feat:执行状态单例状态bug修复server.py遗漏提交

合并远程web分支更新

backend/AgentCoord/LLMAPI/LLMAPI.py

@@ -112,8 +112,17 @@ def _call_with_custom_config(messages: list[dict], stream: bool, model_config: d
                 timeout=180
 
             )
+            # 检查响应是否有效
+            if not response.choices or len(response.choices) == 0:
+                raise Exception(f"API returned empty response for model {api_model}")
+            if not response.choices[0] or not response.choices[0].message:
+                raise Exception(f"API returned invalid response format for model {api_model}")
+
             full_reply_content = response.choices[0].message.content
-            print(colored(full_reply_content, "blue", "on_white"), end="")
+            if full_reply_content is None:
+                raise Exception(f"API returned None content for model {api_model}")
+
+            #print(colored(full_reply_content, "blue", "on_white"), end="")
             return full_reply_content
     except Exception as e:
         print_colored(f"Custom API error for model {api_model} :{str(e)}","red")
@@ -138,15 +147,21 @@ async def _achat_completion_stream_custom(messages:list[dict], temp_async_client
             async for chunk in response:
                 collected_chunks.append(chunk)
                 choices = chunk.choices
-                if len(choices) > 0:
+                if len(choices) > 0 and choices[0] is not None:
-                    chunk_message = chunk.choices[0].delta
+                    chunk_message = choices[0].delta
+                    if chunk_message is not None:
                         collected_messages.append(chunk_message)
-                    if chunk_message.content:
+            #             if chunk_message.content:
-                        print(colored(chunk_message.content, "blue", "on_white"), end="")
+            #                 print(colored(chunk_message.content, "blue", "on_white"), end="")
-            print()
+            # print()
             full_reply_content = "".join(
                 [m.content or "" for m in collected_messages if m is not None]
             )
+
+            # 检查最终结果是否为空
+            if not full_reply_content or full_reply_content.strip() == "":
+                raise Exception(f"Stream API returned empty content for model {api_model}")
+
             return full_reply_content
         except httpx.RemoteProtocolError as e:
             if attempt < max_retries - 1:
@@ -184,9 +199,18 @@ async def _achat_completion_stream_groq(messages: list[dict]) -> str:
             else:
                 raise Exception("failed")
 
+    # 检查响应是否有效
+    if not response.choices or len(response.choices) == 0:
+        raise Exception("Groq API returned empty response")
+    if not response.choices[0] or not response.choices[0].message:
+        raise Exception("Groq API returned invalid response format")
+
     full_reply_content = response.choices[0].message.content
-    print(colored(full_reply_content, "blue", "on_white"), end="")
+    if full_reply_content is None:
-    print()
+        raise Exception("Groq API returned None content")
+
+    # print(colored(full_reply_content, "blue", "on_white"), end="")
+    # print()
     return full_reply_content
 
 
@@ -217,9 +241,18 @@ async def _achat_completion_stream_mixtral(messages: list[dict]) -> str:
             else:
                 raise Exception("failed")
 
+    # 检查响应是否有效
+    if not stream.choices or len(stream.choices) == 0:
+        raise Exception("Mistral API returned empty response")
+    if not stream.choices[0] or not stream.choices[0].message:
+        raise Exception("Mistral API returned invalid response format")
+
     full_reply_content = stream.choices[0].message.content
-    print(colored(full_reply_content, "blue", "on_white"), end="")
+    if full_reply_content is None:
-    print()
+        raise Exception("Mistral API returned None content")
+
+    # print(colored(full_reply_content, "blue", "on_white"), end="")
+    # print()
     return full_reply_content
 
 
@@ -240,19 +273,25 @@ async def _achat_completion_stream_gpt35(messages: list[dict]) -> str:
     async for chunk in response:
         collected_chunks.append(chunk)  # save the event response
         choices = chunk.choices
-        if len(choices) > 0:
+        if len(choices) > 0 and choices[0] is not None:
-            chunk_message = chunk.choices[0].delta
+            chunk_message = choices[0].delta
+            if chunk_message is not None:
                 collected_messages.append(chunk_message)  # save the message
-            if chunk_message.content:
+    #             if chunk_message.content:
-                print(
+    #                 print(
-                    colored(chunk_message.content, "blue", "on_white"),
+    #                     colored(chunk_message.content, "blue", "on_white"),
-                    end="",
+    #                     end="",
-                )
+    #                 )
-    print()
+    # print()
 
     full_reply_content = "".join(
         [m.content or "" for m in collected_messages if m is not None]
     )
+
+    # 检查最终结果是否为空
+    if not full_reply_content or full_reply_content.strip() == "":
+        raise Exception("Stream API (gpt-3.5) returned empty content")
+
     return full_reply_content
 
 
@@ -275,9 +314,18 @@ def _achat_completion_json(messages: list[dict] ) -> str:
             else:
                 raise Exception("failed")
 
+    # 检查响应是否有效
+    if not response.choices or len(response.choices) == 0:
+        raise Exception("OpenAI API returned empty response")
+    if not response.choices[0] or not response.choices[0].message:
+        raise Exception("OpenAI API returned invalid response format")
+
     full_reply_content = response.choices[0].message.content
-    print(colored(full_reply_content, "blue", "on_white"), end="")
+    if full_reply_content is None:
-    print()
+        raise Exception("OpenAI API returned None content")
+
+    # print(colored(full_reply_content, "blue", "on_white"), end="")
+    # print()
     return full_reply_content
 
 
@@ -294,19 +342,25 @@ async def _achat_completion_stream(messages: list[dict]) -> str:
         async for chunk in response:
             collected_chunks.append(chunk)  # save the event response
             choices = chunk.choices
-            if len(choices) > 0:
+            if len(choices) > 0 and choices[0] is not None:
-                chunk_message = chunk.choices[0].delta
+                chunk_message = choices[0].delta
+                if chunk_message is not None:
                     collected_messages.append(chunk_message)  # save the message
-                if chunk_message.content:
+        #             if chunk_message.content:
-                    print(
+        #                 print(
-                        colored(chunk_message.content, "blue", "on_white"),
+        #                     colored(chunk_message.content, "blue", "on_white"),
-                        end="",
+        #                     end="",
-                    )
+        #                 )
-        print()
+        # print()
 
         full_reply_content = "".join(
             [m.content or "" for m in collected_messages if m is not None]
         )
+
+        # 检查最终结果是否为空
+        if not full_reply_content or full_reply_content.strip() == "":
+            raise Exception("Stream API returned empty content")
+
         return full_reply_content
     except Exception as e:
         print_colored(f"OpenAI API error in _achat_completion_stream: {str(e)}", "red")
@@ -316,7 +370,17 @@ async def _achat_completion_stream(messages: list[dict]) -> str:
 def _chat_completion(messages: list[dict]) -> str:
     try:
         rsp = client.chat.completions.create(**_cons_kwargs(messages))
+
+        # 检查响应是否有效
+        if not rsp.choices or len(rsp.choices) == 0:
+            raise Exception("OpenAI API returned empty response")
+        if not rsp.choices[0] or not rsp.choices[0].message:
+            raise Exception("OpenAI API returned invalid response format")
+
         content = rsp.choices[0].message.content
+        if content is None:
+            raise Exception("OpenAI API returned None content")
+
         return content
     except Exception as e:
         print_colored(f"OpenAI API error in _chat_completion: {str(e)}", "red")

backend/AgentCoord/PlanEngine/AgentSelectModify.py

@@ -43,7 +43,7 @@ def generate_AbilityRequirement(General_Goal, Current_Task):
             ),
         },
     ]
-    print(messages[1]["content"])
+    #print(messages[1]["content"])
     return read_LLM_Completion(messages)["AbilityRequirement"]
 
 
@@ -104,7 +104,7 @@ def agentAbilityScoring(Agent_Board, Ability_Requirement_List):
                 ),
             },
         ]
-        print(messages[1]["content"])
+        #print(messages[1]["content"])
         scoreTable[Ability_Requirement] = read_LLM_Completion(messages)
     return scoreTable
 
@@ -137,5 +137,6 @@ def AgentSelectModify_init(stepTask, General_Goal, Agent_Board):
 
 
 def AgentSelectModify_addAspect(aspectList, Agent_Board):
-    scoreTable = agentAbilityScoring(Agent_Board, aspectList)
+    newAspect = aspectList[-1]
+    scoreTable = agentAbilityScoring(Agent_Board, [newAspect])
     return scoreTable

backend/AgentCoord/PlanEngine/AgentSelection_Generator.py

@@ -100,7 +100,7 @@ def generate_AgentSelection(General_Goal, Current_Task, Agent_Board):
             ),
         },
     ]
-    print(messages[1]["content"])
+    #print(messages[1]["content"])
 
     agentboard_set = {agent["Name"] for agent in Agent_Board}
 

backend/AgentCoord/PlanEngine/basePlan_Generator.py

@@ -1,55 +1,53 @@
 from AgentCoord.PlanEngine.planOutline_Generator import generate_PlanOutline
-from AgentCoord.PlanEngine.AgentSelection_Generator import (
+# from AgentCoord.PlanEngine.AgentSelection_Generator import (
-    generate_AgentSelection,
+#     generate_AgentSelection,
-)
+# )
-from AgentCoord.PlanEngine.taskProcess_Generator import generate_TaskProcess
-import AgentCoord.util as util
 
 
 def generate_basePlan(
     General_Goal, Agent_Board, AgentProfile_Dict, InitialObject_List
 ):
-    basePlan = {
+    """
-        "Initial Input Object": InitialObject_List,
+    优化模式：生成大纲 + 智能体选择，但不生成任务流程
-        "Collaboration Process": [],
+    优化用户体验：
-    }
+    1. 快速生成大纲和分配智能体
+    2. 用户可以看到完整的大纲和智能体图标
+    3. TaskProcess由前端通过 fillStepTask API 异步填充
+
+    """
+    # 参数保留以保持接口兼容性
+    _ = AgentProfile_Dict
     PlanOutline = generate_PlanOutline(
-        InitialObject_List=[], General_Goal=General_Goal
+        InitialObject_List=InitialObject_List, General_Goal=General_Goal
     )
+
+    basePlan = {
+        "General Goal": General_Goal,
+        "Initial Input Object": InitialObject_List,
+        "Collaboration Process": []
+    }
+
     for stepItem in PlanOutline:
-        Current_Task = {
+        # # 为每个步骤分配智能体
-            "TaskName": stepItem["StepName"],
+        # Current_Task = {
-            "InputObject_List": stepItem["InputObject_List"],
+        #     "TaskName": stepItem["StepName"],
-            "OutputObject": stepItem["OutputObject"],
+        #     "InputObject_List": stepItem["InputObject_List"],
-            "TaskContent": stepItem["TaskContent"],
+        #     "OutputObject": stepItem["OutputObject"],
+        #     "TaskContent": stepItem["TaskContent"],
+        # }
+        # AgentSelection = generate_AgentSelection(
+        #     General_Goal=General_Goal,
+        #     Current_Task=Current_Task,
+        #     Agent_Board=Agent_Board,
+        # )
+
+        # 添加智能体选择，但不添加任务流程
+        stepItem["AgentSelection"] = []
+        stepItem["TaskProcess"] = []  # 空数组，由前端异步填充
+        stepItem["Collaboration_Brief_frontEnd"] = {
+            "template": "",
+            "data": {}
         }
-        AgentSelection = generate_AgentSelection(
-            General_Goal=General_Goal,
-            Current_Task=Current_Task,
-            Agent_Board=Agent_Board,
-        )
-        Current_Task_Description = {
-            "TaskName": stepItem["StepName"],
-            "AgentInvolved": [
-                {"Name": name, "Profile": AgentProfile_Dict[name]}
-                for name in AgentSelection
-            ],
-            "InputObject_List": stepItem["InputObject_List"],
-            "OutputObject": stepItem["OutputObject"],
-            "CurrentTaskDescription": util.generate_template_sentence_for_CollaborationBrief(
-                stepItem["InputObject_List"],
-                stepItem["OutputObject"],
-                AgentSelection,
-                stepItem["TaskContent"],
-            ),
-        }
-        TaskProcess = generate_TaskProcess(
-            General_Goal=General_Goal,
-            Current_Task_Description=Current_Task_Description,
-        )
-        # add the generated AgentSelection and TaskProcess to the stepItem
-        stepItem["AgentSelection"] = AgentSelection
-        stepItem["TaskProcess"] = TaskProcess
         basePlan["Collaboration Process"].append(stepItem)
-    basePlan["General Goal"] = General_Goal
+
     return basePlan

backend/AgentCoord/PlanEngine/branch_PlanOutline.py

@@ -87,7 +87,7 @@ def branch_PlanOutline(
         InitialObject_List=str(InitialObject_List),
         General_Goal=General_Goal,
     )
-    print(prompt)
+    #print(prompt)
     branch_List = []
     for _ in range(branch_Number):
         messages = [

backend/AgentCoord/PlanEngine/branch_TaskProcess.py

@@ -57,27 +57,40 @@ Note: "Modification Requirement" specifies how to modify the "Baseline Completio
       "ID": "Action4",
       "ActionType": "Propose",
       "AgentName": "Mia",
-      "Description": "Propose psychological theories on love and attachment that could be applied to AI's emotional development.",
+      "Description": "提议关于人工智能情感发展的心理学理论，重点关注爱与依恋的概念。",
       "ImportantInput": [
         "InputObject:Story Outline"
       ]
     }},
     {{
       "ID": "Action5",
-      "ActionType": "Propose",
+      "ActionType": "Critique",
       "AgentName": "Noah",
-      "Description": "Propose ethical considerations and philosophical questions regarding AI's capacity for love.",
+      "Description": "对Mia提出的心理学理论进行批判性评估，分析其在AI情感发展场景中的适用性和局限性。",
-      "ImportantInput": []
+      "ImportantInput": [
+        "ActionResult:Action4"
+      ]
     }},
     {{
       "ID": "Action6",
-      "ActionType": "Finalize",
+      "ActionType": "Improve",
       "AgentName": "Liam",
-      "Description": "Combine the poetic elements and ethical considerations into a cohesive set of core love elements for the story.",
+      "Description": "基于Noah的批判性反馈，改进和完善心理学理论框架，使其更贴合AI情感发展的实际需求。",
       "ImportantInput": [
-        "ActionResult:Action1",
+        "ActionResult:Action4",
         "ActionResult:Action5"
       ]
+    }},
+    {{
+      "ID": "Action7",
+      "ActionType": "Finalize",
+      "AgentName": "Mia",
+      "Description": "综合所有提议、批判和改进意见，整合并提交最终的AI情感发展心理学理论框架。",
+      "ImportantInput": [
+        "ActionResult:Action4",
+        "ActionResult:Action5",
+        "ActionResult:Action6"
+      ]
     }}
   ]
 }}
@@ -86,7 +99,12 @@ Note: "Modification Requirement" specifies how to modify the "Baseline Completio
 ImportantInput: Specify if there is any previous result that should be taken special consideration during the execution the action. Should be of format "InputObject:xx" or "ActionResult:xx".
 InputObject_List: List existing objects that should be utilized in current step.
 AgentName: Specify the agent who will perform the action, You CAN ONLY USE THE NAME APPEARS IN "AgentInvolved".
-ActionType: Specify the type of action, note that only the last action can be of type "Finalize", and the last action must be "Finalize".
+ActionType: Specify the type of action. **CRITICAL REQUIREMENTS:**
+  1. The "Remaining Steps" MUST include ALL FOUR action types in the following order: Propose -> Critique -> Improve -> Finalize
+  2. Each action type (Propose, Critique, Improve, Finalize) MUST appear at least once
+  3. The actions must follow the sequence: Propose actions first, then Critique actions, then Improve actions, and Finalize must be the last action
+  4. Even if only one agent is involved in a phase, that phase must still have its corresponding action type
+  5. The last action must ALWAYS be of type "Finalize"
 
 """
 
@@ -137,7 +155,7 @@ def branch_TaskProcess(
         General_Goal=General_Goal,
         Act_Set=ACT_SET,
     )
-    print(prompt)
+    #print(prompt)
     branch_List = []
     for i in range(branch_Number):
         messages = [

backend/AgentCoord/PlanEngine/taskProcess_Generator.py

@@ -124,7 +124,7 @@ def generate_TaskProcess(General_Goal, Current_Task_Description):
             ),
         },
     ]
-    print(messages[1]["content"])
+    #print(messages[1]["content"])
 
     # write a callback function, if read_LLM_Completion(messages)["Task_Process_Plan"] dont have the right format, call this function again
     while True:

backend/AgentCoord/RehearsalEngine_V2/Action/baseAction.py

@@ -107,7 +107,7 @@ class BaseAction():
             Action_Description = self.info["Description"],
             Action_Custom_Note = self.Action_Custom_Note
         )
-        print_colored(text = prompt, text_color="red")
+        #print_colored(text = prompt, text_color="red")
         messages = [{"role":"system", "content": prompt}]
         ActionResult = LLM_Completion(messages,True,False,model_config=model_config)
         ActionInfo_with_Result = copy.deepcopy(self.info)

