 
             # --- 调用 LLM (普通文本生成) ---
@@ -259,10 +259,22 @@ class ActionPlanner:
         current_available_actions: Dict[str, ActionInfo],
         cycle_info: Optional[str],
         extra_info: list[str],
+        running_memorys: List[Dict[str, Any]],
     ) -> str:
         """构建 Planner LLM 的提示词 (获取模板并填充数据)"""
         try:
-            # --- Determine chat context ---
+            
+            memory_str = ""
+            if global_config.focus_chat.parallel_processing:
+                memory_str = ""
+                if running_memorys:
+                    memory_str = "以下是当前在聊天中，你回忆起的记忆：\n"
+                    for running_memory in running_memorys:
+                        memory_str += f"{running_memory['topic']}: {running_memory['content']}\n"
+            
+            
+            
+            
             chat_context_description = "你现在正在一个群聊中"
             chat_target_name = None  # Only relevant for private
             if not is_group_chat and chat_target_info:
@@ -324,6 +336,7 @@ class ActionPlanner:
             planner_prompt_template = await global_prompt_manager.get_prompt_async("planner_prompt")
             prompt = planner_prompt_template.format(
                 self_info_block=self_info_block,
+                memory_str=memory_str,
                 # bot_name=global_config.bot.nickname,
                 prompt_personality=personality_block,
                 chat_context_description=chat_context_description,
diff --git a/src/chat/heart_flow/observation/actions_observation.py b/src/chat/heart_flow/observation/actions_observation.py
index 8310a17b..6f0cd81c 100644
--- a/src/chat/heart_flow/observation/actions_observation.py
+++ b/src/chat/heart_flow/observation/actions_observation.py
@@ -34,3 +34,13 @@ class ActionObservation:
         action_info_block += "\n注意，除了上面动作选项之外，你在群聊里不能做其他任何事情，这是你能力的边界\n"
 
         self.observe_info = action_info_block
+
+    def to_dict(self) -> dict:
+        """将观察对象转换为可序列化的字典"""
+        return {
+            "observe_info": self.observe_info,
+            "observe_id": self.observe_id,
+            "last_observe_time": self.last_observe_time,
+            "all_actions": self.all_actions,
+            "all_using_actions": self.all_using_actions
+        }
diff --git a/src/chat/heart_flow/observation/chatting_observation.py b/src/chat/heart_flow/observation/chatting_observation.py
index e5c5069e..7e9e562d 100644
--- a/src/chat/heart_flow/observation/chatting_observation.py
+++ b/src/chat/heart_flow/observation/chatting_observation.py
@@ -66,6 +66,24 @@ class ChattingObservation(Observation):
         self.oldest_messages_str = ""
         self.compressor_prompt = ""
 
+    def to_dict(self) -> dict:
+        """将观察对象转换为可序列化的字典"""
+        return {
+            "chat_id": self.chat_id,
+            "platform": self.platform,
+            "is_group_chat": self.is_group_chat,
+            "chat_target_info": self.chat_target_info,
+            "talking_message_str": self.talking_message_str,
+            "talking_message_str_truncate": self.talking_message_str_truncate,
+            "name": self.name,
+            "nick_name": self.nick_name,
+            "mid_memory_info": self.mid_memory_info,
+            "person_list": self.person_list,
+            "oldest_messages_str": self.oldest_messages_str,
+            "compressor_prompt": self.compressor_prompt,
+            "last_observe_time": self.last_observe_time
+        }
+
     async def initialize(self):
         self.is_group_chat, self.chat_target_info = await get_chat_type_and_target_info(self.chat_id)
         logger.debug(f"初始化observation: self.is_group_chat: {self.is_group_chat}")
diff --git a/src/chat/heart_flow/observation/hfcloop_observation.py b/src/chat/heart_flow/observation/hfcloop_observation.py
index 7c782bfd..03341bbf 100644
--- a/src/chat/heart_flow/observation/hfcloop_observation.py
+++ b/src/chat/heart_flow/observation/hfcloop_observation.py
@@ -27,40 +27,62 @@ class HFCloopObservation:
         recent_active_cycles: List[CycleDetail] = []
         for cycle in reversed(self.history_loop):
             # 只关心实际执行了动作的循环
-            action_taken = cycle.loop_action_info["action_taken"]
-            if action_taken:
-                recent_active_cycles.append(cycle)
-                if len(recent_active_cycles) == 5:
-                    break
+            # action_taken = cycle.loop_action_info["action_taken"]
+            # if action_taken:
+            recent_active_cycles.append(cycle)
+            if len(recent_active_cycles) == 5:
+                break
 
         cycle_info_block = ""
+        action_detailed_str = ""
         consecutive_text_replies = 0
         responses_for_prompt = []
 