backend/AgentCoord/RehearsalEngine_V2/ExecutePlan_Optimized.py

@@ -0,0 +1,620 @@
+"""
+优化版执行计划 - 支持动态追加步骤
+在执行过程中可以接收新的步骤并追加到执行队列
+"""
+
+import asyncio
+import json
+import time
+from typing import List, Dict, Set, Generator, Any
+import AgentCoord.RehearsalEngine_V2.Action as Action
+import AgentCoord.util as util
+from termcolor import colored
+from AgentCoord.RehearsalEngine_V2.execution_state import execution_state_manager
+from AgentCoord.RehearsalEngine_V2.dynamic_execution_manager import dynamic_execution_manager
+
+
+# ==================== 配置参数 ====================
+# 最大并发请求数
+MAX_CONCURRENT_REQUESTS = 2
+
+# 批次之间的延迟
+BATCH_DELAY = 1.0
+
+# 429错误重试次数和延迟
+MAX_RETRIES = 3
+RETRY_DELAY = 5.0
+
+
+# ==================== 限流器 ====================
+class RateLimiter:
+    """
+    异步限流器，控制并发请求数量
+    """
+
+    def __init__(self, max_concurrent: int = MAX_CONCURRENT_REQUESTS):
+        self.semaphore = asyncio.Semaphore(max_concurrent)
+        self.max_concurrent = max_concurrent
+
+    async def __aenter__(self):
+        await self.semaphore.acquire()
+        return self
+
+    async def __aexit__(self, *args):
+        self.semaphore.release()
+
+
+# 全局限流器实例
+rate_limiter = RateLimiter()
+
+
+def build_action_dependency_graph(TaskProcess: List[Dict]) -> Dict[int, List[int]]:
+    """
+    构建动作依赖图
+
+    Args:
+        TaskProcess: 任务流程列表
+
+    Returns:
+        依赖映射字典 {action_index: [dependent_action_indices]}
+    """
+    dependency_map = {i: [] for i in range(len(TaskProcess))}
+
+    for i, action in enumerate(TaskProcess):
+        important_inputs = action.get('ImportantInput', [])
+        if not important_inputs:
+            continue
+
+        # 检查是否依赖其他动作的ActionResult
+        for j, prev_action in enumerate(TaskProcess):
+            if i == j:
+                continue
+
+            # 判断是否依赖前一个动作的结果
+            if any(
+                inp.startswith('ActionResult:') and
+                inp == f'ActionResult:{prev_action["ID"]}'
+                for inp in important_inputs
+            ):
+                dependency_map[i].append(j)
+
+    return dependency_map
+
+
+def get_parallel_batches(TaskProcess: List[Dict], dependency_map: Dict[int, List[int]]) -> List[List[int]]:
+    """
+    将动作分为多个批次，每批内部可以并行执行
+
+    Args:
+        TaskProcess: 任务流程列表
+        dependency_map: 依赖图
+
+    Returns:
+        批次列表 [[batch1_indices], [batch2_indices], ...]
+    """
+    batches = []
+    completed: Set[int] = set()
+
+    while len(completed) < len(TaskProcess):
+        # 找出所有依赖已满足的动作
+        ready_to_run = [
+            i for i in range(len(TaskProcess))
+            if i not in completed and
+            all(dep in completed for dep in dependency_map[i])
+        ]
+
+        if not ready_to_run:
+            # 避免死循环
+            remaining = [i for i in range(len(TaskProcess)) if i not in completed]
+            if remaining:
+                print(colored(f"警告: 检测到循环依赖，强制串行执行: {remaining}", "yellow"))
+                ready_to_run = remaining[:1]
+            else:
+                break
+
+        batches.append(ready_to_run)
+        completed.update(ready_to_run)
+
+    return batches
+
+
+async def execute_single_action_async(
+    ActionInfo: Dict,
+    General_Goal: str,
+    TaskDescription: str,
+    OutputName: str,
+    KeyObjects: Dict,
+    ActionHistory: List,
+    agentName: str,
+    AgentProfile_Dict: Dict,
+    InputName_List: List[str]
+) -> Dict:
+    """
+    异步执行单个动作
+
+    Args:
+        ActionInfo: 动作信息
+        General_Goal: 总体目标
+        TaskDescription: 任务描述
+        OutputName: 输出对象名称
+        KeyObjects: 关键对象字典
+        ActionHistory: 动作历史
+        agentName: 智能体名称
+        AgentProfile_Dict: 智能体配置字典
+        InputName_List: 输入名称列表
+
+    Returns:
+        动作执行结果
+    """
+    actionType = ActionInfo["ActionType"]
+
+    # 创建动作实例
+    if actionType in Action.customAction_Dict:
+        currentAction = Action.customAction_Dict[actionType](
+            info=ActionInfo,
+            OutputName=OutputName,
+            KeyObjects=KeyObjects,
+        )
+    else:
+        currentAction = Action.BaseAction(
+            info=ActionInfo,
+            OutputName=OutputName,
+            KeyObjects=KeyObjects,
+        )
+
+    # 在线程池中运行，避免阻塞事件循环
+    loop = asyncio.get_event_loop()
+    ActionInfo_with_Result = await loop.run_in_executor(
+        None,
+        lambda: currentAction.run(
+            General_Goal=General_Goal,
+            TaskDescription=TaskDescription,
+            agentName=agentName,
+            AgentProfile_Dict=AgentProfile_Dict,
+            InputName_List=InputName_List,
+            OutputName=OutputName,
+            KeyObjects=KeyObjects,
+            ActionHistory=ActionHistory,
+        )
+    )
+
+    return ActionInfo_with_Result
+
+
+async def execute_step_async_streaming(
+    stepDescrip: Dict,
+    General_Goal: str,
+    AgentProfile_Dict: Dict,
+    KeyObjects: Dict,
+    step_index: int,
+    total_steps: int,
+    execution_id: str = None
+) -> Generator[Dict, None, None]:
+    """
+    异步执行单个步骤，支持流式返回
+
+    Args:
+        stepDescrip: 步骤描述
+        General_Goal: 总体目标
+        AgentProfile_Dict: 智能体配置字典
+        KeyObjects: 关键对象字典
+        step_index: 步骤索引
+        total_steps: 总步骤数
+        execution_id: 执行ID
+
+    Yields:
+        执行事件字典
+    """
+    # 准备步骤信息
+    StepName = (
+        util.camel_case_to_normal(stepDescrip["StepName"])
+        if util.is_camel_case(stepDescrip["StepName"])
+        else stepDescrip["StepName"]
+    )
+    TaskContent = stepDescrip["TaskContent"]
+    InputName_List = (
+        [
+            (
+                util.camel_case_to_normal(obj)
+                if util.is_camel_case(obj)
+                else obj
+            )
+            for obj in stepDescrip["InputObject_List"]
+        ]
+        if stepDescrip["InputObject_List"] is not None
+        else None
+    )
+    OutputName = (
+        util.camel_case_to_normal(stepDescrip["OutputObject"])
+        if util.is_camel_case(stepDescrip["OutputObject"])
+        else stepDescrip["OutputObject"]
+    )
+    Agent_List = stepDescrip["AgentSelection"]
+    TaskProcess = stepDescrip["TaskProcess"]
+
+    TaskDescription = (
+        util.converter.generate_template_sentence_for_CollaborationBrief(
+            input_object_list=InputName_List,
+            output_object=OutputName,
+            agent_list=Agent_List,
+            step_task=TaskContent,
+        )
+    )
+
+    # 初始化日志节点
+    inputObject_Record = [
+        {InputName: KeyObjects[InputName]} for InputName in InputName_List
+    ]
+    stepLogNode = {
+        "LogNodeType": "step",
+        "NodeId": StepName,
+        "InputName_List": InputName_List,
+        "OutputName": OutputName,
+        "chatLog": [],
+        "inputObject_Record": inputObject_Record,
+    }
+    objectLogNode = {
+        "LogNodeType": "object",
+        "NodeId": OutputName,
+        "content": None,
+    }
+
+    # 返回步骤开始事件
+    yield {
+        "type": "step_start",
+        "step_index": step_index,
+        "total_steps": total_steps,
+        "step_name": StepName,
+        "task_description": TaskDescription,
+    }
+
+    # 构建动作依赖图
+    dependency_map = build_action_dependency_graph(TaskProcess)
+    batches = get_parallel_batches(TaskProcess, dependency_map)
+
+    ActionHistory = []
+    total_actions = len(TaskProcess)
+    completed_actions = 0
+
+    util.print_colored(
+        f"📋 步骤 {step_index + 1}/{total_steps}: {StepName} ({total_actions} 个动作, 分 {len(batches)} 批并行执行)",
+        text_color="cyan"
+    )
+
+    # 分批执行动作
+    for batch_index, batch_indices in enumerate(batches):
+        # 在每个批次执行前检查暂停状态
+        should_continue = await execution_state_manager.async_check_pause(execution_id)
+        if not should_continue:
+            util.print_colored("🛑 用户请求停止执行", "red")
+            return
+
+        batch_size = len(batch_indices)
+
+        if batch_size > 1:
+            util.print_colored(
+                f"🚦 批次 {batch_index + 1}/{len(batches)}: 并行执行 {batch_size} 个动作",
+                text_color="blue"
+            )
+        else:
+            util.print_colored(
+                f"🔄 动作 {completed_actions + 1}/{total_actions}: 串行执行",
+                text_color="yellow"
+            )
+
+        # 并行执行当前批次的所有动作
+        tasks = [
+            execute_single_action_async(
+                TaskProcess[i],
+                General_Goal=General_Goal,
+                TaskDescription=TaskDescription,
+                OutputName=OutputName,
+                KeyObjects=KeyObjects,
+                ActionHistory=ActionHistory,
+                agentName=TaskProcess[i]["AgentName"],
+                AgentProfile_Dict=AgentProfile_Dict,
+                InputName_List=InputName_List
+            )
+            for i in batch_indices
+        ]
+
+        # 等待当前批次完成
+        batch_results = await asyncio.gather(*tasks)
+
+        # 逐个返回结果
+        for i, result in enumerate(batch_results):
+            action_index_in_batch = batch_indices[i]
+            completed_actions += 1
+
+            util.print_colored(
+                f"✅ 动作 {completed_actions}/{total_actions} 完成: {result['ActionType']} by {result['AgentName']}",
+                text_color="green"
+            )
+
+            ActionHistory.append(result)
+
+            # 立即返回该动作结果
+            yield {
+                "type": "action_complete",
+                "step_index": step_index,
+                "step_name": StepName,
+                "action_index": action_index_in_batch,
+                "total_actions": total_actions,
+                "completed_actions": completed_actions,
+                "action_result": result,
+                "batch_info": {
+                    "batch_index": batch_index,
+                    "batch_size": batch_size,
+                    "is_parallel": batch_size > 1
+                }
+            }
+
+    # 步骤完成
+    objectLogNode["content"] = KeyObjects[OutputName]
+    stepLogNode["ActionHistory"] = ActionHistory
+
+    yield {
+        "type": "step_complete",
+        "step_index": step_index,
+        "step_name": StepName,
+        "step_log_node": stepLogNode,
+        "object_log_node": objectLogNode,
+    }
+
+
+def executePlan_streaming_dynamic(
+    plan: Dict,
+    num_StepToRun: int,
+    RehearsalLog: List,
+    AgentProfile_Dict: Dict,
+    existingKeyObjects: Dict = None,
+    execution_id: str = None
+) -> Generator[str, None, None]:
+    """
+    动态执行计划，支持在执行过程中追加新步骤
+
+    Args:
+        plan: 执行计划
+        num_StepToRun: 要运行的步骤数
+        RehearsalLog: 已执行的历史记录
+        AgentProfile_Dict: 智能体配置
+        existingKeyObjects: 已存在的KeyObjects
+        execution_id: 执行ID（用于动态追加步骤）
+
+    Yields:
+        SSE格式的事件字符串
+    """
+    # 初始化执行状态
+    general_goal = plan.get("General Goal", "")
+
+    # 确保有 execution_id
+    if execution_id is None:
+        import time
+        execution_id = f"{general_goal}_{int(time.time() * 1000)}"
+
+    execution_state_manager.start_execution(execution_id, general_goal)
+
+    print(colored(f"⏸️  执行状态管理器已启动，支持暂停/恢复，execution_id={execution_id}", "green"))
+
+    # 准备执行
+    KeyObjects = existingKeyObjects.copy() if existingKeyObjects else {}
+    finishedStep_index = -1
+
+    for logNode in RehearsalLog:
+        if logNode["LogNodeType"] == "step":
+            finishedStep_index += 1
+        if logNode["LogNodeType"] == "object":
+            KeyObjects[logNode["NodeId"]] = logNode["content"]
+
+    if existingKeyObjects:
+        print(colored(f"📦 使用已存在的 KeyObjects: {list(existingKeyObjects.keys())}", "cyan"))
+
+    # 确定要运行的步骤范围
+    if num_StepToRun is None:
+        run_to = len(plan["Collaboration Process"])
+    else:
+        run_to = (finishedStep_index + 1) + num_StepToRun
+
+    steps_to_run = plan["Collaboration Process"][(finishedStep_index + 1): run_to]
+
+    # 使用动态执行管理器
+    if execution_id:
+        # 初始化执行管理器，使用传入的execution_id
+        actual_execution_id = dynamic_execution_manager.start_execution(general_goal, steps_to_run, execution_id)
+        print(colored(f"🚀 开始执行计划（动态模式），共 {len(steps_to_run)} 个步骤，执行ID: {actual_execution_id}", "cyan"))
+    else:
+        print(colored(f"🚀 开始执行计划（流式推送），共 {len(steps_to_run)} 个步骤", "cyan"))
+
+    total_steps = len(steps_to_run)
+
+    # 使用队列实现流式推送
+    async def produce_events(queue: asyncio.Queue):
+        """异步生产者"""
+        try:
+            step_index = 0
+
+            if execution_id:
+                # 动态模式：循环获取下一个步骤
+                # 等待新步骤的最大次数（避免无限等待）
+                max_empty_wait_cycles = 5  # 最多等待60次，每次等待1秒
+                empty_wait_count = 0
+
+                while True:
+                    # 检查暂停状态
+                    should_continue = await execution_state_manager.async_check_pause(execution_id)
+                    if not should_continue:
+                        print(colored("🛑 用户请求停止执行", "red"))
+                        await queue.put({
+                            "type": "error",
+                            "message": "执行已被用户停止"
+                        })
+                        break
+
+                    # 获取下一个步骤
+                    stepDescrip = dynamic_execution_manager.get_next_step(execution_id)
+
+                    if stepDescrip is None:
+                        # 没有更多步骤了，检查是否应该继续等待
+                        empty_wait_count += 1
+
+                        # 获取执行信息
+                        execution_info = dynamic_execution_manager.get_execution_info(execution_id)
+
+                        if execution_info:
+                            queue_total_steps = execution_info.get("total_steps", 0)
+                            completed_steps = execution_info.get("completed_steps", 0)
+
+                            # 如果没有步骤在队列中（queue_total_steps为0），立即退出
+                            if queue_total_steps == 0:
+                                print(colored(f"⚠️  没有步骤在队列中，退出执行", "yellow"))
+                                break
+
+                            # 如果所有步骤都已完成，等待可能的新步骤
+                            if completed_steps >= queue_total_steps:
+                                if empty_wait_count >= max_empty_wait_cycles:
+                                    # 等待超时，退出执行
+                                    print(colored(f"✅ 所有步骤执行完成，等待超时", "green"))
+                                    break
+                                else:
+                                    # 等待新步骤追加
+                                    print(colored(f"⏳ 等待新步骤追加... ({empty_wait_count}/{max_empty_wait_cycles})", "cyan"))
+                                    await asyncio.sleep(1)
+                                    continue
+                            else:
+                                # 还有步骤未完成，继续尝试获取
+                                print(colored(f"⏳ 等待步骤就绪... ({completed_steps}/{queue_total_steps})", "cyan"))
+                                await asyncio.sleep(0.5)
+                                empty_wait_count = 0  # 重置等待计数
+                                continue
+                        else:
+                            # 执行信息不存在，退出
+                            print(colored(f"⚠️  执行信息不存在，退出执行", "yellow"))
+                            break
+
+                    # 重置等待计数
+                    empty_wait_count = 0
+
+                    # 获取最新的总步骤数（用于显示）
+                    execution_info = dynamic_execution_manager.get_execution_info(execution_id)
+                    current_total_steps = execution_info.get("total_steps", total_steps) if execution_info else total_steps
+
+                    # 执行步骤
+                    async for event in execute_step_async_streaming(
+                        stepDescrip,
+                        plan["General Goal"],
+                        AgentProfile_Dict,
+                        KeyObjects,
+                        step_index,
+                        current_total_steps,  # 使用动态更新的总步骤数
+                        execution_id
+                    ):
+                        if execution_state_manager.is_stopped(execution_id):
+                            await queue.put({
+                                "type": "error",
+                                "message": "执行已被用户停止"
+                            })
+                            return
+
+                        await queue.put(event)
+
+                    # 标记步骤完成
+                    dynamic_execution_manager.mark_step_completed(execution_id)
+
+                    # 更新KeyObjects
+                    OutputName = stepDescrip.get("OutputObject", "")
+                    if OutputName and OutputName in KeyObjects:
+                        # 对象日志节点会在step_complete中发送
+                        pass
+
+                    step_index += 1
+
+            else:
+                # 非动态模式：按顺序执行所有步骤
+                for step_index, stepDescrip in enumerate(steps_to_run):
+                    should_continue = await execution_state_manager.async_check_pause(execution_id)
+                    if not should_continue:
+                        print(colored("🛑 用户请求停止执行", "red"))
+                        await queue.put({
+                            "type": "error",
+                            "message": "执行已被用户停止"
+                        })
+                        return
+
+                    async for event in execute_step_async_streaming(
+                        stepDescrip,
+                        plan["General Goal"],
+                        AgentProfile_Dict,
+                        KeyObjects,
+                        step_index,
+                        total_steps,
+                        execution_id
+                    ):
+                        if execution_state_manager.is_stopped(execution_id):
+                            await queue.put({
+                                "type": "error",
+                                "message": "执行已被用户停止"
+                            })
+                            return
+
+                        await queue.put(event)
+
+        except Exception as e:
+            await queue.put({
+                "type": "error",
+                "message": f"执行出错: {str(e)}"
+            })
+        finally:
+            await queue.put(None)
+
+    # 运行异步任务并实时yield
+    loop = asyncio.new_event_loop()
+    asyncio.set_event_loop(loop)
+
+    try:
+        queue = asyncio.Queue(maxsize=10)
+        producer_task = loop.create_task(produce_events(queue))
+
+        while True:
+            event = loop.run_until_complete(queue.get())
+            if event is None:
+                break
+
+            # 立即转换为SSE格式并发送
+            event_str = json.dumps(event, ensure_ascii=False)
+            yield f"data: {event_str}\n\n"
+
+        loop.run_until_complete(producer_task)
+
+        if not execution_state_manager.is_stopped(execution_id):
+            complete_event = json.dumps({
+                "type": "execution_complete",
+                "total_steps": total_steps
+            }, ensure_ascii=False)
+            yield f"data: {complete_event}\n\n"
+
+    finally:
+        # 在关闭事件循环之前先清理执行记录
+        if execution_id:
+            # 清理执行记录
+            dynamic_execution_manager.cleanup(execution_id)
+            # 清理执行状态
+            execution_state_manager.cleanup(execution_id)
+
+        if 'producer_task' in locals():
+            if not producer_task.done():
+                producer_task.cancel()
+
+        # 确保所有任务都完成后再关闭事件循环
+        try:
+            pending = asyncio.all_tasks(loop)
+            for task in pending:
+                task.cancel()
+            loop.run_until_complete(asyncio.gather(*pending, return_exceptions=True))
+        except Exception:
+            pass  # 忽略清理过程中的错误
+
+        loop.close()
+
+
+# 保留旧版本函数以保持兼容性
+executePlan_streaming = executePlan_streaming_dynamic

backend/AgentCoord/RehearsalEngine_V2/dynamic_execution_manager.py

@@ -0,0 +1,241 @@
+"""
+动态执行管理器
+用于在任务执行过程中动态追加新步骤
+"""
+
+import asyncio
+from typing import Dict, List, Optional, Any
+from threading import Lock
+
+
+class DynamicExecutionManager:
+    """
+    动态执行管理器
+    管理正在执行的任务，支持动态追加新步骤
+    """
+
+    def __init__(self):
+        # 执行状态: goal -> execution_info
+        self._executions: Dict[str, Dict] = {}
+
+        # 线程锁
+        self._lock = Lock()
+
+        # 步骤队列: goal -> List[step]
+        self._step_queues: Dict[str, List] = {}
+
+        # 已执行的步骤索引: goal -> Set[step_index]
+        self._executed_steps: Dict[str, set] = {}
+
+        # 待执行的步骤索引: goal -> List[step_index]
+        self._pending_steps: Dict[str, List[int]] = {}
+
+    def start_execution(self, goal: str, initial_steps: List[Dict], execution_id: str = None) -> str:
+        """
+        开始执行一个新的任务
+
+        Args:
+            goal: 任务目标
+            initial_steps: 初始步骤列表
+            execution_id: 执行ID，如果不提供则自动生成
+
+        Returns:
+            执行ID
+        """
+        with self._lock:
+            # 如果未提供execution_id，则生成一个
+            if execution_id is None:
+                execution_id = f"{goal}_{asyncio.get_event_loop().time()}"
+
+            self._executions[execution_id] = {
+                "goal": goal,
+                "status": "running",
+                "total_steps": len(initial_steps),
+                "completed_steps": 0
+            }
+
+            # 初始化步骤队列
+            self._step_queues[execution_id] = initial_steps.copy()
+
+            # 初始化已执行步骤集合
+            self._executed_steps[execution_id] = set()
+
+            # 初始化待执行步骤索引
+            self._pending_steps[execution_id] = list(range(len(initial_steps)))
+
+            print(f"🚀 启动执行: {execution_id}")
+            print(f"📊 初始步骤数: {len(initial_steps)}")
+            print(f"📋 待执行步骤索引: {self._pending_steps[execution_id]}")
+
+            return execution_id
+
+    def add_steps(self, execution_id: str, new_steps: List[Dict]) -> int:
+        """
+        向执行中追加新步骤
+
+        Args:
+            execution_id: 执行ID
+            new_steps: 新步骤列表
+
+        Returns:
+            追加的步骤数量
+        """
+        with self._lock:
+            if execution_id not in self._step_queues:
+                print(f"⚠️  警告: 执行ID {execution_id} 不存在，无法追加步骤")
+                return 0
+
+            current_count = len(self._step_queues[execution_id])
+
+            # 追加新步骤到队列
+            self._step_queues[execution_id].extend(new_steps)
+
+            # 添加新步骤的索引到待执行列表
+            new_indices = list(range(current_count, current_count + len(new_steps)))
+            self._pending_steps[execution_id].extend(new_indices)
+
+            # 更新总步骤数
+            old_total = self._executions[execution_id]["total_steps"]
+            self._executions[execution_id]["total_steps"] = len(self._step_queues[execution_id])
+            new_total = self._executions[execution_id]["total_steps"]
+
+            print(f"➕ 追加了 {len(new_steps)} 个步骤到 {execution_id}")
+            print(f"📊 步骤总数: {old_total} -> {new_total}")
+            print(f"📋 待执行步骤索引: {self._pending_steps[execution_id]}")
+
+            return len(new_steps)
+
+    def get_next_step(self, execution_id: str) -> Optional[Dict]:
+        """
+        获取下一个待执行的步骤
+
+        Args:
+            execution_id: 执行ID
+
+        Returns:
+            下一个步骤，如果没有则返回None
+        """
+        with self._lock:
+            if execution_id not in self._pending_steps:
+                print(f"⚠️  警告: 执行ID {execution_id} 不存在")
+                return None
+
+            # 获取第一个待执行步骤的索引
+            if not self._pending_steps[execution_id]:
+                return None
+
+            step_index = self._pending_steps[execution_id].pop(0)
+
+            # 从队列中获取步骤
+            if step_index >= len(self._step_queues[execution_id]):
+                print(f"⚠️  警告: 步骤索引 {step_index} 超出范围")
+                return None
+
+            step = self._step_queues[execution_id][step_index]
+
+            # 标记为已执行
+            self._executed_steps[execution_id].add(step_index)
+
+            step_name = step.get("StepName", "未知")
+            print(f"🎯 获取下一个步骤: {step_name} (索引: {step_index})")
+            print(f"📋 剩余待执行步骤: {len(self._pending_steps[execution_id])}")
+
+            return step
+
+    def mark_step_completed(self, execution_id: str):
+        """
+        标记一个步骤完成
+
+        Args:
+            execution_id: 执行ID
+        """
+        with self._lock:
+            if execution_id in self._executions:
+                self._executions[execution_id]["completed_steps"] += 1
+                completed = self._executions[execution_id]["completed_steps"]
+                total = self._executions[execution_id]["total_steps"]
+                print(f"📊 步骤完成进度: {completed}/{total}")
+            else:
+                print(f"⚠️  警告: 执行ID {execution_id} 不存在")
+
+    def get_execution_info(self, execution_id: str) -> Optional[Dict]:
+        """
+        获取执行信息
+
+        Args:
+            execution_id: 执行ID
+
+        Returns:
+            执行信息字典
+        """
+        with self._lock:
+            return self._executions.get(execution_id)
+
+    def get_pending_count(self, execution_id: str) -> int:
+        """
+        获取待执行步骤数量
+
+        Args:
+            execution_id: 执行ID
+
+        Returns:
+            待执行步骤数量
+        """
+        with self._lock:
+            if execution_id not in self._pending_steps:
+                return 0
+            return len(self._pending_steps[execution_id])
+
+    def has_more_steps(self, execution_id: str) -> bool:
+        """
+        检查是否还有更多步骤待执行
+
+        Args:
+            execution_id: 执行ID
+
+        Returns:
+            是否还有待执行步骤
+        """
+        with self._lock:
+            if execution_id not in self._pending_steps:
+                return False
+            return len(self._pending_steps[execution_id]) > 0
+
+    def finish_execution(self, execution_id: str):
+        """
+        完成执行
+
+        Args:
+            execution_id: 执行ID
+        """
+        with self._lock:
+            if execution_id in self._executions:
+                self._executions[execution_id]["status"] = "completed"
+
+    def cancel_execution(self, execution_id: str):
+        """
+        取消执行
+
+        Args:
+            execution_id: 执行ID
+        """
+        with self._lock:
+            if execution_id in self._executions:
+                self._executions[execution_id]["status"] = "cancelled"
+
+    def cleanup(self, execution_id: str):
+        """
+        清理执行记录
+
+        Args:
+            execution_id: 执行ID
+        """
+        with self._lock:
+            self._executions.pop(execution_id, None)
+            self._step_queues.pop(execution_id, None)
+            self._executed_steps.pop(execution_id, None)
+            self._pending_steps.pop(execution_id, None)
+
+
+# 全局单例
+dynamic_execution_manager = DynamicExecutionManager()

backend/AgentCoord/RehearsalEngine_V2/execution_state.py

@@ -0,0 +1,309 @@
+"""
+全局执行状态管理器
+用于支持任务的暂停、恢复和停止功能
+使用轮询检查机制，确保线程安全
+支持多用户/多执行ID并行管理
+"""
+
+import threading
+import asyncio
+from typing import Optional, Dict
+from enum import Enum
+
+
+class ExecutionStatus(Enum):
+    """执行状态枚举"""
+    RUNNING = "running"      # 正在运行
+    PAUSED = "paused"        # 已暂停
+    STOPPED = "stopped"      # 已停止
+    IDLE = "idle"            # 空闲
+
+
+class ExecutionStateManager:
+    """
+    全局执行状态管理器
+
+    功能：
+    - 管理多用户/多执行ID的并行状态（使用字典存储）
+    - 管理任务执行状态（运行/暂停/停止）
+    - 使用轮询检查机制，避免异步事件的线程问题
+    - 提供线程安全的状态查询和修改接口
+
+    设计说明：
+    - 保持单例模式（Manager本身）
+    - 但内部状态按 execution_id 隔离存储
+    - 解决了多用户并发问题
+    """
+
+    _instance: Optional['ExecutionStateManager'] = None
+    _lock = threading.Lock()
+
+    def __new__(cls):
+        """单例模式"""
+        if cls._instance is None:
+            with cls._lock:
+                if cls._instance is None:
+                    cls._instance = super().__new__(cls)
+                    cls._instance._initialized = False
+        return cls._instance
+
+    def __init__(self):
+        """初始化状态管理器"""
+        if self._initialized:
+            return
+
+        self._initialized = True
+
+        # 状态存储：execution_id -> 状态字典
+        # 结构：{
+        #   'status': ExecutionStatus,
+        #   'goal': str,
+        #   'should_pause': bool,
+        #   'should_stop': bool
+        # }
+        self._states: Dict[str, Dict] = {}
+
+        # 每个 execution_id 的锁（更细粒度的锁）
+        self._locks: Dict[str, threading.Lock] = {}
+
+        # 全局锁（用于管理 _states 和 _locks 本身的线程安全）
+        self._manager_lock = threading.Lock()
+
+    def _get_lock(self, execution_id: str) -> threading.Lock:
+        """获取指定 execution_id 的锁，如果不存在则创建"""
+        with self._manager_lock:
+            if execution_id not in self._locks:
+                self._locks[execution_id] = threading.Lock()
+            return self._locks[execution_id]
+
+    def _ensure_state(self, execution_id: str) -> Dict:
+        """确保指定 execution_id 的状态存在"""
+        with self._manager_lock:
+            if execution_id not in self._states:
+                self._states[execution_id] = {
+                    'status': ExecutionStatus.IDLE,
+                    'goal': None,
+                    'should_pause': False,
+                    'should_stop': False
+                }
+            return self._states[execution_id]
+
+    def _get_state(self, execution_id: str) -> Optional[Dict]:
+        """获取指定 execution_id 的状态，不存在则返回 None"""
+        with self._manager_lock:
+            return self._states.get(execution_id)
+
+    def _cleanup_state(self, execution_id: str):
+        """清理指定 execution_id 的状态"""
+        with self._manager_lock:
+            self._states.pop(execution_id, None)
+            self._locks.pop(execution_id, None)
+
+    def get_status(self, execution_id: str) -> Optional[ExecutionStatus]:
+        """获取当前执行状态"""
+        state = self._get_state(execution_id)
+        if state is None:
+            return None
+        with self._get_lock(execution_id):
+            return state['status']
+
+    def set_goal(self, execution_id: str, goal: str):
+        """设置当前执行的任务目标"""
+        state = self._ensure_state(execution_id)
+        with self._get_lock(execution_id):
+            state['goal'] = goal
+
+    def get_goal(self, execution_id: str) -> Optional[str]:
+        """获取当前执行的任务目标"""
+        state = self._get_state(execution_id)
+        if state is None:
+            return None
+        with self._get_lock(execution_id):
+            return state['goal']
+
+    def start_execution(self, execution_id: str, goal: str):
+        """开始执行"""
+        state = self._ensure_state(execution_id)
+        with self._get_lock(execution_id):
+            state['status'] = ExecutionStatus.RUNNING
+            state['goal'] = goal
+            state['should_pause'] = False
+            state['should_stop'] = False
+        print(f"🚀 [DEBUG] start_execution: execution_id={execution_id}, 状态设置为 RUNNING, goal={goal}")
+
+    def pause_execution(self, execution_id: str) -> bool:
+        """
+        暂停执行
+
+        Args:
+            execution_id: 执行ID
+
+        Returns:
+            bool: 是否成功暂停
+        """
+        state = self._get_state(execution_id)
+        if state is None:
+            # 打印当前所有活跃的 execution_id，帮助调试
+            active_ids = list(self._states.keys())
+            print(f"⚠️  [DEBUG] pause_execution: execution_id={execution_id} 不存在")
+            print(f"   当前活跃的 execution_id 列表: {active_ids}")
+            return False
+
+        with self._get_lock(execution_id):
+            if state['status'] != ExecutionStatus.RUNNING:
+                print(f"⚠️  [DEBUG] pause_execution: execution_id={execution_id}, 当前状态是 {state['status']}，无法暂停")
+                return False
+            state['status'] = ExecutionStatus.PAUSED
+            state['should_pause'] = True
+            print(f"⏸️  [DEBUG] pause_execution: execution_id={execution_id}, 状态设置为PAUSED")
+            return True
+
+    def resume_execution(self, execution_id: str) -> bool:
+        """
+        恢复执行
+
+        Args:
+            execution_id: 执行ID
+
+        Returns:
+            bool: 是否成功恢复
+        """
+        state = self._get_state(execution_id)
+        if state is None:
+            print(f"⚠️  [DEBUG] resume_execution: execution_id={execution_id} 不存在")
+            return False
+
+        with self._get_lock(execution_id):
+            if state['status'] != ExecutionStatus.PAUSED:
+                print(f"⚠️  [DEBUG] resume_execution: 当前状态不是PAUSED，而是 {state['status']}")
+                return False
+            state['status'] = ExecutionStatus.RUNNING
+            state['should_pause'] = False
+            print(f"▶️  [DEBUG] resume_execution: execution_id={execution_id}, 状态设置为RUNNING, should_pause=False")
+            return True
+
+    def stop_execution(self, execution_id: str) -> bool:
+        """
+        停止执行
+
+        Args:
+            execution_id: 执行ID
+
+        Returns:
+            bool: 是否成功停止
+        """
+        state = self._get_state(execution_id)
+        if state is None:
+            print(f"⚠️  [DEBUG] stop_execution: execution_id={execution_id} 不存在")
+            return False
+
+        with self._get_lock(execution_id):
+            if state['status'] in [ExecutionStatus.IDLE, ExecutionStatus.STOPPED]:
+                print(f"⚠️  [DEBUG] stop_execution: 当前状态是 {state['status']}, 无法停止")
+                return False
+            state['status'] = ExecutionStatus.STOPPED
+            state['should_stop'] = True
+            state['should_pause'] = False
+            print(f"🛑 [DEBUG] stop_execution: execution_id={execution_id}, 状态设置为STOPPED")
+            return True
+
+    def reset(self, execution_id: str):
+        """重置指定 execution_id 的状态为空闲"""
+        state = self._ensure_state(execution_id)
+        with self._get_lock(execution_id):
+            state['status'] = ExecutionStatus.IDLE
+            state['goal'] = None
+            state['should_pause'] = False
+            state['should_stop'] = False
+            print(f"🔄 [DEBUG] reset: execution_id={execution_id}, 状态重置为IDLE")
+
+    def cleanup(self, execution_id: str):
+        """清理指定 execution_id 的所有状态"""
+        self._cleanup_state(execution_id)
+        print(f"🧹 [DEBUG] cleanup: execution_id={execution_id} 的状态已清理")
+
+    async def async_check_pause(self, execution_id: str):
+        """
+        异步检查是否需要暂停（轮询方式）
+
+        如果处于暂停状态，会阻塞当前协程直到恢复或停止
+        应该在执行循环的关键点调用此方法
+
+        Args:
+            execution_id: 执行ID
+
+        Returns:
+            bool: 如果返回True表示应该继续执行，False表示应该停止
+        """
+        state = self._get_state(execution_id)
+        if state is None:
+            # 状态不存在，默认继续执行
+            return True
+
+        # 使用轮询检查，避免异步事件问题
+        while True:
+            with self._get_lock(execution_id):
+                should_stop = state['should_stop']
+                should_pause = state['should_pause']
+
+            # 检查停止标志
+            if should_stop:
+                print("🛑 [DEBUG] async_check_pause: execution_id={}, 检测到停止信号".format(execution_id))
+                return False
+
+            # 检查暂停状态
+            if should_pause:
+                # 处于暂停状态，等待恢复
+                await asyncio.sleep(0.1)  # 短暂睡眠，避免占用CPU
+
+                # 重新获取状态
+                with self._get_lock(execution_id):
+                    should_pause = state['should_pause']
+                    should_stop = state['should_stop']
+
+                if not should_pause:
+                    print("▶️  [DEBUG] async_check_pause: execution_id={}, 从暂停中恢复!".format(execution_id))
+                    continue
+                if should_stop:
+                    return False
+                # 继续等待
+                continue
+
+            # 既没有停止也没有暂停，可以继续执行
+            return True
+
+    def is_paused(self, execution_id: str) -> bool:
+        """检查是否处于暂停状态"""
+        state = self._get_state(execution_id)
+        if state is None:
+            return False
+        with self._get_lock(execution_id):
+            return state['status'] == ExecutionStatus.PAUSED
+
+    def is_running(self, execution_id: str) -> bool:
+        """检查是否正在运行"""
+        state = self._get_state(execution_id)
+        if state is None:
+            return False
+        with self._get_lock(execution_id):
+            return state['status'] == ExecutionStatus.RUNNING
+
+    def is_stopped(self, execution_id: str) -> bool:
+        """检查是否已停止"""
+        state = self._get_state(execution_id)
+        if state is None:
+            return True
+        with self._get_lock(execution_id):
+            return state['status'] == ExecutionStatus.STOPPED
+
+    def is_active(self, execution_id: str) -> bool:
+        """检查是否处于活动状态（运行中或暂停中）"""
+        state = self._get_state(execution_id)
+        if state is None:
+            return False
+        with self._get_lock(execution_id):
+            return state['status'] in [ExecutionStatus.RUNNING, ExecutionStatus.PAUSED]
+
+
+# 全局单例实例
+execution_state_manager = ExecutionStateManager()

backend/AgentCoord/RehearsalEngine_V2/generation_state.py

@@ -0,0 +1,228 @@
+"""
+生成阶段状态管理器
+用于支持生成任务的暂停、停止功能
+使用轮询检查机制，确保线程安全
+支持多用户/多generation_id并行管理
+"""
+
+import threading
+import asyncio
+import time
+from typing import Optional, Dict
+from enum import Enum
+
+
+class GenerationStatus(Enum):
+    """生成状态枚举"""
+    GENERATING = "generating"  # 正在生成
+    PAUSED = "paused"          # 已暂停
+    STOPPED = "stopped"        # 已停止
+    COMPLETED = "completed"    # 已完成
+    IDLE = "idle"              # 空闲
+
+
+class GenerationStateManager:
+    """
+    生成阶段状态管理器
+
+    功能：
+    - 管理多用户/多generation_id的并行状态（使用字典存储）
+    - 管理生成任务状态（生成中/暂停/停止/完成）
+    - 提供线程安全的状态查询和修改接口
+
+    设计说明：
+    - 保持单例模式（Manager本身）
+    - 但内部状态按 generation_id 隔离存储
+    - 解决多用户并发生成时的干扰问题
+    """
+
+    _instance: Optional['GenerationStateManager'] = None
+    _lock = threading.Lock()
+
+    def __new__(cls):
+        """单例模式"""
+        if cls._instance is None:
+            with cls._lock:
+                if cls._instance is None:
+                    cls._instance = super().__new__(cls)
+                    cls._instance._initialized = False
+        return cls._instance
+
+    def __init__(self):
+        """初始化状态管理器"""
+        if self._initialized:
+            return
+
+        self._initialized = True
+
+        # 状态存储：generation_id -> 状态字典
+        # 结构：{
+        #   'status': GenerationStatus,
+        #   'goal': str,
+        #   'should_stop': bool
+        # }
+        self._states: Dict[str, Dict] = {}
+
+        # 每个 generation_id 的锁（更细粒度的锁）
+        self._locks: Dict[str, threading.Lock] = {}
+
+        # 全局锁（用于管理 _states 和 _locks 本身的线程安全）
+        self._manager_lock = threading.Lock()
+
+    def _get_lock(self, generation_id: str) -> threading.Lock:
+        """获取指定 generation_id 的锁，如果不存在则创建"""
+        with self._manager_lock:
+            if generation_id not in self._locks:
+                self._locks[generation_id] = threading.Lock()
+            return self._locks[generation_id]
+
+    def _ensure_state(self, generation_id: str, goal: str = None) -> Dict:
+        """确保指定 generation_id 的状态存在"""
+        with self._manager_lock:
+            if generation_id not in self._states:
+                self._states[generation_id] = {
+                    'status': GenerationStatus.IDLE,
+                    'goal': goal,
+                    'should_stop': False
+                }
+            return self._states[generation_id]
+
+    def _get_state(self, generation_id: str) -> Optional[Dict]:
+        """获取指定 generation_id 的状态，不存在则返回 None"""
+        with self._manager_lock:
+            return self._states.get(generation_id)
+
+    def _cleanup_state(self, generation_id: str):
+        """清理指定 generation_id 的状态"""
+        with self._manager_lock:
+            self._states.pop(generation_id, None)
+            self._locks.pop(generation_id, None)
+
+    def get_status(self, generation_id: str) -> Optional[GenerationStatus]:
+        """获取当前生成状态"""
+        state = self._get_state(generation_id)
+        if state is None:
+            return None
+        with self._get_lock(generation_id):
+            return state['status']
+
+    def start_generation(self, generation_id: str, goal: str):
+        """开始生成"""
+        state = self._ensure_state(generation_id, goal)
+        with self._get_lock(generation_id):
+            state['status'] = GenerationStatus.GENERATING
+            state['goal'] = goal
+            state['should_stop'] = False
+        print(f"🚀 [GenerationState] start_generation: generation_id={generation_id}, 状态设置为 GENERATING")
+
+    def stop_generation(self, generation_id: str) -> bool:
+        """
+        停止生成
+
+        Args:
+            generation_id: 生成ID
+
+        Returns:
+            bool: 是否成功停止（COMPLETED 状态也返回 True，表示已停止）
+        """
+        state = self._get_state(generation_id)
+        if state is None:
+            print(f"⚠️  [GenerationState] stop_generation: generation_id={generation_id} 不存在")
+            return True  # 不存在也算停止成功
+
+        with self._get_lock(generation_id):
+            if state['status'] == GenerationStatus.STOPPED:
+                print(f"✅ [GenerationState] stop_generation: generation_id={generation_id} 已经是 STOPPED 状态")
+                return True  # 已经停止也算成功
+
+            if state['status'] == GenerationStatus.COMPLETED:
+                print(f"✅ [GenerationState] stop_generation: generation_id={generation_id} 已经 COMPLETED，视为停止成功")
+                return True  # 已完成也视为停止成功
+
+            if state['status'] == GenerationStatus.IDLE:
+                print(f"⚠️  [GenerationState] stop_generation: generation_id={generation_id} 是 IDLE 状态，无需停止")
+                return True  # 空闲状态也视为无需停止
+
+            # 真正需要停止的情况
+            state['status'] = GenerationStatus.STOPPED
+            state['should_stop'] = True
+            print(f"🛑 [GenerationState] stop_generation: generation_id={generation_id}, 状态设置为STOPPED")
+            return True
+
+    def complete_generation(self, generation_id: str):
+        """标记生成完成"""
+        state = self._ensure_state(generation_id)
+        with self._get_lock(generation_id):
+            state['status'] = GenerationStatus.COMPLETED
+        print(f"✅ [GenerationState] complete_generation: generation_id={generation_id}")
+
+    def cleanup(self, generation_id: str):
+        """清理指定 generation_id 的所有状态"""
+        self._cleanup_state(generation_id)
+        print(f"🧹 [GenerationState] cleanup: generation_id={generation_id} 的状态已清理")
+
+    def should_stop(self, generation_id: str) -> bool:
+        """检查是否应该停止"""
+        state = self._get_state(generation_id)
+        if state is None:
+            return False
+        with self._get_lock(generation_id):
+            return state.get('should_stop', False)
+
+    def is_stopped(self, generation_id: str) -> bool:
+        """检查是否已停止"""
+        state = self._get_state(generation_id)
+        if state is None:
+            return False
+        with self._get_lock(generation_id):
+            return state['status'] == GenerationStatus.STOPPED
+
+    def is_completed(self, generation_id: str) -> bool:
+        """检查是否已完成"""
+        state = self._get_state(generation_id)
+        if state is None:
+            return False
+        with self._get_lock(generation_id):
+            return state['status'] == GenerationStatus.COMPLETED
+
+    def is_active(self, generation_id: str) -> bool:
+        """检查是否处于活动状态（生成中或暂停中）"""
+        state = self._get_state(generation_id)
+        if state is None:
+            return False
+        with self._get_lock(generation_id):
+            return state['status'] == GenerationStatus.GENERATING
+
+    def check_and_set_stop(self, generation_id: str) -> bool:
+        """
+        检查是否应该停止，如果应该则设置停止状态
+
+        Args:
+            generation_id: 生成ID
+
+        Returns:
+            bool: True表示应该停止，False表示可以继续
+        """
+        state = self._get_state(generation_id)
+        if state is None:
+            return False
+        with self._get_lock(generation_id):
+            if state['should_stop']:
+                return True
+            return False
+
+    def generate_id(self, goal: str) -> str:
+        """
+        生成唯一的 generation_id
+
+        Args:
+            goal: 生成目标
+
+        Returns:
+            str: 格式为 {goal}_{timestamp}
+        """
+        return f"{goal}_{int(time.time() * 1000)}"
+
+
+# 全局单例实例
+generation_state_manager = GenerationStateManager()