         # 检查这最近的活动循环中有多少是连续的文本回复 (从最近的开始看)
         for cycle in recent_active_cycles:
             action_type = cycle.loop_plan_info["action_result"]["action_type"]
+            action_reasoning = cycle.loop_plan_info["action_result"]["reasoning"]
+            is_taken = cycle.loop_action_info["action_taken"]
+            action_taken_time = cycle.loop_action_info["taken_time"]
+            action_taken_time_str = datetime.fromtimestamp(action_taken_time).strftime("%H:%M:%S")
+            # print(action_type)
+            # print(action_reasoning)
+            # print(is_taken)
+            # print(action_taken_time_str)
+            # print("--------------------------------")
             if action_type == "reply":
                 consecutive_text_replies += 1
                 response_text = cycle.loop_plan_info["action_result"]["action_data"].get("text", "[空回复]")
                 responses_for_prompt.append(response_text)
+                
+                if is_taken:
+                    action_detailed_str += f"{action_taken_time_str}时，你选择回复(action:{action_type},内容是:'{response_text}')。你选择这个action的原因是:{action_reasoning}\n"
+                else:
+                    action_detailed_str += f"{action_taken_time_str}时，你选择回复(action:{action_type},内容是:'{response_text}')，但是动作失败了。你选择这个action的原因是:{action_reasoning}\n"
+            elif action_type == "no_reply":
+                action_detailed_str += f"{action_taken_time_str}时，你选择不回复(action:{action_type})，你选择了沉默，原因是:{action_reasoning}\n"
             else:
-                break
-
+                if is_taken:
+                    action_detailed_str += f"{action_taken_time_str}时，你选择执行了(action:{action_type})，你选择这个action的原因是:{action_reasoning}\n"
+                else:
+                    action_detailed_str += f"{action_taken_time_str}时，你选择执行了(action:{action_type})，但是动作失败了。你选择这个action的原因是:{action_reasoning}\n"
+                
+        if action_detailed_str:
+            cycle_info_block = f"\n你最近做的事：\n{action_detailed_str}\n"
+        else:
+            cycle_info_block = "\n"
+        
         # 根据连续文本回复的数量构建提示信息
-        # 注意: responses_for_prompt 列表是从最近到最远排序的
         if consecutive_text_replies >= 3:  # 如果最近的三个活动都是文本回复
             cycle_info_block = f'你已经连续回复了三条消息（最近: "{responses_for_prompt[0]}"，第二近: "{responses_for_prompt[1]}"，第三近: "{responses_for_prompt[2]}"）。你回复的有点多了，请注意'
         elif consecutive_text_replies == 2:  # 如果最近的两个活动是文本回复
             cycle_info_block = f'你已经连续回复了两条消息（最近: "{responses_for_prompt[0]}"，第二近: "{responses_for_prompt[1]}"），请注意'
-        elif consecutive_text_replies == 1:  # 如果最近的一个活动是文本回复
-            cycle_info_block = f'你刚刚已经回复一条消息（内容: "{responses_for_prompt[0]}"）'
 
         # 包装提示块，增加可读性，即使没有连续回复也给个标记
-        if cycle_info_block:
-            cycle_info_block = f"\n你最近的回复\n{cycle_info_block}\n"
-        else:
-            cycle_info_block = "\n"
+        # if cycle_info_block:
+        #     cycle_info_block = f"\n你最近的回复\n{cycle_info_block}\n"
+        # else:
+        #     cycle_info_block = "\n"
 