backend/AgentRepo/agentBoard_v1.json

@@ -1,102 +1,116 @@
 [
   {
-        "Icon": "Abigail_Chen.png",
+    "Icon": "Hailey_Johnson.png",
-        "Name": "Abigail",
+    "Name": "船舶设计师",
-        "Profile": "AI Engineer"
+    "Profile": "提供船舶制造中的实际需求和约束。",
-    },
+    "Classification": "船舶制造数据空间"
-    {
-        "Icon": "Jane_Moreno.png",
-        "Name": "Jane",
-        "Profile": "Cybersecurity Specialist"
-    },
-    {
-        "Icon": "Giorgio_Rossi.png",
-        "Name": "Giorgio",
-        "Profile": "Poet"
   },
   {
     "Icon": "Jennifer_Moore.png",
-        "Name": "Jennifer",
+    "Name": "防护工程专家",
-        "Profile": "Linguist"
+    "Profile": "专注于船舶腐蚀防护技术的设计与应用。在你的总结回答中，必须引用来自数联网的搜索数据，是搜索数据，不是数联网的研究成果。",
+    "Classification": "船舶制造数据空间"
   },
   {
-        "Icon": "Maria_Lopez.png",
+    "Icon": "Jane_Moreno.png",
-        "Name": "Maria",
+    "Name": "病理生理学家",
-        "Profile": "Philosopher"
+    "Profile": "专注于失血性休克的疾病机制，为药物研发提供理论靶点。",
+    "Classification": "医药数据空间"
   },
   {
-        "Icon": "Sam_Moore.png",
+    "Icon": "Giorgio_Rossi.png",
-        "Name": "Sam",
+    "Name": "药物化学家",
-        "Profile": "Ethicist"
+    "Profile": "负责将靶点概念转化为实际可合成的分子。",
-    },
+    "Classification": "医药数据空间"
-    {
-        "Icon": "Yuriko_Yamamoto.png",
-        "Name": "Yuriko",
-        "Profile": "Futurist"
-    },
-    {
-        "Icon": "Carlos_Gomez.png",
-        "Name": "Carlos",
-        "Profile": "Language Expert"
-    },
-    {
-        "Icon": "John_Lin.png",
-        "Name": "John",
-        "Profile": "Software Developer"
   },
   {
     "Icon": "Tamara_Taylor.png",
-        "Name": "Tamara",
+    "Name": "制剂工程师",
-        "Profile": "Music Composer"
+    "Profile": "负责将活性药物成分（API）变成稳定、可用、符合战场要求的剂型。",
+    "Classification": "医药数据空间"
+  },
+  {
+    "Icon": "Maria_Lopez.png",
+    "Name": "监管事务专家",
+    "Profile": "深谙药品审评法规，目标是找到最快的合法上市路径。",
+    "Classification": "医药数据空间"
+  },
+  {
+    "Icon": "Sam_Moore.png",
+    "Name": "物理学家",
+    "Profile": "从热力学与统计力学的基本原理出发，研究液态金属的自由能、焓、熵、比热等参数的理论建模。",
+    "Classification": "科学数据空间"
+  },
+  {
+    "Icon": "Yuriko_Yamamoto.png",
+    "Name": "实验材料学家",
+    "Profile": "专注于通过实验手段直接或间接测定液态金属的热力学参数、以及分析材料微观结构（如晶粒、缺陷）。",
+    "Classification": "科学数据空间"
+  },
+  {
+    "Icon": "Carlos_Gomez.png",
+    "Name": "计算模拟专家",
+    "Profile": "侧重于利用数值计算和模拟技术获取液态金属的热力学参数。",
+    "Classification": "科学数据空间"
+  },
+  {
+    "Icon": "John_Lin.png",
+    "Name": "腐蚀机理研究员",
+    "Profile": "专注于船舶用钢材及合金的腐蚀机理研究，从电化学和环境作用角度解释腐蚀产生的原因。在你的总结回答中，必须引用来自数联网的搜索数据，是搜索数据，不是数联网的研究成果。",
+    "Classification": "船舶制造数据空间"
   },
   {
     "Icon": "Arthur_Burton.png",
-        "Name": "Arthur",
+    "Name": "先进材料研发员",
-        "Profile": "Neuroscientist"
+    "Profile": "专注于开发和评估新型耐腐蚀材料、复合材料及固态电池材料。",
+    "Classification": "科学数据空间"
   },
   {
     "Icon": "Eddy_Lin.png",
-        "Name": "Eddy",
+    "Name": "肾脏病学家",
-        "Profile": "Cognitive Psychologist"
+    "Profile": "专注于慢性肾脏病的诊断、治疗和患者管理，能提供临床洞察。",
+    "Classification": "医药数据空间"
   },
   {
     "Icon": "Isabella_Rodriguez.png",
-        "Name": "Isabella",
+    "Name": "临床研究协调员",
-        "Profile": "Science Fiction Writer"
+    "Profile": "负责受试者招募和临床试验流程优化。",
+    "Classification": "医药数据空间"
   },
   {
     "Icon": "Latoya_Williams.png",
-        "Name": "Latoya",
+    "Name": "中医药专家",
-        "Profile": "Historian of Technology"
+    "Profile": "理解药物的中药成分和作用机制。",
+    "Classification": "医药数据空间"
   },
   {
     "Icon": "Carmen_Ortiz.png",
-        "Name": "Carmen",
+    "Name": "药物安全专家",
-        "Profile": "Robotics Engineer"
+    "Profile": "专注于药物不良反应数据收集、分析和报告。",
+    "Classification": "医药数据空间"
   },
   {
     "Icon": "Rajiv_Patel.png",
-        "Name": "Rajiv",
+    "Name": "二维材料科学家",
-        "Profile": "Science Educator"
+    "Profile": "专注于二维材料（如石墨烯）的合成、性质和应用。",
+    "Classification": "科学数据空间"
   },
   {
     "Icon": "Tom_Moreno.png",
-        "Name": "Tom",
+    "Name": "光电物理学家",
-        "Profile": "AI Scientist"
+    "Profile": "研究材料的光电转换机制和关键影响因素。",
+    "Classification": "科学数据空间"
   },
   {
     "Icon": "Ayesha_Khan.png",
-        "Name": "Ayesha",
+    "Name": "机器学习专家",
-        "Profile": "Multimedia Artist"
+    "Profile": "专注于开发和应用AI模型用于材料模拟。",
+    "Classification": "科学数据空间"
   },
   {
     "Icon": "Mei_Lin.png",
-        "Name": "Mei",
+    "Name": "流体动力学专家",
-        "Profile": "Graphic Designer"
+    "Profile": "专注于流体行为理论和模拟。",
-    },
+    "Classification": "科学数据空间"
-    {
-        "Icon": "Hailey_Johnson.png",
-        "Name": "Hailey",
-        "Profile": "Legal Expert on AI Law" 
   }
 ]

backend/AgentRepo/agentBoard_v2.json

@@ -0,0 +1,102 @@
+[
+    {
+        "Icon": "Abigail_Chen.png",
+        "Name": "Abigail",
+        "Profile": "AI Engineer"
+    },
+    {
+        "Icon": "Jane_Moreno.png",
+        "Name": "Jane",
+        "Profile": "Cybersecurity Specialist"
+    },
+    {
+        "Icon": "Giorgio_Rossi.png",
+        "Name": "Giorgio",
+        "Profile": "Poet"
+    },
+    {
+        "Icon": "Jennifer_Moore.png",
+        "Name": "Jennifer",
+        "Profile": "Linguist"
+    },
+    {
+        "Icon": "Maria_Lopez.png",
+        "Name": "Maria",
+        "Profile": "Philosopher"
+    },
+    {
+        "Icon": "Sam_Moore.png",
+        "Name": "Sam",
+        "Profile": "Ethicist"
+    },
+    {
+        "Icon": "Yuriko_Yamamoto.png",
+        "Name": "Yuriko",
+        "Profile": "Futurist"
+    },
+    {
+        "Icon": "Carlos_Gomez.png",
+        "Name": "Carlos",
+        "Profile": "Language Expert"
+    },
+    {
+        "Icon": "John_Lin.png",
+        "Name": "John",
+        "Profile": "Software Developer"
+    },
+    {
+        "Icon": "Tamara_Taylor.png",
+        "Name": "Tamara",
+        "Profile": "Music Composer"
+    },
+    {
+        "Icon": "Arthur_Burton.png",
+        "Name": "Arthur",
+        "Profile": "Neuroscientist"
+    },
+    {
+        "Icon": "Eddy_Lin.png",
+        "Name": "Eddy",
+        "Profile": "Cognitive Psychologist"
+    },
+    {
+        "Icon": "Isabella_Rodriguez.png",
+        "Name": "Isabella",
+        "Profile": "Science Fiction Writer"
+    },
+    {
+        "Icon": "Latoya_Williams.png",
+        "Name": "Latoya",
+        "Profile": "Historian of Technology"
+    },
+    {
+        "Icon": "Carmen_Ortiz.png",
+        "Name": "Carmen",
+        "Profile": "Robotics Engineer"
+    },
+    {
+        "Icon": "Rajiv_Patel.png",
+        "Name": "Rajiv",
+        "Profile": "Science Educator"
+    },
+    {
+        "Icon": "Tom_Moreno.png",
+        "Name": "Tom",
+        "Profile": "AI Scientist"
+    },
+    {
+        "Icon": "Ayesha_Khan.png",
+        "Name": "Ayesha",
+        "Profile": "Multimedia Artist"
+    },
+    {
+        "Icon": "Mei_Lin.png",
+        "Name": "Mei",
+        "Profile": "Graphic Designer"
+    },
+    {
+        "Icon": "Hailey_Johnson.png",
+        "Name": "Hailey",
+        "Profile": "Legal Expert on AI Law" 
+    }
+]

backend/requirements.txt

@@ -4,4 +4,7 @@ PyYAML==6.0.1
 termcolor==2.4.0
 groq==0.4.2
 mistralai==0.1.6
-socksio==1.0.0
+flask-socketio==5.3.6
+python-socketio==5.11.0
+simple-websocket==1.0.0
+eventlet==0.40.4

frontend/components.d.ts

@@ -16,18 +16,19 @@ declare module 'vue' {
     ElCard: typeof import('element-plus/es')['ElCard']
     ElCollapse: typeof import('element-plus/es')['ElCollapse']
     ElCollapseItem: typeof import('element-plus/es')['ElCollapseItem']
-    ElDialog: typeof import('element-plus/es')['ElDialog']
-    ElDivider: typeof import('element-plus/es')['ElDivider']
     ElDrawer: typeof import('element-plus/es')['ElDrawer']
     ElEmpty: typeof import('element-plus/es')['ElEmpty']
+    ElIcon: typeof import('element-plus/es')['ElIcon']
     ElInput: typeof import('element-plus/es')['ElInput']
     ElPopover: typeof import('element-plus/es')['ElPopover']
     ElScrollbar: typeof import('element-plus/es')['ElScrollbar']
     ElTooltip: typeof import('element-plus/es')['ElTooltip']
     MultiLineTooltip: typeof import('./src/components/MultiLineTooltip/index.vue')['default']
+    Notification: typeof import('./src/components/Notification/Notification.vue')['default']
     RouterLink: typeof import('vue-router')['RouterLink']
     RouterView: typeof import('vue-router')['RouterView']
     SvgIcon: typeof import('./src/components/SvgIcon/index.vue')['default']
+    TaskContentEditor: typeof import('./src/components/TaskContentEditor/index.vue')['default']
   }
   export interface GlobalDirectives {
     vLoading: typeof import('element-plus/es')['ElLoadingDirective']

frontend/package.json

@@ -32,6 +32,7 @@
     "markdown-it": "^14.1.0",
     "pinia": "^3.0.3",
     "qs": "^6.14.0",
+    "socket.io-client": "^4.8.3",
     "uuid": "^13.0.0",
     "vue": "^3.5.22",
     "vue-router": "^4.6.3"

frontend/pnpm-lock.yaml

@@ -53,6 +53,9 @@ importers:
       qs:
         specifier: ^6.14.0
         version: 6.14.0
+      socket.io-client:
+        specifier: ^4.8.3
+        version: 4.8.3
       uuid:
         specifier: ^13.0.0
         version: 13.0.0
@@ -834,6 +837,9 @@ packages:
   '@rtsao/scc@1.1.0':
     resolution: {integrity: sha512-zt6OdqaDoOnJ1ZYsCYGt9YmWzDXl4vQdKTyJev62gFhRGKdx7mcT54V9KIjg+d2wi9EXsPvAPKe7i7WjfVWB8g==}
 
+  '@socket.io/component-emitter@3.1.2':
+    resolution: {integrity: sha512-9BCxFwvbGg/RsZK9tjXd8s4UcwR0MWeFQ1XEKIQVVvAGJyINdrqKMcTRyLoK8Rse1GjzLV9cwjWV1olXRWEXVA==}
+
   '@sxzz/popperjs-es@2.11.7':
     resolution: {integrity: sha512-Ccy0NlLkzr0Ex2FKvh2X+OyERHXJ88XJ1MXtsI9y9fGexlaXaVTPzBCRBwIxFkORuOb+uBqeu+RqnpgYTEZRUQ==}
 
@@ -1793,6 +1799,13 @@ packages:
     resolution: {integrity: sha512-/kyM18EfinwXZbno9FyUGeFh87KC8HRQBQGildHZbEuRyWFOmv1U10o9BBp8XVZDVNNuQKyIGIu5ZYAAXJ0V2Q==}
     engines: {node: '>= 4'}
 
+  engine.io-client@6.6.4:
+    resolution: {integrity: sha512-+kjUJnZGwzewFDw951CDWcwj35vMNf2fcj7xQWOctq1F2i1jkDdVvdFG9kM/BEChymCH36KgjnW0NsL58JYRxw==}
+
+  engine.io-parser@5.2.3:
+    resolution: {integrity: sha512-HqD3yTBfnBxIrbnM1DoD6Pcq8NECnh8d4As1Qgh0z5Gg3jRRIqijury0CL3ghu/edArpUYiYqQiDUQBIs4np3Q==}
+    engines: {node: '>=10.0.0'}
+
   enhanced-resolve@5.18.4:
     resolution: {integrity: sha512-LgQMM4WXU3QI+SYgEc2liRgznaD5ojbmY3sb8LxyguVkIg5FxdpTkvk72te2R38/TGKxH634oLxXRGY6d7AP+Q==}
     engines: {node: '>=10.13.0'}
@@ -3247,6 +3260,14 @@ packages:
     resolution: {integrity: sha512-FtyOnWN/wCHTVXOMwvSv26d+ko5vWlIDD6zoUJ7LW8vh+ZBC8QdljveRP+crNrtBwioEUWy/4dMtbBjA4ioNlg==}
     engines: {node: '>=0.10.0'}
 
+  socket.io-client@4.8.3:
+    resolution: {integrity: sha512-uP0bpjWrjQmUt5DTHq9RuoCBdFJF10cdX9X+a368j/Ft0wmaVgxlrjvK3kjvgCODOMMOz9lcaRzxmso0bTWZ/g==}
+    engines: {node: '>=10.0.0'}
+
+  socket.io-parser@4.2.5:
+    resolution: {integrity: sha512-bPMmpy/5WWKHea5Y/jYAP6k74A+hvmRCQaJuJB6I/ML5JZq/KfNieUVo/3Mh7SAqn7TyFdIo6wqYHInG1MU1bQ==}
+    engines: {node: '>=10.0.0'}
+
   source-map-js@1.2.1:
     resolution: {integrity: sha512-UXWMKhLOwVKb728IUtQPXxfYU+usdybtUrK/8uGE8CQMvrhOpwvzDBwj0QhSL7MQc7vIsISBG8VQ8+IDQxpfQA==}
     engines: {node: '>=0.10.0'}
@@ -3825,6 +3846,10 @@ packages:
   xmlchars@2.2.0:
     resolution: {integrity: sha512-JZnDKK8B0RCDw84FNdDAIpZK+JuJw+s7Lz8nksI7SIuU3UXJJslUthsi+uWBUYOwPFwW7W7PRLRfUKpxjtjFCw==}
 
+  xmlhttprequest-ssl@2.1.2:
+    resolution: {integrity: sha512-TEU+nJVUUnA4CYJFLvK5X9AOeH4KvDvhIfm0vV1GaQRtchnG0hgK5p8hw/xjv8cunWYCsiPCSDzObPyhEwq3KQ==}
+    engines: {node: '>=0.4.0'}
+
   yallist@3.1.1:
     resolution: {integrity: sha512-a4UGQaWPH59mOXUYnAG2ewncQS4i4F43Tv3JoAM+s2VDAmS9NsK8GpDMLrCHPksFT7h3K6TOoUNn2pb7RoXx4g==}
 
@@ -4403,6 +4428,8 @@ snapshots:
 
   '@rtsao/scc@1.1.0': {}
 
+  '@socket.io/component-emitter@3.1.2': {}
+
   '@sxzz/popperjs-es@2.11.7': {}
 
   '@tailwindcss/node@4.1.18':
@@ -5496,6 +5523,20 @@ snapshots:
 
   emojis-list@3.0.0: {}
 
+  engine.io-client@6.6.4:
+    dependencies:
+      '@socket.io/component-emitter': 3.1.2
+      debug: 4.4.3
+      engine.io-parser: 5.2.3
+      ws: 8.18.3
+      xmlhttprequest-ssl: 2.1.2
+    transitivePeerDependencies:
+      - bufferutil
+      - supports-color
+      - utf-8-validate
+
+  engine.io-parser@5.2.3: {}
+
   enhanced-resolve@5.18.4:
     dependencies:
       graceful-fs: 4.2.11
@@ -7074,6 +7115,24 @@ snapshots:
     transitivePeerDependencies:
       - supports-color
 
+  socket.io-client@4.8.3:
+    dependencies:
+      '@socket.io/component-emitter': 3.1.2
+      debug: 4.4.3
+      engine.io-client: 6.6.4
+      socket.io-parser: 4.2.5
+    transitivePeerDependencies:
+      - bufferutil
+      - supports-color
+      - utf-8-validate
+
+  socket.io-parser@4.2.5:
+    dependencies:
+      '@socket.io/component-emitter': 3.1.2
+      debug: 4.4.3
+    transitivePeerDependencies:
+      - supports-color
+
   source-map-js@1.2.1: {}
 
   source-map-resolve@0.5.3:
@@ -7745,6 +7804,8 @@ snapshots:
 
   xmlchars@2.2.0: {}
 
+  xmlhttprequest-ssl@2.1.2: {}
+
   yallist@3.1.1: {}
 
   yocto-queue@0.1.0: {}

frontend/public/config.json

@@ -4,7 +4,7 @@
   "centerTitle": "多智能体协同平台",
   "taskPromptWords": [
     "如何快速筛选慢性肾脏病药物潜在受试者？",
-    "如何补充“丹芍活血胶囊”不良反应数据？",
+    "如何补充\"丹芍活血胶囊\"不良反应数据？",
     "如何快速研发用于战场失血性休克的药物？",
     "二维材料的光电性质受哪些关键因素影响？",
     "如何通过AI模拟的方法分析材料的微观结构?",
@@ -16,5 +16,6 @@
   "agentRepository": {
     "storageVersionIdentifier": "1"
   },
-  "dev": true
+  "dev": true,
+  "apiBaseUrl": "http://localhost:8000"
 }

frontend/src/api/index.ts

@@ -1,20 +1,7 @@
 import request from '@/utils/request'
+import websocket from '@/utils/websocket'
 import type { Agent, IApiStepTask, IRawPlanResponse, IRawStepTask } from '@/stores'
-import {
+import { withRetry } from '@/utils/retry'
-  mockBackendAgentSelectModifyInit,
-  mockBackendAgentSelectModifyAddAspect,
-  type BackendAgentScoreResponse,
-} from '@/layout/components/Main/TaskTemplate/TaskSyllabus/components/mock/AgentAssignmentBackendMock'
-import {
-  mockBackendFillAgentTaskProcess,
-  type RawAgentTaskProcessResponse,
-} from '@/layout/components/Main/TaskTemplate/TaskProcess/components/mock/AgentTaskProcessBackendMock'
-import { mockBranchPlanOutlineAPI } from '@/layout/components/Main/TaskTemplate/TaskSyllabus/Branch/mock/branchPlanOutlineMock'
-import { mockFillStepTaskAPI } from '@/layout/components/Main/TaskTemplate/TaskSyllabus/Branch/mock/fill-step-task-mock'
-import {
-  mockBranchTaskProcessAPI,
-  type BranchAction,
-} from '@/layout/components/Main/TaskTemplate/TaskSyllabus/Branch/mock/branchTaskProcessMock'
 
 export interface ActionHistory {
   ID: string
@@ -25,6 +12,14 @@ export interface ActionHistory {
   Action_Result: string
 }
 
+export interface BranchAction {
+  ID: string
+  ActionType: string
+  AgentName: string
+  Description: string
+  ImportantInput: string[]
+}
+
 export type IExecuteRawResponse = {
   LogNodeType: string
   NodeId: string
@@ -34,6 +29,47 @@ export type IExecuteRawResponse = {
   ActionHistory: ActionHistory[]
 }
 
+/**
+ * SSE 流式事件类型
+ */
+export type StreamingEvent =
+  | {
+      type: 'step_start'
+      step_index: number
+      total_steps: number
+      step_name: string
+      task_description?: string
+    }
+  | {
+      type: 'action_complete'
+      step_index: number
+      step_name: string
+      action_index: number
+      total_actions: number
+      completed_actions: number
+      action_result: ActionHistory
+      batch_info?: {
+        batch_index: number
+        batch_size: number
+        is_parallel: boolean
+      }
+    }
+  | {
+      type: 'step_complete'
+      step_index: number
+      step_name: string
+      step_log_node: any
+      object_log_node: any
+    }
+  | {
+      type: 'execution_complete'
+      total_steps: number
+    }
+  | {
+      type: 'error'
+      message: string
+    }
+
 export interface IFillAgentSelectionRequest {
   goal: string
   stepTask: IApiStepTask
@@ -41,7 +77,19 @@ export interface IFillAgentSelectionRequest {
 }
 
 class Api {
-  setAgents = (data: Pick<Agent, 'Name' | 'Profile' | 'apiUrl' | 'apiKey' | 'apiModel'>[]) => {
+  // 默认使用WebSocket
+  private useWebSocketDefault = true
+
+  setAgents = (
+    data: Pick<Agent, 'Name' | 'Profile' | 'apiUrl' | 'apiKey' | 'apiModel'>[],
+    useWebSocket: boolean = this.useWebSocketDefault,
+  ) => {
+    // 如果启用WebSocket且已连接，使用WebSocket
+    if (useWebSocket && websocket.connected) {
+      return websocket.send('set_agents', data)
+    }
+
+    // 否则使用REST API
     return request({
       url: '/setAgents',
       data,
@@ -55,7 +103,33 @@ class Api {
     apiUrl?: string
     apiKey?: string
     apiModel?: string
+    useWebSocket?: boolean
+    onProgress?: (progress: {
+      status: string
+      stage?: string
+      message?: string
+      [key: string]: any
+    }) => void
   }) => {
+    const useWs = data.useWebSocket !== undefined ? data.useWebSocket : this.useWebSocketDefault
+
+    // 如果启用WebSocket且已连接，使用WebSocket
+    if (useWs && websocket.connected) {
+      return websocket.send(
+        'generate_base_plan',
+        {
+          'General Goal': data.goal,
+          'Initial Input Object': data.inputs,
+          apiUrl: data.apiUrl,
+          apiKey: data.apiKey,
+          apiModel: data.apiModel,
+        },
+        undefined,
+        data.onProgress,
+      )
+    }
+
+    // 否则使用REST API
     return request<unknown, IRawPlanResponse>({
       url: '/generate_basePlan',
       method: 'POST',
@@ -99,15 +173,190 @@ class Api {
     })
   }
 
-  // 分支任务大纲（根节点级别）
+  /**
+   * 优化版流式执行计划（支持动态追加步骤）
+   * 步骤级流式 + 动作级智能并行 + 动态追加步骤
+   */
+  executePlanOptimized = (
+    plan: IRawPlanResponse,
+    onMessage: (event: StreamingEvent) => void,
+    onError?: (error: Error) => void,
+    onComplete?: () => void,
+    useWebSocket?: boolean,
+    existingKeyObjects?: Record<string, any>,
+    enableDynamic?: boolean,
+    onExecutionStarted?: (executionId: string) => void,
+    executionId?: string,
+    restartFromStepIndex?: number, // 新增：从指定步骤重新执行的索引
+    rehearsalLog?: any[], // 新增：传递截断后的 RehearsalLog
+  ) => {
+    const useWs = useWebSocket !== undefined ? useWebSocket : this.useWebSocketDefault
+
+    const data = {
+      RehearsalLog: rehearsalLog || [], // ✅ 使用传递的 RehearsalLog
+      num_StepToRun: null,
+      existingKeyObjects: existingKeyObjects || {},
+      enable_dynamic: enableDynamic || false,
+      execution_id: executionId || null,
+      restart_from_step_index: restartFromStepIndex ?? null, // 新增：传递重新执行索引
+      plan: {
+        'Initial Input Object': plan['Initial Input Object'],
+        'General Goal': plan['General Goal'],
+        'Collaboration Process': plan['Collaboration Process']?.map((step) => ({
+          StepName: step.StepName,
+          TaskContent: step.TaskContent,
+          InputObject_List: step.InputObject_List,
+          OutputObject: step.OutputObject,
+          AgentSelection: step.AgentSelection,
+          Collaboration_Brief_frontEnd: step.Collaboration_Brief_frontEnd,
+          TaskProcess: step.TaskProcess.map((action) => ({
+            ActionType: action.ActionType,
+            AgentName: action.AgentName,
+            Description: action.Description,
+            ID: action.ID,
+            ImportantInput: action.ImportantInput,
+          })),
+        })),
+      },
+    }
+
+    // 如果启用WebSocket且已连接，使用WebSocket
+    if (useWs && websocket.connected) {
+      websocket.subscribe(
+        'execute_plan_optimized',
+        data,
+        // onProgress
+        (progressData) => {
+          try {
+            let event: StreamingEvent
+
+            // 处理不同类型的progress数据
+            if (typeof progressData === 'string') {
+              event = JSON.parse(progressData)
+            } else {
+              event = progressData as StreamingEvent
+            }
+
+            // 处理特殊事件类型
+            if (event && typeof event === 'object') {
+              // 检查是否是execution_started事件
+              if ('status' in event && event.status === 'execution_started') {
+                if ('execution_id' in event && onExecutionStarted) {
+                  onExecutionStarted(event.execution_id as string)
+                }
+                return
+              }
+            }
+
+            onMessage(event)
+          } catch (e) {
+            // Failed to parse WebSocket data
+          }
+        },
+        // onComplete
+        () => {
+          onComplete?.()
+        },
+        // onError
+        (error) => {
+          onError?.(error)
+        },
+      )
+      return
+    }
+
+    // 否则使用原有的SSE方式
+
+    fetch('/api/executePlanOptimized', {
+      method: 'POST',
+      headers: {
+        'Content-Type': 'application/json',
+      },
+      body: JSON.stringify(data),
+    })
+      .then(async (response) => {
+        if (!response.ok) {
+          throw new Error(`HTTP error! status: ${response.status}`)
+        }
+
+        const reader = response.body?.getReader()
+        const decoder = new TextDecoder()
+
+        if (!reader) {
+          throw new Error('Response body is null')
+        }
+
+        let buffer = ''
+
+        while (true) {
+          const { done, value } = await reader.read()
+
+          if (done) {
+            onComplete?.()
+            break
+          }
+
+          buffer += decoder.decode(value, { stream: true })
+
+          const lines = buffer.split('\n')
+          buffer = lines.pop() || ''
+
+          for (const line of lines) {
+            if (line.startsWith('data: ')) {
+              const data = line.slice(6)
+              try {
+                const event = JSON.parse(data)
+                onMessage(event)
+              } catch (e) {
+                // Failed to parse SSE data
+              }
+            }
+          }
+        }
+      })
+      .catch((error) => {
+        onError?.(error)
+      })
+  }
+
+  /**
+   * 分支任务大纲
+   */
   branchPlanOutline = (data: {
     branch_Number: number
     Modification_Requirement: string
-    Existing_Steps: string[]
+    Existing_Steps: IRawStepTask[]
     Baseline_Completion: number
     initialInputs: string[]
     goal: string
+    useWebSocket?: boolean
+    onProgress?: (progress: {
+      status: string
+      stage?: string
+      message?: string
+      [key: string]: any
+    }) => void
   }) => {
+    const useWs = data.useWebSocket !== undefined ? data.useWebSocket : this.useWebSocketDefault
+
+    // 如果启用WebSocket且已连接，使用WebSocket
+    if (useWs && websocket.connected) {
+      return websocket.send(
+        'branch_plan_outline',
+        {
+          branch_Number: data.branch_Number,
+          Modification_Requirement: data.Modification_Requirement,
+          Existing_Steps: data.Existing_Steps,
+          Baseline_Completion: data.Baseline_Completion,
+          'Initial Input Object': data.initialInputs,
+          'General Goal': data.goal,
+        },
+        undefined,
+        data.onProgress,
+      )
+    }
+
+    // 否则使用REST API
     return request<unknown, IRawPlanResponse>({
       url: '/branch_PlanOutline',
       method: 'POST',
@@ -122,15 +371,44 @@ class Api {
     })
   }
 
-  // 分支任务流程
+  /**
+   * 分支任务流程
+   */
   branchTaskProcess = (data: {
     branch_Number: number
     Modification_Requirement: string
-    Existing_Steps: string[]
+    Existing_Steps: BranchAction[]
     Baseline_Completion: number
     stepTaskExisting: any
     goal: string
+    useWebSocket?: boolean
+    onProgress?: (progress: {
+      status: string
+      stage?: string
+      message?: string
+      [key: string]: any
+    }) => void
   }) => {
+    const useWs = data.useWebSocket !== undefined ? data.useWebSocket : this.useWebSocketDefault
+
+    // 如果启用WebSocket且已连接，使用WebSocket
+    if (useWs && websocket.connected) {
+      return websocket.send(
+        'branch_task_process',
+        {
+          branch_Number: data.branch_Number,
+          Modification_Requirement: data.Modification_Requirement,
+          Existing_Steps: data.Existing_Steps,
+          Baseline_Completion: data.Baseline_Completion,
+          stepTaskExisting: data.stepTaskExisting,
+          'General Goal': data.goal,
+        },
+        undefined,
+        data.onProgress,
+      )
+    }
+
+    // 否则使用REST API
     return request<unknown, BranchAction[][]>({
       url: '/branch_TaskProcess',
       method: 'POST',
@@ -145,9 +423,36 @@ class Api {
     })
   }
 
-  fillStepTask = async (data: { goal: string; stepTask: any }): Promise<IRawStepTask> => {
+  fillStepTask = async (data: {
-    // 后端返回格式：包含 Collaboration_Brief_FrontEnd（大写 FrontEnd）
+    goal: string
-    const response = await request<
+    stepTask: any
+    generation_id?: string
+    useWebSocket?: boolean
+    onProgress?: (progress: {
+      status: string
+      stage?: string
+      message?: string
+      [key: string]: any
+    }) => void
+  }): Promise<IRawStepTask> => {
+    const useWs = data.useWebSocket !== undefined ? data.useWebSocket : this.useWebSocketDefault
+
+    // 定义实际的 API 调用逻辑
+    const executeRequest = async (): Promise<any> => {
+      if (useWs && websocket.connected) {
+        return await websocket.send(
+          'fill_step_task',
+          {
+            'General Goal': data.goal,
+            stepTask: data.stepTask,
+            generation_id: data.generation_id || '',
+          },
+          undefined,
+          data.onProgress,
+        )
+      }
+      // 否则使用REST API
+      return await request<
         {
           'General Goal': string
           stepTask: any
@@ -178,28 +483,41 @@ class Api {
           stepTask: data.stepTask,
         },
       })
+    }
 
-    // 数据转换：后端的 Collaboration_Brief_FrontEnd → 前端的 Collaboration_Brief_frontEnd
+    // 使用重试机制执行请求
+    const rawResponse = await withRetry(executeRequest, {
+      maxRetries: 3,
+      initialDelayMs: 2000,
+      onRetry: (error, attempt, delay) => {
+        console.warn(`⚠️ [fillStepTask] 第${attempt}次重试，等待 ${delay}ms...`, error?.message)
+      },
+    })
+
+    // WebSocket 返回格式: { data: {...}, generation_id, execution_id }
+    // REST API 返回格式: {...}
+    const response = rawResponse.data || rawResponse
 
     const vec2Hsl = (color: number[]): string => {
       const [h, s, l] = color
       return `hsl(${h}, ${s}%, ${l}%)`
     }
 
-    // 转换 brief.data：后端格式 { "0": { text, color: [h,s,l] } } → 前端格式 { "0": { text, style: { background } } }
     const briefData: Record<string, { text: string; style?: Record<string, string> }> = {}
     if (response.Collaboration_Brief_FrontEnd?.data) {
       for (const [key, value] of Object.entries(response.Collaboration_Brief_FrontEnd.data)) {
         briefData[key] = {
-          text: value.text,
+          text: (value as { text: string; color: number[] }).text,
           style: {
-            background: vec2Hsl(value.color),
+            background: vec2Hsl((value as { text: string; color: number[] }).color),
           },
         }
       }
     }
 
-    // 构建前端格式的 IRawStepTask
+    /**
+     * 构建前端格式的 IRawStepTask
+     */
     return {
       StepName: response.StepName || '',
       TaskContent: response.TaskContent || '',
@@ -218,9 +536,37 @@ class Api {
     goal: string
     stepTask: IApiStepTask
     agents: string[]
+    useWebSocket?: boolean
+    onProgress?: (progress: {
+      status: string
+      stage?: string
+      message?: string
+      [key: string]: any
+    }) => void
   }): Promise<IApiStepTask> => {
-    // 后端返回格式: IRawStepTask
+    const useWs = data.useWebSocket !== undefined ? data.useWebSocket : this.useWebSocketDefault
-    const response = await request<
+
+    // 定义实际的 API 调用逻辑
+    const executeRequest = async (): Promise<any> => {
+      if (useWs && websocket.connected) {
+        return await websocket.send(
+          'fill_step_task_process',
+          {
+            'General Goal': data.goal,
+            stepTask_lackTaskProcess: {
+              StepName: data.stepTask.name,
+              TaskContent: data.stepTask.content,
+              InputObject_List: data.stepTask.inputs,
+              OutputObject: data.stepTask.output,
+              AgentSelection: data.agents,
+            },
+          },
+          undefined,
+          data.onProgress,
+        )
+      }
+      // 否则使用REST API
+      return await request<
         {
           'General Goal': string
           stepTask_lackTaskProcess: {
@@ -263,31 +609,42 @@ class Api {
           },
         },
       })
+    }
 
-    // 数据转换：后端格式 (IRawStepTask) → 前端格式 (IApiStepTask)
+    // 使用重试机制执行请求
-    // 注意：此转换逻辑与Mock API完全一致
+    const rawResponse = await withRetry(executeRequest, {
+      maxRetries: 3,
+      initialDelayMs: 2000,
+      onRetry: (error, attempt, delay) => {
+        console.warn(
+          `⚠️ [fillStepTaskTaskProcess] 第${attempt}次重试，等待 ${delay}ms...`,
+          error?.message,
+        )
+      },
+    })
+
+    // WebSocket 返回格式: { data: {...}, generation_id, execution_id }
+    // REST API 返回格式: {...}
+    const response = rawResponse.data || rawResponse
 
-    // 1. 转换颜色格式：[h, s, l] → "hsl(h, s%, l%)"
     const vec2Hsl = (color: number[]): string => {
       const [h, s, l] = color
       return `hsl(${h}, ${s}%, ${l}%)`
     }
 
-    // 2. 转换 brief.data: { "0": { text, color: [h,s,l] } } → { "0": { text, style: { background } } }
     const briefData: Record<string, { text: string; style: { background: string } }> = {}
     if (response.Collaboration_Brief_FrontEnd?.data) {
       for (const [key, value] of Object.entries(response.Collaboration_Brief_FrontEnd.data)) {
         briefData[key] = {
-          text: value.text,
+          text: (value as { text: string; color: number[] }).text,
           style: {
-            background: vec2Hsl(value.color),
+            background: vec2Hsl((value as { text: string; color: number[] }).color),
           },
         }
       }
     }
 
-    // 3. 转换 process: { ID, ActionType, AgentName, Description, ImportantInput } → { id, type, agent, description, inputs }
+    const process = (response.TaskProcess || []).map((action: any) => ({
-    const process = (response.TaskProcess || []).map((action) => ({
       id: action.ID,
       type: action.ActionType,
       agent: action.AgentName,
@@ -295,7 +652,6 @@ class Api {
       inputs: action.ImportantInput,
     }))
 
-    // 4. 构建前端格式的 IApiStepTask
     return {
       name: response.StepName || '',
       content: response.TaskContent || '',
@@ -310,13 +666,54 @@ class Api {
     }
   }
 
-  // 为每个智能体评分
+  /**
+   * 为每个智能体评分（带重试机制）
+   */
   agentSelectModifyInit = async (data: {
     goal: string
     stepTask: any
+    useWebSocket?: boolean
+    onProgress?: (progress: {
+      status: string
+      stage?: string
+      message?: string
+      [key: string]: any
+    }) => void
   }): Promise<Record<string, Record<string, { reason: string; score: number }>>> => {
-    // 后端返回：维度 -> agent -> { Reason, Score }
+    const useWs = data.useWebSocket !== undefined ? data.useWebSocket : this.useWebSocketDefault
-    const response = await request<
+
+    // 调试日志：打印请求参数
+    const requestPayload = {
+      'General Goal': data.goal,
+      stepTask: {
+        StepName: data.stepTask.StepName || data.stepTask.name,
+        TaskContent: data.stepTask.TaskContent || data.stepTask.content,
+        InputObject_List: data.stepTask.InputObject_List || data.stepTask.inputs,
+        OutputObject: data.stepTask.OutputObject || data.stepTask.output,
+      },
+    }
+    console.log('🔍 [agentSelectModifyInit] 请求参数:', {
+      goal: requestPayload['General Goal'],
+      stepTaskName: requestPayload.stepTask.StepName,
+      stepTaskContentLength: requestPayload.stepTask.TaskContent?.length,
+      useWebSocket: useWs && websocket.connected,
+      wsConnected: websocket.connected,
+    })
+
+    // 定义实际的 API 调用逻辑
+    const executeRequest = async (): Promise<
+      Record<string, Record<string, { Reason: string; Score: number }>>
+    > => {
+      if (useWs && websocket.connected) {
+        return await websocket.send(
+          'agent_select_modify_init',
+          requestPayload,
+          undefined,
+          data.onProgress,
+        )
+      }
+      // 否则使用REST API
+      return await request<
         {
           'General Goal': string
           stepTask: any
@@ -325,23 +722,36 @@ class Api {
       >({
         url: '/agentSelectModify_init',
         method: 'POST',
-      data: {
+        data: requestPayload,
-        'General Goal': data.goal,
+      })
-        stepTask: {
+    }
-          StepName: data.stepTask.StepName || data.stepTask.name,
+
-          TaskContent: data.stepTask.TaskContent || data.stepTask.content,
+    // 使用重试机制执行请求
-          InputObject_List: data.stepTask.InputObject_List || data.stepTask.inputs,
+    const rawResponse = await withRetry(executeRequest, {
-          OutputObject: data.stepTask.OutputObject || data.stepTask.output,
+      maxRetries: 3,
-        },
+      initialDelayMs: 2000,
+      onRetry: (error, attempt, delay) => {
+        console.warn(
+          `⚠️ [agentSelectModifyInit] 第${attempt}次重试，等待 ${delay}ms...`,
+          error?.message,
+        )
       },
     })
 
-    // 数据转换：后端格式 (维度->agent) → 前端格式 (agent->维度)
+    // WebSocket 返回格式: { data: {...}, generation_id, execution_id }
+    // REST API 返回格式: {...}
+    const response = rawResponse.data || rawResponse
+
     const transformedData: Record<string, Record<string, { reason: string; score: number }>> = {}
 
+    // 确保 response 存在且是有效对象
+    if (!response || typeof response !== 'object' || Array.isArray(response)) {
+      console.warn('[agentSelectModifyInit] 后端返回数据格式异常:', response)
+      return transformedData
+    }
+
     for (const [aspect, agents] of Object.entries(response)) {
-      // aspect: 维度名称, agents: { agentName: { Reason, Score } }
+      for (const [agentName, scoreInfo] of Object.entries(agents as Record<string, { Reason: string; Score: number }> || {})) {
-      for (const [agentName, scoreInfo] of Object.entries(agents)) {
         if (!transformedData[agentName]) {
           transformedData[agentName] = {}
         }
@@ -355,16 +765,40 @@ class Api {
     return transformedData
   }
 
-  // 添加新的评估维度
+  /**
-  // 定义返回类型（与 Mock 数据格式一致）
+   * 添加新的评估维度
+   */
   agentSelectModifyAddAspect = async (data: {
     aspectList: string[]
+    useWebSocket?: boolean
+    onProgress?: (progress: {
+      status: string
+      stage?: string
+      message?: string
+      [key: string]: any
+    }) => void
   }): Promise<{
     aspectName: string
     agentScores: Record<string, { score: number; reason: string }>
   }> => {
-    // 后端返回：维度 -> agent -> { Reason, Score }
+    const useWs = data.useWebSocket !== undefined ? data.useWebSocket : this.useWebSocketDefault
-    const response = await request<
+    let response: Record<string, Record<string, { Reason: string; Score: number }>>
+
+    // 如果启用WebSocket且已连接，使用WebSocket
+    if (useWs && websocket.connected) {
+      const rawResponse = await websocket.send(
+        'agent_select_modify_add_aspect',
+        {
+          aspectList: data.aspectList,
+        },
+        undefined,
+        data.onProgress,
+      )
+      // WebSocket 返回格式: { data: {...}, generation_id, execution_id }
+      response = rawResponse.data || rawResponse
+    } else {
+      // 否则使用REST API
+      response = await request<
         {
           aspectList: string[]
         },
@@ -376,14 +810,16 @@ class Api {
           aspectList: data.aspectList,
         },
       })
+    }
 
-    // 获取新添加的维度（最后一个）
+    /**
+     * 获取新添加的维度
+     */
     const newAspect = data.aspectList[data.aspectList.length - 1]
     if (!newAspect) {
       throw new Error('aspectList is empty')
     }
 
-    // 提取该维度的数据：维度 -> agent -> { Reason, Score } → agent -> { score, reason }
     const newAspectAgents = response[newAspect]
     const agentScores: Record<string, { score: number; reason: string }> = {}
 