         # 获取history_loop中最新添加的
         if self.history_loop:
@@ -70,10 +92,19 @@ class HFCloopObservation:
             if start_time is not None and end_time is not None:
                 time_diff = int(end_time - start_time)
                 if time_diff > 60:
-                    cycle_info_block += f"\n距离你上一次阅读消息已经过去了{time_diff / 60}分钟\n"
+                    cycle_info_block += f"距离你上一次阅读消息并思考和规划，已经过去了{int(time_diff / 60)}分钟\n"
                 else:
-                    cycle_info_block += f"\n距离你上一次阅读消息已经过去了{time_diff}秒\n"
+                    cycle_info_block += f"距离你上一次阅读消息并思考和规划，已经过去了{time_diff}秒\n"
             else:
-                cycle_info_block += "\n你还没看过消息\n"
+                cycle_info_block += "你还没看过消息\n"
 
         self.observe_info = cycle_info_block
+
+    def to_dict(self) -> dict:
+        """将观察对象转换为可序列化的字典"""
+        return {
+            "observe_info": self.observe_info,
+            "observe_id": self.observe_id,
+            "last_observe_time": self.last_observe_time,
+            "history_loop": [cycle.to_dict() for cycle in self.history_loop]
+        }
diff --git a/src/chat/heart_flow/observation/observation.py b/src/chat/heart_flow/observation/observation.py
index 97e254fc..5c8b5fda 100644
--- a/src/chat/heart_flow/observation/observation.py
+++ b/src/chat/heart_flow/observation/observation.py
@@ -13,5 +13,13 @@ class Observation:
         self.observe_id = observe_id
         self.last_observe_time = datetime.now().timestamp()  # 初始化为当前时间
 
+    def to_dict(self) -> dict:
+        """将观察对象转换为可序列化的字典"""
+        return {
+            "observe_info": self.observe_info,
+            "observe_id": self.observe_id,
+            "last_observe_time": self.last_observe_time
+        }
+
     async def observe(self):
         pass
diff --git a/src/chat/heart_flow/observation/structure_observation.py b/src/chat/heart_flow/observation/structure_observation.py
index 73b5bf75..6e670f5e 100644
--- a/src/chat/heart_flow/observation/structure_observation.py
+++ b/src/chat/heart_flow/observation/structure_observation.py
@@ -15,6 +15,16 @@ class StructureObservation:
         self.history_loop = []
         self.structured_info = []
 
+    def to_dict(self) -> dict:
+        """将观察对象转换为可序列化的字典"""
+        return {
+            "observe_info": self.observe_info,
+            "observe_id": self.observe_id,
+            "last_observe_time": self.last_observe_time,
+            "history_loop": self.history_loop,
+            "structured_info": self.structured_info
+        }
+
     def get_observe_info(self):
         return self.structured_info
 
diff --git a/src/chat/heart_flow/observation/working_observation.py b/src/chat/heart_flow/observation/working_observation.py
index 7013c3a2..3cab4a37 100644
--- a/src/chat/heart_flow/observation/working_observation.py
+++ b/src/chat/heart_flow/observation/working_observation.py
@@ -32,3 +32,13 @@ class WorkingMemoryObservation:
 
     async def observe(self):
         pass
+
+    def to_dict(self) -> dict:
+        """将观察对象转换为可序列化的字典"""
+        return {
+            "observe_info": self.observe_info,
+            "observe_id": self.observe_id,
+            "last_observe_time": self.last_observe_time,
+            "working_memory": self.working_memory.to_dict() if hasattr(self.working_memory, 'to_dict') else str(self.working_memory),
+            "retrieved_working_memory": [item.to_dict() if hasattr(item, 'to_dict') else str(item) for item in self.retrieved_working_memory]
+        }
diff --git a/src/chat/heart_flow/sub_heartflow.py b/src/chat/heart_flow/sub_heartflow.py
index 435a7b61..984b3638 100644
--- a/src/chat/heart_flow/sub_heartflow.py
+++ b/src/chat/heart_flow/sub_heartflow.py
@@ -278,10 +278,6 @@ class SubHeartflow:
             self.update_last_chat_state_time()
             self.history_chat_state.append((current_state, self.chat_state_last_time))
 
-            # logger.info(
-            # f"{log_prefix} 麦麦的聊天状态从 {current_state.value} （持续了 {int(self.chat_state_last_time)} 秒） 变更为 {new_state.value}"
-            # )
-
             self.chat_state.chat_status = new_state
             self.chat_state_last_time = 0
             self.chat_state_changed_time = time.time()
diff --git a/src/config/official_configs.py b/src/config/official_configs.py
index af729db3..1a1469fe 100644
--- a/src/config/official_configs.py
+++ b/src/config/official_configs.py
@@ -143,9 +143,15 @@ class FocusChatConfig(ConfigBase):
 
     think_interval: float = 1
     """思考间隔（秒）"""
-    
+
     consecutive_replies: float = 1
     """连续回复能力，值越高，麦麦连续回复的概率越高"""
+    
+    parallel_processing: bool = False
+    """是否允许处理器阶段和回忆阶段并行执行"""
+    
+    processor_max_time: int = 25
+    """处理器最大时间，单位秒，如果超过这个时间，处理器会自动停止"""
 