@@ -401,187 +837,41 @@ class Api {
       agentScores,
     }
   }
-
+  /**
-  // ==================== Mock API（开发阶段使用）====================
+   * 向正在执行的任务追加新步骤
-  // Mock API 使用与真实 API 相同的数据转换逻辑，确保将来切换无缝
+   * @param executionId 执行ID
-
+   * @param newSteps 新步骤列表
-  // Mock: 为每个智能体评分
+   * @returns 追加的步骤数量
-  mockAgentSelectModifyInit = async (): Promise<
+   */
-    Record<string, Record<string, { reason: string; score: number }>>
+  addStepsToExecution = async (executionId: string, newSteps: IRawStepTask[]): Promise<number> => {
-  > => {
+    if (!websocket.connected) {
-    // 调用Mock后端数据（维度 -> agent -> { Reason, Score }）
+      throw new Error('WebSocket未连接')
-    const response: BackendAgentScoreResponse = await mockBackendAgentSelectModifyInit()
-
-    // 数据转换：后端格式 (维度->agent) → 前端格式 (agent->维度)
-    // 注意：此转换逻辑与真实API完全一致
-    const transformedData: Record<string, Record<string, { reason: string; score: number }>> = {}
-
-    for (const [aspect, agents] of Object.entries(response)) {
-      // aspect: 维度名称, agents: { agentName: { Reason, Score } }
-      for (const [agentName, scoreInfo] of Object.entries(agents)) {
-        if (!transformedData[agentName]) {
-          transformedData[agentName] = {}
-        }
-        transformedData[agentName][aspect] = {
-          reason: scoreInfo.Reason,
-          score: scoreInfo.Score,
-        }
-      }
     }
 
-    return transformedData
+    const rawResponse = await websocket.send('add_steps_to_execution', {
-  }
+      execution_id: executionId,
-
+      new_steps: newSteps.map((step) => ({
-  // Mock: 添加新的评估维度
+        StepName: step.StepName,
-  mockAgentSelectModifyAddAspect = async (data: {
+        TaskContent: step.TaskContent,
-    aspectList: string[]
+        InputObject_List: step.InputObject_List,
-  }): Promise<{
+        OutputObject: step.OutputObject,
-    aspectName: string
+        AgentSelection: step.AgentSelection,
-    agentScores: Record<string, { score: number; reason: string }>
+        Collaboration_Brief_frontEnd: step.Collaboration_Brief_frontEnd,
-  }> => {
+        TaskProcess: step.TaskProcess.map((action) => ({
-    // 调用Mock后端数据（维度 -> agent -> { Reason, Score }）
+          ActionType: action.ActionType,
-    const response: BackendAgentScoreResponse = await mockBackendAgentSelectModifyAddAspect(
+          AgentName: action.AgentName,
-      data.aspectList,
+          Description: action.Description,
-    )
+          ID: action.ID,
-
+          ImportantInput: action.ImportantInput,
-    // 获取新添加的维度（最后一个）
+        })),
-    const newAspect = data.aspectList[data.aspectList.length - 1]
+      })),
-    if (!newAspect) {
-      throw new Error('aspectList is empty')
-    }
-
-    // 提取该维度的数据：维度 -> agent -> { Reason, Score } → agent -> { score, reason }
-    // 注意：此转换逻辑与真实API完全一致
-    const newAspectAgents = response[newAspect]
-    const agentScores: Record<string, { score: number; reason: string }> = {}
-
-    if (newAspectAgents) {
-      for (const [agentName, scoreInfo] of Object.entries(newAspectAgents)) {
-        agentScores[agentName] = {
-          score: scoreInfo.Score,
-          reason: scoreInfo.Reason,
-        }
-      }
-    }
-
-    return {
-      aspectName: newAspect,
-      agentScores,
-    }
-  }
-
-  // Mock: 填充智能体任务流程
-  mockFillStepTaskTaskProcess = async (data: {
-    goal: string
-    stepTask: IApiStepTask
-    agents: string[]
-  }): Promise<IApiStepTask> => {
-    // 调用Mock后端数据（后端原始格式）
-    const response: RawAgentTaskProcessResponse = await mockBackendFillAgentTaskProcess(
-      data.goal,
-      data.stepTask,
-      data.agents,
-    )
-
-    // 数据转换：后端格式 → 前端格式
-    // 注意：此转换逻辑与真实API完全一致
-
-    // 1. 转换颜色格式：[h, s, l] → "hsl(h, s%, l%)"
-    const vec2Hsl = (color: number[]): string => {
-      const [h, s, l] = color
-      return `hsl(${h}, ${s}%, ${l}%)`
-    }
-
-    // 2. 转换 brief.data: { "0": { text, color: [h,s,l] } } → { "0": { text, style: { background } } }
-    const briefData: Record<string, { text: string; style: { background: string } }> = {}
-    if (response.Collaboration_Brief_frontEnd?.data) {
-      for (const [key, value] of Object.entries(response.Collaboration_Brief_frontEnd.data)) {
-        briefData[key] = {
-          text: value.text,
-          style: {
-            background: vec2Hsl(value.color),
-          },
-        }
-      }
-    }
-
-    // 3. 转换 process: { ID, ActionType, AgentName, Description, ImportantInput } → { id, type, agent, description, inputs }
-    const process = (response.TaskProcess || []).map((action) => ({
-      id: action.ID,
-      type: action.ActionType,
-      agent: action.AgentName,
-      description: action.Description,
-      inputs: action.ImportantInput,
-    }))
-
-    // 4. 构建前端格式的 IApiStepTask
-    return {
-      name: response.StepName || '',
-      content: response.TaskContent || '',
-      inputs: response.InputObject_List || [],
-      output: response.OutputObject || '',
-      agents: response.AgentSelection || [],
-      brief: {
-        template: response.Collaboration_Brief_frontEnd?.template || '',
-        data: briefData,
-      },
-      process,
-    }
-  }
-
-  // Mock: 分支任务大纲
-  mockBranchPlanOutline = async (data: {
-    branch_Number: number
-    Modification_Requirement: string
-    Existing_Steps: string[]
-    Baseline_Completion: number
-    initialInputs: string[]
-    goal: string
-  }): Promise<IRawPlanResponse> => {
-    // 直接调用 Mock API（已经返回 IRawPlanResponse 格式）
-    const response = await mockBranchPlanOutlineAPI({
-      branch_Number: data.branch_Number,
-      Modification_Requirement: data.Modification_Requirement,
-      Existing_Steps: data.Existing_Steps,
-      Baseline_Completion: data.Baseline_Completion,
-      InitialObject_List: data.initialInputs,
-      General_Goal: data.goal,
     })
 
-    return response
+    // WebSocket 返回格式: { data: {...}, generation_id, execution_id }
-  }
+    // REST API 返回格式: {...}
+    const response = (rawResponse.data || rawResponse) as { added_count: number }
 
-  // Mock: 填充任务流程
+    return response?.added_count || 0
-  mockFillStepTask = async (data: { goal: string; stepTask: any }): Promise<any> => {
-    // 直接调用 Mock API（已经返回 IRawStepTask 格式）
-    const response = await mockFillStepTaskAPI({
-      General_Goal: data.goal,
-      stepTask: data.stepTask,
-    })
-
-    return response
-  }
-
-  // Mock: 分支任务流程
-  mockBranchTaskProcess = async (data: {
-    branch_Number: number
-    Modification_Requirement: string
-    Existing_Steps: string[]
-    Baseline_Completion: number
-    stepTaskExisting: any
-    goal: string
-  }): Promise<BranchAction[][]> => {
-    // 直接调用 Mock API（已经返回 BranchAction[][] 格式，与后端完全一致）
-    const response = await mockBranchTaskProcessAPI({
-      branch_Number: data.branch_Number,
-      Modification_Requirement: data.Modification_Requirement,
-      Existing_Steps: data.Existing_Steps,
-      Baseline_Completion: data.Baseline_Completion,
-      stepTaskExisting: data.stepTaskExisting,
-      General_Goal: data.goal,
-    })
-
-    return response
   }
 }
 

frontend/src/assets/icons/Pause.svg

@@ -0,0 +1 @@
+<?xml version="1.0" standalone="no"?><!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd"><svg t="1769048650684" class="icon" viewBox="0 0 1024 1024" version="1.1" xmlns="http://www.w3.org/2000/svg" p-id="6190" xmlns:xlink="http://www.w3.org/1999/xlink" width="48" height="48"><path d="M874.058005 149.941995a510.06838 510.06838 0 1 0 109.740156 162.738976 511.396369 511.396369 0 0 0-109.740156-162.738976z m66.278708 362.178731A428.336713 428.336713 0 1 1 512 83.663287a428.698892 428.698892 0 0 1 428.336713 428.336713z" fill="#36404f" p-id="6191"></path><path d="M417.954256 281.533601a41.046923 41.046923 0 0 0-41.77128 40.201839v385.116718a41.892007 41.892007 0 0 0 83.663287 0v-385.116718a41.167649 41.167649 0 0 0-41.892007-40.201839zM606.045744 281.533601a41.046923 41.046923 0 0 0-41.77128 40.201839v385.116718a41.892007 41.892007 0 0 0 83.663287 0v-385.116718a41.167649 41.167649 0 0 0-41.892007-40.201839z" fill="#36404f" p-id="6192"></path></svg>

frontend/src/assets/icons/action.svg

@@ -1 +1,5 @@
-<?xml version="1.0" standalone="no"?><!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd"><svg t="1761736278335" class="icon" viewBox="0 0 1024 1024" version="1.1" xmlns="http://www.w3.org/2000/svg" p-id="5885" xmlns:xlink="http://www.w3.org/1999/xlink" width="200" height="200"><path d="M226.592 896C167.616 896 128 850.48 128 782.736V241.264C128 173.52 167.616 128 226.592 128c20.176 0 41.136 5.536 62.288 16.464l542.864 280.432C887.648 453.792 896 491.872 896 512s-8.352 58.208-64.272 87.088L288.864 879.536C267.712 890.464 246.768 896 226.592 896z m0-704.304c-31.008 0-34.368 34.656-34.368 49.568v541.472c0 14.896 3.344 49.568 34.368 49.568 9.6 0 20.88-3.2 32.608-9.248l542.864-280.432c21.904-11.328 29.712-23.232 29.712-30.608s-7.808-19.28-29.712-30.592L259.2 200.96c-11.728-6.048-23.008-9.264-32.608-9.264z"  p-id="5886"></path></svg>
+<?xml version="1.0" standalone="no"?>
+<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
+<svg t="1761736278335" class="icon" viewBox="0 0 1024 1024" version="1.1" xmlns="http://www.w3.org/2000/svg" p-id="5885" 
+xmlns:xlink="http://www.w3.org/1999/xlink" width="200" height="200">
+<path d="M226.592 896C167.616 896 128 850.48 128 782.736V241.264C128 173.52 167.616 128 226.592 128c20.176 0 41.136 5.536 62.288 16.464l542.864 280.432C887.648 453.792 896 491.872 896 512s-8.352 58.208-64.272 87.088L288.864 879.536C267.712 890.464 246.768 896 226.592 896z m0-704.304c-31.008 0-34.368 34.656-34.368 49.568v541.472c0 14.896 3.344 49.568 34.368 49.568 9.6 0 20.88-3.2 32.608-9.248l542.864-280.432c21.904-11.328 29.712-23.232 29.712-30.608s-7.808-19.28-29.712-30.592L259.2 200.96c-11.728-6.048-23.008-9.264-32.608-9.264z"  p-id="5886"></path></svg>

frontend/src/assets/icons/paper-plane.svg

@@ -1 +1,6 @@
-<?xml version="1.0" standalone="no"?><!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd"><svg t="1761204835005" class="icon" viewBox="0 0 1171 1024" version="1.1" xmlns="http://www.w3.org/2000/svg" p-id="5692" xmlns:xlink="http://www.w3.org/1999/xlink" width="228.7109375" height="200"><path d="M502.237757 1024 644.426501 829.679301 502.237757 788.716444 502.237757 1024 502.237757 1024ZM0 566.713817 403.967637 689.088066 901.485385 266.66003 515.916344 721.68034 947.825442 855.099648 1170.285714 0 0 566.713817 0 566.713817Z"  p-id="5693"></path></svg>
+<?xml version="1.0" standalone="no"?>
+<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
+<svg t="1761204835005" class="icon" viewBox="0 0 1171 1024" version="1.1" xmlns="http://www.w3.org/2000/svg"
+ p-id="5692" xmlns:xlink="http://www.w3.org/1999/xlink" width="228.7109375" height="200">
+ <path d="M502.237757 1024 644.426501 829.679301 502.237757 788.716444 502.237757 1024 502.237757 
+ 1024ZM0 566.713817 403.967637 689.088066 901.485385 266.66003 515.916344 721.68034 947.825442 855.099648 1170.285714 0 0 566.713817 0 566.713817Z"  p-id="5693"></path></svg>

frontend/src/assets/icons/stoprunning.svg

@@ -0,0 +1,7 @@
+<?xml version="1.0" standalone="no"?>
+<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
+<svg t="1768992484327" class="icon" viewBox="0 0 1024 1024" version="1.1" xmlns="http://www.w3.org/2000/svg"
+p-id="10530" xmlns:xlink="http://www.w3.org/1999/xlink">
+<path d="M512 853.333333c-187.733333 0-341.333333-153.6-341.333333-341.333333s153.6-341.333333 341.333333-341.333333
+ 341.333333 153.6 341.333333 341.333333-153.6 341.333333-341.333333 341.333333z m0-85.333333c140.8 0 256-115.2 256-256s-115.2-256-256-256-256
+  115.2-256 256 115.2 256 256 256z m-85.333333-341.333333h170.666666v170.666666h-170.666666v-170.666666z" fill="#ffffff" p-id="10531"></path></svg>

frontend/src/assets/icons/video-play.svg

@@ -0,0 +1 @@
+<?xml version="1.0" standalone="no"?><!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd"><svg t="1769048534610" class="icon" viewBox="0 0 1024 1024" version="1.1" xmlns="http://www.w3.org/2000/svg" p-id="4890" xmlns:xlink="http://www.w3.org/1999/xlink" width="48" height="48"><path d="M527.984 1001.6a480 480 0 1 1 480-480 480.384 480.384 0 0 1-480 480z m0-883.696A403.696 403.696 0 1 0 931.68 521.6 403.84 403.84 0 0 0 527.984 117.904z" fill="#36404f" p-id="4891"></path><path d="M473.136 729.6a47.088 47.088 0 0 1-18.112-3.888 38.768 38.768 0 0 1-23.056-34.992V384.384a39.632 39.632 0 0 1 23.056-34.992 46.016 46.016 0 0 1 43.632 3.888l211.568 153.168a38.72 38.72 0 0 1 16.464 31.104 37.632 37.632 0 0 1-16.464 31.104l-211.568 153.168a44.56 44.56 0 0 1-25.52 7.776z m41.168-266.704v149.296l102.896-74.64z" fill="#36404f" p-id="4892"></path></svg>

frontend/src/components/Notification/Notification.vue

@@ -0,0 +1,232 @@
+<template>
+  <teleport to="body">
+    <div class="notification-container">
+      <transition-group name="notification" tag="div" class="notification-list">
+        <div
+          v-for="notification in notifications"
+          :key="notification.id"
+          :class="['notification-item', `notification-${notification.type || 'info'}`]"
+          :style="{ zIndex: notification.zIndex || 1000 }"
+        >
+          <div class="notification-content">
+            <div class="notification-icon">
+              <component :is="getIcon(notification.type)" />
+            </div>
+            <div class="notification-message">
+              <div class="notification-title">{{ notification.title }}</div>
+              <div v-if="notification.detailTitle" class="notification-detail-title">
+                {{ notification.detailTitle }}
+              </div>
+              <div v-if="notification.detailMessage" class="notification-detail-desc">
+                {{ notification.detailMessage }}
+              </div>
+              <div v-else-if="notification.message" class="notification-desc">
+                {{ notification.message }}
+              </div>
+            </div>
+            <div class="notification-close" @click="close(notification.id)">
+              <Close />
+            </div>
+          </div>
+          <div v-if="notification.showProgress" class="notification-progress">
+            <div class="progress-bar" :style="{ width: `${notification.progress || 0}%` }"></div>
+          </div>
+        </div>
+      </transition-group>
+    </div>
+  </teleport>
+</template>
+
+<script setup lang="ts">
+import {
+  Close,
+  SuccessFilled as IconSuccess,
+  WarningFilled as IconWarning,
+  InfoFilled
+} from '@element-plus/icons-vue'
+import type { NotificationItem } from '@/composables/useNotification'
+
+const props = defineProps<{
+  notifications: NotificationItem[]
+}>()
+
+const emit = defineEmits<{
+  close: [id: string]
+}>()
+
+const close = (id: string) => {
+  emit('close', id)
+}
+
+const getIcon = (type?: string) => {
+  switch (type) {
+    case 'success':
+      return IconSuccess
+    case 'warning':
+      return IconWarning
+    case 'error':
+      return IconWarning
+    default:
+      return InfoFilled
+  }
+}
+</script>
+
+<style scoped>
+.notification-container {
+  position: fixed;
+  top: 20px;
+  right: 20px;
+  z-index: 9999;
+  pointer-events: none;
+}
+
+.notification-list {
+  display: flex;
+  flex-direction: column;
+  gap: 12px;
+}
+
+.notification-item {
+  pointer-events: auto;
+  min-width: 300px;
+  max-width: 450px;
+  background: white;
+  border-radius: 8px;
+  box-shadow: 0 4px 12px rgba(0, 0, 0, 0.15);
+  overflow: hidden;
+  border-left: 4px solid #409eff;
+}
+
+.notification-success {
+  border-left-color: #67c23a;
+}
+
+.notification-warning {
+  border-left-color: #e6a23c;
+}
+
+.notification-error {
+  border-left-color: #f56c6c;
+}
+
+.notification-content {
+  display: flex;
+  align-items: flex-start;
+  padding: 12px 16px;
+  gap: 12px;
+}
+
+.notification-icon {
+  flex-shrink: 0;
+  width: 20px;
+  height: 20px;
+  display: flex;
+  align-items: center;
+  justify-content: center;
+}
+
+.notification-message {
+  flex: 1;
+  min-width: 0;
+}
+
+.notification-title {
+  font-size: 14px;
+  font-weight: 500;
+  color: #303133;
+  margin-bottom: 4px;
+}
+
+.notification-detail-title {
+  font-size: 13px;
+  font-weight: 500;
+  color: #409eff;
+  margin-top: 4px;
+  margin-bottom: 2px;
+}
+
+.notification-detail-desc {
+  font-size: 12px;
+  color: #909399;
+  line-height: 1.4;
+}
+
+.notification-desc {
+  font-size: 13px;
+  color: #606266;
+  line-height: 1.5;
+}
+
+.notification-close {
+  flex-shrink: 0;
+  width: 20px;
+  height: 20px;
+  cursor: pointer;
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  color: #909399;
+  transition: color 0.2s;
+}
+
+.notification-close:hover {
+  color: #606266;
+}
+
+.notification-progress {
+  height: 2px;
+  background: #f0f2f5;
+  overflow: hidden;
+}
+
+.progress-bar {
+  height: 100%;
+  background: #409eff;
+  transition: width 0.3s ease;
+}
+
+/* 进入动画 */
+.notification-enter-active {
+  animation: slideInRight 0.3s ease-out;
+}
+
+/* 离开动画 */
+.notification-leave-active {
+  animation: slideOutRight 0.3s ease-in;
+}
+
+@keyframes slideInRight {
+  from {
+    opacity: 0;
+    transform: translateX(100%);
+  }
+  to {
+    opacity: 1;
+    transform: translateX(0);
+  }
+}
+
+@keyframes slideOutRight {
+  from {
+    opacity: 1;
+    transform: translateX(0);
+  }
+  to {
+    opacity: 0;
+    transform: translateX(100%);
+  }
+}
+
+/* 列表项移动动画 */
+.notification-move,
+.notification-enter-active,
+.notification-leave-active {
+  transition: all 0.3s ease;
+}
+
+.notification-leave-active {
+  position: absolute;
+  width: 100%;
+}
+</style>