 
 @dataclass
@@ -160,6 +166,11 @@ class FocusChatProcessorConfig(ConfigBase):
 
     working_memory_processor: bool = True
     """是否启用工作记忆处理器"""
+    
+    lite_chat_mind_processor: bool = False
+    """是否启用轻量级聊天思维处理器，可以节省token消耗和时间"""
+
+
 
 
 @dataclass
diff --git a/src/individuality/expression_style.py b/src/individuality/expression_style.py
index f4eed60b..264110b7 100644
--- a/src/individuality/expression_style.py
+++ b/src/individuality/expression_style.py
@@ -14,13 +14,13 @@ def init_prompt() -> None:
     personality_expression_prompt = """
 {personality}
 
-请从以上人设中总结出这个角色可能的语言风格
+请从以上人设中总结出这个角色可能的语言风格，你必须严格根据人设引申，不要输出例子
 思考回复的特殊内容和情感
 思考有没有特殊的梗，一并总结成语言风格
 总结成如下格式的规律，总结的内容要详细，但具有概括性：
 当"xxx"时，可以"xxx", xxx不超过10个字
 
-例如：
+例如（不要输出例子）：
 当"表示十分惊叹"时，使用"我嘞个xxxx"
 当"表示讽刺的赞同，不想讲道理"时，使用"对对对"
 当"想说明某个观点，但懒得明说"，使用"懂的都懂"
diff --git a/src/plugins/test_plugin/actions/mute_action.py b/src/plugins/test_plugin/actions/mute_action.py
index 21a8e057..5bae71e0 100644
--- a/src/plugins/test_plugin/actions/mute_action.py
+++ b/src/plugins/test_plugin/actions/mute_action.py
@@ -23,7 +23,7 @@ class MuteAction(PluginAction):
         "当有人戳你两次以上时，防止刷屏，禁言他，必须牢记",
         "当你想回避某个话题时使用",
     ]
-    default = True  # 默认动作，是否手动添加到使用集
+    default = False  # 默认动作，是否手动添加到使用集
     associated_types = ["command", "text"]
     # associated_types = ["text"]
 
diff --git a/template/bot_config_template.toml b/template/bot_config_template.toml
index 95d329cd..439a6e12 100644
--- a/template/bot_config_template.toml
+++ b/template/bot_config_template.toml
@@ -41,7 +41,7 @@ identity_detail = [
 [expression]
 # 表达方式
 expression_style = "描述麦麦说话的表达风格，表达习惯，例如：(回复尽量简短一些。可以参考贴吧，知乎和微博的回复风格，回复不要浮夸，不要用夸张修辞，平淡一些。不要有额外的符号，尽量简单简短)"
-enable_expression_learning = true # 是否启用表达学习，麦麦会学习人类说话风格
+enable_expression_learning = false # 是否启用表达学习，麦麦会学习人类说话风格
 learning_interval = 600 # 学习间隔 单位秒
 
 [relationship]
@@ -94,8 +94,11 @@ talk_frequency_down_groups = []  #降低回复频率的群号码
 think_interval = 3 # 思考间隔 单位秒，可以有效减少消耗
 consecutive_replies = 1 # 连续回复能力，值越高，麦麦连续回复的概率越高
 
+parallel_processing = true # 是否并行处理回忆和处理器阶段，可以节省时间
 
-observation_context_size = 16 # 观察到的最长上下文大小,建议15，太短太长都会导致脑袋尖尖
+processor_max_time = 25 # 处理器最大时间，单位秒，如果超过这个时间，处理器会自动停止
+
+observation_context_size = 16 # 观察到的最长上下文大小
 compressed_length = 8 # 不能大于observation_context_size,心流上下文压缩的最短压缩长度，超过心流观察到的上下文长度，会压缩，最短压缩长度为5
 compress_length_limit = 4 #最多压缩份数，超过该数值的压缩上下文会被删除