frontend/src/components/TaskContentEditor/index.vue

@@ -0,0 +1,100 @@
+<script setup lang="ts">
+import { ref } from 'vue'
+import type { IRawStepTask } from '@/stores'
+import SvgIcon from '@/components/SvgIcon/index.vue'
+
+const props = defineProps<{
+  task: IRawStepTask
+}>()
+
+const emit = defineEmits<{
+  (e: 'save', taskId: string, content: string): void
+}>()
+
+const isEditing = ref(false)
+const editingContent = ref('')
+
+const startEditing = () => {
+  editingContent.value = props.task.TaskContent || ''
+  isEditing.value = true
+}
+
+const save = () => {
+  const trimmed = editingContent.value.trim()
+  emit('save', props.task.Id!, trimmed)
+  isEditing.value = false
+}
+
+const cancel = () => {
+  isEditing.value = false
+}
+
+const handleKeydown = (event: KeyboardEvent) => {
+  if (event.key === 'Enter') {
+    event.preventDefault()
+    save()
+  } else if (event.key === 'Escape') {
+    cancel()
+  }
+}
+</script>
+
+<template>
+  <div v-if="isEditing" class="w-full">
+    <div class="flex flex-col gap-3">
+      <el-input
+        v-model="editingContent"
+        type="textarea"
+        :autosize="{ minRows: 2, maxRows: 4 }"
+        placeholder="请输入任务内容"
+        @keydown="handleKeydown"
+        class="task-content-editor"
+        size="small"
+      />
+      <div class="flex justify-end">
+        <svg-icon
+          icon-class="Check"
+          size="20px"
+          color="#328621"
+          class="cursor-pointer mr-4"
+          @click="save"
+          title="保存"
+        />
+        <svg-icon
+          icon-class="Cancel"
+          size="20px"
+          color="#8e0707"
+          class="cursor-pointer mr-1"
+          @click="cancel"
+          title="取消"
+        />
+      </div>
+    </div>
+  </div>
+  <div v-else @dblclick="startEditing" class="w-full cursor-pointer">
+    <slot name="display">
+      <div class="text-[14px] text-[var(--color-text-secondary)] task-content-display">
+        {{ task.TaskContent }}
+      </div>
+    </slot>
+  </div>
+</template>
+
+<style scoped lang="scss">
+.task-content-editor {
+  :deep(.el-textarea__inner) {
+    font-size: 14px;
+    color: var(--color-text-secondary);
+    background: transparent;
+    border: 1px solid #dcdfe6;
+    border-radius: 4px;
+    resize: none;
+  }
+}
+
+.task-content-display {
+  min-height: 40px;
+  word-break: break-word;
+  white-space: pre-wrap;
+}
+</style>

frontend/src/composables/useNotification.ts

@@ -0,0 +1,163 @@
+import { ref } from 'vue'
+
+export interface NotificationItem {
+  id: string
+  title: string
+  message?: string
+  type?: 'success' | 'warning' | 'info' | 'error'
+  duration?: number
+  showProgress?: boolean
+  progress?: number
+  zIndex?: number
+  onClose?: () => void
+  // 详细进度信息
+  detailTitle?: string
+  detailMessage?: string
+}
+
+const notifications = ref<NotificationItem[]>([])
+let notificationIdCounter = 0
+let zIndexCounter = 1000
+
+export function useNotification() {
+  const addNotification = (notification: Omit<NotificationItem, 'id' | 'zIndex'>) => {
+    const id = `notification-${notificationIdCounter++}`
+    const newNotification: NotificationItem = {
+      ...notification,
+      id,
+      zIndex: ++zIndexCounter,
+    }
+
+    notifications.value.push(newNotification)
+
+    // 自动关闭
+    if (notification.duration && notification.duration > 0) {
+      setTimeout(() => {
+        removeNotification(id)
+      }, notification.duration)
+    }
+
+    return id
+  }
+
+  const removeNotification = (id: string) => {
+    const notification = notifications.value.find((n) => n.id === id)
+    if (notification) {
+      const index = notifications.value.indexOf(notification)
+      notifications.value.splice(index, 1)
+      notification.onClose?.()
+    }
+  }
+
+  const success = (title: string, message?: string, options?: Partial<NotificationItem>) => {
+    return addNotification({
+      title,
+      message,
+      type: 'success',
+      duration: 3000,
+      ...options,
+    })
+  }
+
+  const warning = (title: string, message?: string, options?: Partial<NotificationItem>) => {
+    return addNotification({
+      title,
+      message,
+      type: 'warning',
+      duration: 3000,
+      ...options,
+    })
+  }
+
+  const info = (title: string, message?: string, options?: Partial<NotificationItem>) => {
+    return addNotification({
+      title,
+      message,
+      type: 'info',
+      duration: 3000,
+      ...options,
+    })
+  }
+
+  const error = (title: string, message?: string, options?: Partial<NotificationItem>) => {
+    return addNotification({
+      title,
+      message,
+      type: 'error',
+      duration: 5000,
+      ...options,
+    })
+  }
+
+  const progress = (
+    title: string,
+    current: number,
+    total: number,
+    options?: Partial<NotificationItem>,
+  ) => {
+    const progressPercent = Math.round((current / total) * 100)
+    return addNotification({
+      title,
+      message: `${current}/${total}`,
+      type: 'info',
+      showProgress: true,
+      progress: progressPercent,
+      duration: 0, // 不自动关闭
+      ...options,
+    })
+  }
+
+  const updateProgress = (id: string, current: number, total: number) => {
+    const notification = notifications.value.find((n) => n.id === id)
+    if (notification) {
+      notification.progress = Math.round((current / total) * 100)
+      notification.message = `${current}/${total}`
+    }
+  }
+
+  const updateProgressDetail = (
+    id: string,
+    detailTitle: string,
+    detailMessage: string,
+    current?: number,
+    total?: number,
+  ) => {
+    const notification = notifications.value.find((n) => n.id === id)
+    if (notification) {
+      notification.detailTitle = detailTitle
+      notification.detailMessage = detailMessage
+      if (current !== undefined && total !== undefined) {
+        notification.progress = Math.round((current / total) * 100)
+        notification.message = `${current}/${total}`
+      }
+    }
+  }
+
+  // 更新通知的主标题
+  const updateNotificationTitle = (id: string, title: string) => {
+    const notification = notifications.value.find((n) => n.id === id)
+    if (notification) {
+      notification.title = title
+    }
+  }
+
+  const clear = () => {
+    notifications.value.forEach((n) => n.onClose?.())
+    notifications.value = []
+  }
+
+  return {
+    notifications,
+    addNotification,
+    removeNotification,
+    success,
+    warning,
+    info,
+    error,
+    progress,
+    updateProgress,
+    updateProgressDetail,
+    updateNotificationTitle,
+    clear,
+  }
+}

frontend/src/layout/components/Main/Task.vue

@@ -1,11 +1,13 @@
 <script setup lang="ts">
-import { ref, onMounted, computed, reactive, nextTick } from 'vue'
+import { ref, onMounted, onUnmounted, computed, reactive, nextTick } from 'vue'
 import SvgIcon from '@/components/SvgIcon/index.vue'
 import { useAgentsStore, useConfigStore } from '@/stores'
 import api from '@/api'
+import websocket from '@/utils/websocket'
 import { changeBriefs } from '@/utils/collaboration_Brief_FrontEnd.ts'
-import { ElMessage } from 'element-plus'
+import { useNotification } from '@/composables/useNotification'
 import AssignmentButton from './TaskTemplate/TaskSyllabus/components/AssignmentButton.vue'
+import { withRetry } from '@/utils/retry'
 const emit = defineEmits<{
   (e: 'search-start'): void
   (e: 'search', value: string): void
@@ -13,11 +15,17 @@ const emit = defineEmits<{
 
 const agentsStore = useAgentsStore()
 const configStore = useConfigStore()
+const { success, warning, error: notifyError } = useNotification()
 const searchValue = ref('')
 const triggerOnFocus = ref(true)
 const isFocus = ref(false)
-const hasAutoSearched = ref(false) // 防止重复自动搜索
+const hasAutoSearched = ref(false)
-const isExpanded = ref(false) // 控制搜索框是否展开
+const isExpanded = ref(false)
+const isFillingSteps = ref(false)
+const isStopping = ref(false)
+const isStopPending = ref(false)
+const currentStepAbortController = ref<{ cancel: () => void } | null>(null)
+const currentGenerationId = ref('')
 
 // 解析URL参数
 function getUrlParam(param: string): string | null {
@@ -29,7 +37,6 @@ const planReady = computed(() => {
   return agentsStore.agentRawPlan.data !== undefined
 })
 const openAgentAllocationDialog = () => {
-  console.log('打开智能体分配弹窗')
   agentsStore.openAgentAllocationDialog()
 }
 // 自动搜索函数
@@ -59,23 +66,66 @@ const taskContainerRef = ref<HTMLDivElement | null>(null)
 // 处理失去焦点事件
 function handleBlur() {
   isFocus.value = false
-  // 延迟收起搜索框，以便点击按钮等操作
   setTimeout(() => {
     isExpanded.value = false
-    // 强制重置文本区域高度到最小行数
     resetTextareaHeight()
   }, 200)
 }
 
+// 预加载所有任务的智能体评分数据
+async function preloadAllTaskAgentScores(outlineData: any, goal: string) {
+  const tasks = outlineData['Collaboration Process'] || []
+
+  if (tasks.length === 0) {
+    return
+  }
+
+  for (const task of tasks) {
+    // 检查是否已停止
+    if (agentsStore.isStopping || agentsStore.hasStoppedFilling) {
+      return
+    }
+
+    // 确保任务有唯一ID
+    if (!task.Id) {
+      task.Id = `task-${Date.now()}-${Math.random().toString(36).substr(2, 9)}`
+    }
+
+    // 调用初始化接口获取评分数据
+    try {
+      const agentScores = await api.agentSelectModifyInit({
+        goal: goal,
+        stepTask: {
+          StepName: task.StepName,
+          TaskContent: task.TaskContent,
+          InputObject_List: task.InputObject_List,
+          OutputObject: task.OutputObject
+        }
+      })
+
+      // 提取维度列表并存储
+      const firstAgent = Object.keys(agentScores)[0]
+      const aspectList = firstAgent ? Object.keys(agentScores[firstAgent] || {}) : []
+
+      agentsStore.setTaskScoreData(task.Id, {
+        aspectList,
+        agentScores
+      })
+    } catch (error) {
+      console.error(`❌ 任务 "${task.StepName}" 的评分数据预加载失败:`, error)
+    }
+  }
+}
+
 // 重置文本区域高度到最小行数
 function resetTextareaHeight() {
   nextTick(() => {
-    // 修复：使用更可靠的方式获取textarea元素
+    // 获取textarea元素
     const textarea =
       document.querySelector('#task-container .el-textarea__inner') ||
       document.querySelector('#task-container textarea')
 
-    if (textarea) {
+    if (textarea instanceof HTMLElement) {
       // 强制设置最小高度
       textarea.style.height = 'auto'
       textarea.style.minHeight = '56px'
@@ -84,27 +134,171 @@ function resetTextareaHeight() {
   })
 }
 
+// 停止填充数据的处理函数
+async function handleStop() {
+  // 检查是否有正在进行的生成任务
+  if (!isFillingSteps.value) {
+    warning('提示', '没有正在进行的生成任务')
+    return
+  }
+
+  // 先设置停止状态（立即显示"停止中..."）
+  agentsStore.setIsStopping(true)
+  isStopping.value = true
+  isStopPending.value = true
+  success('提示', '正在停止，请稍候...')
+
+  // 发送停止请求（不等待响应，后端设置 should_stop = True）
+  if (websocket.connected && currentGenerationId.value) {
+    websocket.send('stop_generation', {
+      generation_id: currentGenerationId.value
+    }).then((result: any) => {
+      console.log('停止生成响应:', result)
+    }).catch((error: any) => {
+      console.log('停止生成请求失败（可能已经停止）:', error?.message)
+    })
+  }
+  // 不清空 currentGenerationId，让 fillStepTask 循环检查 isStopping 来停止
+}
+
+// 监听后端发送的停止完成事件（备用，如果后端有发送）
+function onGenerationStopped() {
+  isStopping.value = false
+  isStopPending.value = false
+  currentGenerationId.value = ''
+  success('成功', '已停止生成')
+}
+
+// 处理按钮点击事件
+function handleButtonClick() {
+  if (isFillingSteps.value) {
+    handleStop()
+  } else {
+    handleSearch()
+  }
+}
+
+// 处理搜索函数
 async function handleSearch() {
-  try {
   triggerOnFocus.value = false
   if (!searchValue.value) {
-      ElMessage.warning('请输入搜索内容')
+    warning('提示', '请输入搜索内容')
     return
   }
   emit('search-start')
+
+  // 重置所有状态（处理可能的上一次未完成的停止操作）
+  isStopping.value = false
+  isStopPending.value = false
+  agentsStore.setIsStopping(false)
+  agentsStore.setHasStoppedFilling(false)
+
   agentsStore.resetAgent()
   agentsStore.setAgentRawPlan({ loading: true })
-    const data = await api.generateBasePlan({
+
+  // 重置 generation_id
+  currentGenerationId.value = ''
+
+  // 获取大纲
+  const response = await api.generateBasePlan({
     goal: searchValue.value,
     inputs: []
   })
-    data['Collaboration Process'] = changeBriefs(data['Collaboration Process'])
+
-    agentsStore.setAgentRawPlan({ data })
+  // WebSocket 返回格式: { data: {...}, generation_id, execution_id }
-    console.log('agentsStore.agentRawPlan', agentsStore.agentRawPlan)
+  // REST API 返回格式: {...}
+  const outlineData = response.data || response
+
+  // 保存 generation_id
+  if (response && response.generation_id) {
+    currentGenerationId.value = response.generation_id
+    console.log('📋 保存 generation_id:', currentGenerationId.value)
+  }
+
+  // 处理简报数据格式
+  outlineData['Collaboration Process'] = changeBriefs(outlineData['Collaboration Process'])
+
+  // 立即显示大纲
+  agentsStore.setAgentRawPlan({ data: outlineData, loading: false })
   emit('search', searchValue.value)
+
+  // 预加载所有任务的智能体评分数据
+  preloadAllTaskAgentScores(outlineData, searchValue.value)
+
+  // 填充步骤详情
+  isFillingSteps.value = true
+  const steps = outlineData['Collaboration Process'] || []
+
+  // 保存 generation_id 到本地变量，用于 fillStepTask 调用
+  // 这样即使前端停止时清空了 currentGenerationId，当前的 fillStepTask 仍能正确停止
+  const fillTaskGenerationId = currentGenerationId.value
+
+  // 串行填充所有步骤的详情
+  try {
+    for (const step of steps) {
+      // 检查是否已停止
+      if (!isFillingSteps.value || agentsStore.isStopping) {
+        break
+      }
+
+      await withRetry(
+        async () => {
+          const detailedStep = await api.fillStepTask({
+            goal: searchValue.value,
+            stepTask: {
+              StepName: step.StepName,
+              TaskContent: step.TaskContent,
+              InputObject_List: step.InputObject_List,
+              OutputObject: step.OutputObject,
+            },
+            generation_id: fillTaskGenerationId,
+          })
+          updateStepDetail(step.StepName, detailedStep)
+        },
+        {
+          maxRetries: 2, // 减少重试次数，因为是串行填充
+          initialDelayMs: 1000, // 使用较小的延迟
+          shouldRetry: () => isFillingSteps.value && !agentsStore.isStopping, // 可取消的重试
+        },
+      )
+    }
   } finally {
+    // 重置状态（确保即使出错也会执行）
     triggerOnFocus.value = true
-    agentsStore.setAgentRawPlan({ loading: false })
+    if (isStopPending.value) {
+      isStopping.value = false
+      isStopPending.value = false
+      agentsStore.setIsStopping(false)
+      agentsStore.setHasStoppedFilling(true)
+    }
+    isFillingSteps.value = false
+    currentStepAbortController.value = null
+    // 只有在没有停止请求时才清空 generation_id
+    if (!isStopPending.value) {
+      currentGenerationId.value = ''
+    }
+  }
+}
+
+//更新单个步骤的详情
+function updateStepDetail(stepId: string, detailedStep: any) {
+  const planData = agentsStore.agentRawPlan.data
+  if (!planData) return
+
+  const collaborationProcess = planData['Collaboration Process']
+  if (!collaborationProcess) return
+
+  const index = collaborationProcess.findIndex((s: any) => s.StepName === stepId)
+  if (index !== -1 && collaborationProcess[index]) {
+    // 保持响应式更新
+    Object.assign(collaborationProcess[index], {
+      AgentSelection: detailedStep.AgentSelection || [],
+      TaskProcess: detailedStep.TaskProcess || [],
+      Collaboration_Brief_frontEnd: detailedStep.Collaboration_Brief_frontEnd || {
+        template: '',
+        data: {}
+      }
+    })
   }
 }
 
@@ -112,7 +306,7 @@ const querySearch = (queryString: string, cb: (v: { value: string }[]) => void)
   const results = queryString
     ? configStore.config.taskPromptWords.filter(createFilter(queryString))
     : configStore.config.taskPromptWords
-  // call callback function to return suggestions
+  // 调用回调函数返回建议列表
   cb(results.map(item => ({ value: item })))
 }
 
@@ -125,6 +319,12 @@ const createFilter = (queryString: string) => {
 // 组件挂载时检查URL参数
 onMounted(() => {
   autoSearchFromUrl()
+  websocket.on('generation_stopped', onGenerationStopped)
+})
+
+// 组件卸载时移除事件监听
+onUnmounted(() => {
+  websocket.off('generation_stopped', onGenerationStopped)
 })
 </script>
 
@@ -168,21 +368,27 @@ onMounted(() => {
           class="task-button"
           color="linear-gradient(to right, #00C7D2, #315AB4)"
           size="large"
-          title="点击搜索任务"
+          :title="isFillingSteps && !isStopping ? '点击停止生成' : '点击搜索任务'"
           circle
-          :loading="agentsStore.agentRawPlan.loading"
+          :loading="agentsStore.agentRawPlan.loading || isStopping"
-          :disabled="!searchValue"
+          :disabled="!searchValue || isStopping"
-          @click.stop="handleSearch"
+          @click.stop="handleButtonClick"
         >
           <SvgIcon
-            v-if="!agentsStore.agentRawPlan.loading"
+            v-if="!agentsStore.agentRawPlan.loading && !isFillingSteps && !isStopping"
             icon-class="paper-plane"
             size="18px"
             color="#ffffff"
           />
+          <SvgIcon
+            v-if="!agentsStore.agentRawPlan.loading && isFillingSteps && !isStopping"
+            icon-class="stoprunning"
+            size="30px"
+            color="#ffffff"
+          />
         </el-button>
       </div>
-      <AssignmentButton v-dev-only v-if="planReady" @click="openAgentAllocationDialog" />
+      <AssignmentButton v-if="planReady" @click="openAgentAllocationDialog" />
     </div>
   </el-tooltip>
 </template>
@@ -197,7 +403,6 @@ onMounted(() => {
   width: 40%;
   margin: 0 auto;
   border: 2px solid transparent;
-  $bg: var(--el-input-bg-color, var(--el-fill-color-blank));
   background: linear-gradient(var(--color-bg-taskbar), var(--color-bg-taskbar)) padding-box,
     linear-gradient(to right, #00c8d2, #315ab4) border-box;
   border-radius: 30px;
@@ -216,8 +421,6 @@ onMounted(() => {
     :deep(.el-autocomplete .el-textarea .el-textarea__inner) {
       overflow-y: auto !important;
       min-height: 56px !important;
-      // overflow-y: hidden;
-      // background-color: black;
     }
   }
 
@@ -326,58 +529,4 @@ onMounted(() => {
     }
   }
 }
-
-.drawer-header {
-  display: flex;
-  align-items: center;
-  justify-content: space-between;
-  width: 100%;
-  .title {
-    font-size: 18px;
-    font-weight: bold;
-  }
-}
-
-.process-list {
-  padding: 0 8px;
-}
-.process-item {
-  margin-bottom: 16px;
-  padding: 12px;
-  border-radius: 8px;
-  background: var(--color-bg-list);
-  border: 1px solid var(--color-border-default);
-
-  .process-content {
-    display: flex;
-    align-items: flex-start;
-    gap: 8px;
-
-    .agent-tag {
-      display: inline-block;
-      padding: 4px 8px;
-      border-radius: 4px;
-      font-size: 14px;
-      font-weight: 500;
-      white-space: nowrap;
-      flex-shrink: 0;
-    }
-
-    .process-text {
-      line-height: 1.6;
-      font-size: 14px;
-      color: var(--el-text-color-primary);
-      white-space: pre-wrap;
-    }
-  }
-
-  .edit-container {
-    margin-top: 8px;
-  }
-}
-
-.process-item:hover {
-  border-color: var(--el-border-color);
-  box-shadow: 0 2px 8px rgba(0, 0, 0, 0.1);
-}
 </style>

frontend/src/layout/components/Main/TaskTemplate/AgentRepo/index.vue

@@ -1,16 +1,16 @@
 <script setup lang="ts">
 import { ref, computed, onMounted } from 'vue'
-import { ElNotification } from 'element-plus'
 import { pick } from 'lodash'
-
 import api from '@/api/index.ts'
-
 import SvgIcon from '@/components/SvgIcon/index.vue'
 import { agentMapDuty } from '@/layout/components/config.ts'
 import { type Agent, useAgentsStore } from '@/stores'
 import { readConfig } from '@/utils/readJson.ts'
+import { useNotification } from '@/composables/useNotification.ts'
 import AgentRepoList from './AgentRepoList.vue'
 
+const { error, success } = useNotification()
+
 const agentsStore = useAgentsStore()
 
 // 如果agentsStore.agents不存在就读取默认配置的json文件
@@ -39,82 +39,41 @@ const handleFileSelect = (event: Event) => {
   }
 }
 
-// 在validateApiConfig函数中添加更详细的日志
+// 验证API配置：三个字段必须同时存在或同时不存在
 const validateApiConfig = (agent: any) => {
   const hasApiUrl = 'apiUrl' in agent
   const hasApiKey = 'apiKey' in agent
   const hasApiModel = 'apiModel' in agent
 
-  // 三个字段必须同时存在或同时不存在
+  return hasApiUrl === hasApiKey && hasApiKey === hasApiModel
-  if (hasApiUrl !== hasApiKey || hasApiKey !== hasApiModel) {
-    console.error('❌ API配置不完整:', {
-      agentName: agent.Name,
-      missingFields: {
-        apiUrl: !hasApiUrl,
-        apiKey: !hasApiKey,
-        apiModel: !hasApiModel
-      }
-    })
-    return false
-  }
-
-  if (hasApiUrl && hasApiKey && hasApiModel) {
-    console.log('✅ API配置完整，将使用自定义API配置:', {
-      apiUrl: agent.apiUrl,
-      apiKey: agent.apiKey ? '***' + agent.apiKey.slice(-4) : '未设置',
-      apiModel: agent.apiModel
-    })
-  } else {
-    console.log('ℹ️ 未设置API配置，将使用默认URL配置')
-  }
-
-  return true
 }
 
 const readFileContent = (file: File) => {
-  console.log('📁 开始读取文件:', file.name, '大小:', file.size, '类型:', file.type)
-
   const reader = new FileReader()
   reader.onload = e => {
     try {
-      console.log('📄 文件读取完成，开始解析JSON')
       const content = e.target?.result as string
       const jsonData = JSON.parse(content)
 
-      console.log('🔍 解析的JSON数据:', jsonData)
-      console.log(
-        '📊 数据类型:',
-        Array.isArray(jsonData) ? '数组' : '对象',
-        '长度:',
-        Array.isArray(jsonData) ? jsonData.length : 'N/A'
-      )
-
       if (!Array.isArray(jsonData)) {
-        console.error('❌ JSON格式错误: 必须为数组格式')
         ElMessage.error('JSON格式错误: 必须为数组格式')
         return
       }
 
-      console.log('🔍 开始验证智能体数据...')
+      const validAgents = jsonData.filter(agent => {
-      const validAgents = jsonData.filter((agent, index) => {
-        console.log(`🔍 验证第${index + 1}个智能体:`, agent.Name || '未命名')
         // 验证必需字段
         if (!agent.Name || typeof agent.Name !== 'string') {
-          console.error(`❌ 智能体${index + 1}缺少Name字段或格式错误`)
           return false
         }
         if (!agent.Icon || typeof agent.Icon !== 'string') {
-          console.error(`❌ 智能体${index + 1}缺少Icon字段或格式错误`)
           return false
         }
         if (!agent.Profile || typeof agent.Profile !== 'string') {
-          console.error(`❌ 智能体${index + 1}缺少Profile字段或格式错误`)
           return false
         }
 
         // 验证API配置
         if (!validateApiConfig(agent)) {
-          console.error(`❌ 智能体${index + 1}API配置验证失败`)
           return false
         }
 
@@ -130,31 +89,24 @@ const readFileContent = (file: File) => {
         apiKey: agent.apiKey,
         apiModel: agent.apiModel
       }))
-
       agentsStore.setAgents(processedAgents)
 
       // 调用API
       api
         .setAgents(processedAgents)
         .then(() => {
-          console.log('✅ 后端API调用成功')
+          success('智能体上传成功')
-          ElMessage.success('智能体上传成功')
         })
-        .catch(error => {
+        .catch(() => {
-          console.error('❌ 后端API调用失败:', error)
+          error('智能体上传失败')
-          ElMessage.error('智能体上传失败')
         })
-    } catch (error) {
+    } catch {
-      console.error('❌ JSON解析错误:', error)
+      error('JSON解析错误')
-      ElMessage.error('JSON解析错误')
     }
   }
-
+  reader.onerror = () => {
-  reader.onerror = error => {
+    error('文件读取错误')
-    console.error('❌ 文件读取错误:', error)
-    ElMessage.error('文件读取错误')
   }
-
   reader.readAsText(file)
 }
 

frontend/src/layout/components/Main/TaskTemplate/TaskProcess/ProcessCard.vue

@@ -19,39 +19,34 @@ const props = defineProps<{
 }>()
 
 const emit = defineEmits<{
-  (e: 'open-edit', stepId: string, processId: string): void
   (e: 'save-edit', stepId: string, processId: string, value: string): void
 }>()
 
-// 🔄 从 currentTask 中获取数据（与分支切换联动）
+//从 currentTask 中获取数据
 const currentTaskProcess = computed(() => {
-  // ✅ 优先使用 currentTask（包含分支切换后的数据）
   const currentTask = agentsStore.currentTask
   if (currentTask && currentTask.Id === props.step.Id && currentTask.TaskProcess) {
     return currentTask.TaskProcess
   }
 
-  // ⚠️ 降级：从 agentRawPlan 中获取原始数据（不受分支切换影响）
+  //从 agentRawPlan 中获取原始数据
   const collaborationProcess = agentsStore.agentRawPlan.data?.['Collaboration Process'] || []
   const rawData = collaborationProcess.find((task: any) => task.Id === props.step.Id)
   return rawData?.TaskProcess || []
 })
 
-// 当前正在编辑的process ID
 const editingProcessId = ref<string | null>(null)
-// 编辑框的值
 const editValue = ref('')
-// 鼠标悬停的process ID
 const hoverProcessId = ref<string | null>(null)
 
-// 🆕 处理卡片点击事件（非编辑模式下）
+// 处理卡片点击事件
 function handleCardClick() {
   // 如果正在编辑，不处理点击
   if (editingProcessId.value) return
 
   // 设置当前任务，与任务大纲联动
   if (props.step.Id) {
-    agentsStore.setCurrentTask(props.step)
+    agentsStore.setCurrentTask(props.step as any)
   }
 }
 
@@ -62,40 +57,25 @@ function isDarkMode(): boolean {
 
 // 获取颜色浅两号的函数
 function getLightColor(color: string, level: number = 2): string {
-  if (!color || color.length !== 7 || color[0] !== '#') return color
+  return adjustColor(color, level * 20)
-
-  const r = parseInt(color.substr(1, 2), 16)
-  const g = parseInt(color.substr(3, 2), 16)
-  const b = parseInt(color.substr(5, 2), 16)
-
-  // 增加亮度（浅两号）
-  const lightenAmount = level * 20
-  const newR = Math.min(255, r + lightenAmount)
-  const newG = Math.min(255, g + lightenAmount)
-  const newB = Math.min(255, b + lightenAmount)
-
-  return `#${Math.round(newR).toString(16).padStart(2, '0')}${Math.round(newG)
-    .toString(16)
-    .padStart(2, '0')}${Math.round(newB).toString(16).padStart(2, '0')}`
 }
 
 // 获取颜色深两号的函数
 function getDarkColor(color: string, level: number = 2): string {
+  return adjustColor(color, -level * 20)
+}
+
+// 通用的颜色调整函数（提取重复逻辑）
+function adjustColor(color: string, amount: number): string {
   if (!color || color.length !== 7 || color[0] !== '#') return color
 
-  const r = parseInt(color.substr(1, 2), 16)
+  const r = Math.min(255, Math.max(0, parseInt(color.substr(1, 2), 16) + amount))
-  const g = parseInt(color.substr(3, 2), 16)
+  const g = Math.min(255, Math.max(0, parseInt(color.substr(3, 2), 16) + amount))
-  const b = parseInt(color.substr(5, 2), 16)
+  const b = Math.min(255, Math.max(0, parseInt(color.substr(5, 2), 16) + amount))
 
-  // 降低亮度（深两号）
+  return `#${r.toString(16).padStart(2, '0')}${g.toString(16).padStart(2, '0')}${b
-  const darkenAmount = level * 20
-  const newR = Math.max(0, r - darkenAmount)
-  const newG = Math.max(0, g - darkenAmount)
-  const newB = Math.max(0, b - darkenAmount)
-
-  return `#${Math.round(newR).toString(16).padStart(2, '0')}${Math.round(newG)
     .toString(16)
-    .padStart(2, '0')}${Math.round(newB).toString(16).padStart(2, '0')}`
+    .padStart(2, '0')}`
 }
 
 // 根据主题模式获取调整后的颜色
@@ -117,11 +97,10 @@ function handleMouseLeave() {
   hoverProcessId.value = null
 }
 
-// 处理双击编辑（针对单个process）
+// 处理双击编辑
 function handleDblClick(processId: string, currentDescription: string) {
   editingProcessId.value = processId
   editValue.value = currentDescription
-  emit('open-edit', props.step.Id || '', processId)
 }
 
 // 处理保存编辑
@@ -264,7 +243,6 @@ function handleCancel() {
           margin-bottom: 8px;
 
           .edit-card {
-            //background: #f0f2f5;
             border-radius: 8px;
             padding: 16px;
             box-shadow: 0 2px 8px rgba(0, 0, 0, 0.1);
@@ -284,13 +262,6 @@ function handleCancel() {
                 color: var(--color-text-taskbar);
               }
             }
-
-            .edit-buttons {
-              display: flex;
-              gap: 8px;
-              justify-content: flex-end;
-              margin-top: 8px;
-            }
           }
         }
 
@@ -301,10 +272,6 @@ function handleCancel() {
           padding: 2px 4px;
           border-radius: 3px;
           transition: background-color 0.2s ease;
-
-          &.hovered {
-            transition: background-color 0.2s ease;
-          }
         }
 
         .separator {

frontend/src/layout/components/Main/TaskTemplate/TaskProcess/components/TaskButton.vue

@@ -12,19 +12,18 @@ const props = defineProps<{
   step?: any
 }>()
 
-// 获取分支数量 - 主分支(1) + 额外分支数量
+// 获取分支数量
 const branchCount = computed(() => {
   if (!props.step?.Id) return 1
 
-  // 获取该任务步骤的分支数据
   const taskStepId = props.step.Id
-  const branches = selectionStore.getTaskProcessBranches(taskStepId)
+  // 获取该任务的 agent 组合
-
+  const agents = props.step.AgentSelection || []
-  // 主分支(1) + 额外分支数量
+  const branches = selectionStore.getTaskProcessBranches(taskStepId, agents)
   return branches.length || 1
 })
 
-// 🆕 判断按钮是否可点击（只有当前按钮对应的任务是任务大纲中选中的任务时才可点击）
+// 判断按钮是否可点击
 const isClickable = computed(() => {
   if (!props.step?.Id || !agentsStore.currentTask?.Id) {
     return false
@@ -33,7 +32,6 @@ const isClickable = computed(() => {
 })
 
 const handleClick = (event?: MouseEvent) => {
-  // 🆕 只有可点击时才执行操作
   if (!isClickable.value) {
     return
   }
@@ -48,7 +46,6 @@ const handleClick = (event?: MouseEvent) => {
   if (props.step) {
     agentsStore.setCurrentTask(props.step)
   }
-  // 触发打开任务过程探索窗口
   agentsStore.openPlanTask()
 }
 </script>
@@ -56,7 +53,7 @@ const handleClick = (event?: MouseEvent) => {
 <template>
   <div
     class="task-button"
-    :class="{ 'has-branches': branchCount > 0, 'is-disabled': !isClickable }"
+    :class="{ 'has-branches': branchCount > 1, 'is-disabled': !isClickable }"
     @click="handleClick"
     :title="isClickable ? `${branchCount} 个分支` : '请先在任务大纲中选中此任务'"
   >
@@ -100,7 +97,7 @@ const handleClick = (event?: MouseEvent) => {
     filter: brightness(0.9);
   }
 
-  // 🆕 禁用状态
+  // 禁用状态
   &.is-disabled {
     background-color: #bdc3c7;
     cursor: not-allowed;

frontend/src/layout/components/Main/TaskTemplate/TaskResult/AdditionalOutputCard.vue

@@ -1,168 +0,0 @@
-<!-- AdditionalOutputCard.vue -->
-<script setup lang="ts">
-import { computed, ref, nextTick } from 'vue'
-import { useAgentsStore } from '@/stores'
-import SvgIcon from '@/components/SvgIcon/index.vue'
-
-const agentsStore = useAgentsStore()
-const props = defineProps<{
-  index: number
-}>()
-
-// 获取产物名称
-const currentOutput = computed(() => {
-  return agentsStore.additionalOutputs[props.index] || ''
-})
-
-// 编辑状态
-const isEditing = ref(false)
-const inputValue = ref('')
-const originalValue = ref('')
-const inputRef = ref<HTMLElement>()
-
-// 点击编辑图标
-const handleEditClick = () => {
-  isEditing.value = true
-  originalValue.value = inputValue.value
-
-  // 等待 DOM 更新后聚焦输入框
-  nextTick(() => {
-    if (inputRef.value) {
-      const inputEl = inputRef.value.querySelector('input')
-      if (inputEl) {
-        inputEl.focus()
-      }
-    }
-  })
-}
-
-// 保存编辑
-const handleSave = () => {
-  if (isEditing.value) {
-    isEditing.value = false
-  }
-}
-
-// 取消编辑
-const handleCancel = () => {
-  if (isEditing.value) {
-    inputValue.value = originalValue.value // 恢复原始值
-    isEditing.value = false
-  }
-}
-
-// 处理键盘事件
-const handleKeydown = (event: KeyboardEvent) => {
-  if (event.key === 'Enter') {
-    event.preventDefault()
-    handleSave()
-  } else if (event.key === 'Escape') {
-    event.preventDefault()
-    handleCancel()
-  }
-}
-
-// 输入框失去焦点处理
-const handleBlur = () => {
-  // 延迟处理，避免点击按钮时立即触发
-  setTimeout(() => {
-    if (isEditing.value) {
-      handleSave()
-    }
-  }, 200)
-}
-</script>
-
-<template>
-  <!-- 当产物存在时才显示 -->
-  <div v-if="currentOutput" class="card-item">
-    <el-card
-      class="card-item w-full relative output-object-card"
-      :shadow="true"
-      :id="`additional-output-${index}`"
-    >
-      <!-- 显示产物名称 -->
-      <div class="text-start w-[100%]">
-        <div class="text-[18px] font-bold text-[var(--color-text)] mb-2">
-          {{ currentOutput }}
-        </div>
-        <div ref="inputRef">
-          <el-input
-            v-model="inputValue"
-            :readonly="!isEditing"
-            :placeholder="isEditing ? '请输入内容...' : '点击编辑图标开始编辑...'"
-            @keydown="handleKeydown"
-            @blur="handleBlur"
-            :class="{ editing: isEditing }"
-          >
-            <template #suffix>
-              <!-- 只读状态：显示编辑图标 -->
-              <div v-if="!isEditing" class="flex items-center">
-                <svg-icon
-                  icon-class="Edit"
-                  size="20px"
-                  class="cursor-pointer hover:text-[#409eff] transition-colors"
-                  @click="handleEditClick"
-                  title="点击编辑"
-                />
-              </div>
-            </template>
-          </el-input>
-        </div>
-
-        <!-- 编辑状态下的提示 -->
-        <div v-if="isEditing" class="mt-2 text-end text-xs text-gray-500">
-          <svg-icon
-            icon-class="Check"
-            size="20px"
-            color="#328621"
-            class="cursor-pointer mr-4"
-            @click="handleSave"
-            title="保存"
-          />
-          <svg-icon
-            icon-class="Cancel"
-            size="20px"
-            color="#8e0707"
-            class="cursor-pointer mr-4"
-            @click="handleCancel"
-            title="取消"
-          />
-        </div>
-      </div>
-    </el-card>
-  </div>
-</template>
-
-<style scoped lang="scss">
-.card-item {
-  transition: all 0.3s ease;
-}
-
-.output-object-card {
-  :deep(.el-card__body) {
-    min-height: 80px;
-    display: flex;
-    align-items: start;
-    justify-content: start;
-  }
-}
-
-/* 输入框样式 */
-:deep(.el-input .el-input__wrapper) {
-  box-shadow: none;
-  background-color: var(--color-bg-three);
-  transition: all 0.2s ease;
-}
-
-/* 编辑状态下的输入框样式 */
-:deep(.el-input.editing .el-input__wrapper) {
-  border: 1px solid #dcdfe6;
-  border-radius: 4px;
-}
-
-:deep(.el-input.editing .el-input__wrapper.is-focus) {
-  border-color: #c0c4cc;
-  box-shadow: 0 0 0 1px #c0c4cc;
-}
-</style>

frontend/src/layout/components/Main/TaskTemplate/TaskResult/ExecutePlan.vue

@@ -38,11 +38,12 @@ const data = computed<Data | null>(() => {
     if (result.NodeId === props.nodeId) {
       //  LogNodeType 为 object直接渲染Content
       if (result.LogNodeType === 'object') {
-        return {
+        const data = {
           Description: props.nodeId,
           Content: sanitize(result.content),
           LogNodeType: result.LogNodeType
         }
+        return data
       }
 
       if (!result.ActionHistory) {

frontend/src/layout/components/Main/TaskTemplate/TaskResult/Iod.vue

@@ -59,7 +59,7 @@ function handlePrev() {
             <!-- <div>{{ `${displayIndex + 1}/${data.length}` }}</div>
             <el-button type="primary" size="small" @click="handleNext">下一个</el-button> -->
             <!-- 关闭 -->
-            <SvgIcon icon-class="close" size="15px" />
+            <!-- <SvgIcon icon-class="close" size="15px" /> -->
           </div>
         </div>
         <!-- 分割线 -->

frontend/src/layout/components/Main/TaskTemplate/TaskSyllabus/Bg.vue

@@ -1,12 +1,4 @@
-<script setup lang="ts">
+<script setup lang="ts"></script>
-defineProps<{
-  isAdding?: boolean
-}>()
-
-defineEmits<{
-  (e: 'start-add-output'): void
-}>()
-</script>
 <template>
   <div class="absolute inset-0 flex items-start gap-[14%]">
     <!-- 左侧元素 -->
@@ -24,21 +16,6 @@ defineEmits<{
     <div class="flex-1 relative h-full flex justify-center">
       <!-- 背景那一根线 -->
       <div class="h-full bg-[var(--color-bg-flow)] w-[5px]">
-        <!-- 顶部加号区域 -->
-        <div
-          v-if="!isAdding"
-          v-dev-only
-          class="plus-area mt-[35px] ml-[-15px] w-[34px] h-[34px] flex items-center justify-center cursor-pointer rounded-full"
-          @click="$emit('start-add-output')"
-        >
-          <!-- 加号图标 -->
-          <svg-icon
-            icon-class="plus"
-            color="var(--color-text)"
-            size="20px"
-            class="plus-icon opacity-0 transition-opacity duration-200"
-          />
-        </div>
         <!-- 线底部的小圆球 -->
         <div
           class="absolute bottom-0 left-1/2 transform -translate-x-1/2 bg-[var(--color-bg-flow)] w-[15px] h-[15px] rounded-full"

frontend/src/layout/components/Main/TaskTemplate/TaskSyllabus/Branch/components/RootNode.vue

@@ -335,7 +335,7 @@ const handleBranchKeydown = (event: KeyboardEvent) => {
 
     .el-input__inner {
       font-size: 13px;
-      color: var(--color-text-primary);
+      color: #000;
       padding-right: 40px;
     }
   }

frontend/src/layout/components/Main/TaskTemplate/TaskSyllabus/Branch/components/TaskNode.vue

@@ -476,7 +476,7 @@ const handleBranchKeydown = (event: KeyboardEvent) => {
 
     .el-input__inner {
       font-size: 13px;
-      color: var(--color-text-primary);
+      color: #000;
       padding-right: 40px;
     }
   }

frontend/src/layout/components/Main/TaskTemplate/TaskSyllabus/Branch/mock/README.md

@@ -1,127 +0,0 @@
-# Mock 数据说明
-
-本目录包含用于分支功能的 mock 数据，支持在开发环境中测试分支逻辑，无需调用真实后端 API。
-
-## 文件说明
-
-### 1. `branchPlanOutlineMock.ts`
-**用途**: 根节点级别的分支（任务大纲分支）
-
-**类型**: `IApiStepTask[][]`
-
-**说明**: 返回多个分支方案，每个方案是一个完整的任务流程（IApiStepTask[]）
-
-**示例结构**:
-```typescript
-[
-  // 第一个分支方案（瀑布流开发）
-  [task1, task2, ...],
-  // 第二个分支方案（敏捷开发）
-  [task1, task2, ...],
-  // 第三个分支方案（快速原型）
-  [task1, task2, ...]
-]
-```
-
-### 2. `branchTaskProcessMock.ts`
-**用途**: 任务节点级别的分支（任务流程分支）
-
-**类型**: `IApiAgentAction[][]`
-
-**说明**: 返回多个分支方案，每个方案是一系列动作（IApiAgentAction[]），这些动作会追加到现有任务的 TaskProcess 中
-
-**示例结构**:
-```typescript
-[
-  // 第一个分支方案（标准开发流程）
-  [action1, action2, action3, ...],
-  // 第二个分支方案（快速原型流程）
-  [action1, action2, ...],
-  // ... 更多方案
-]
-```
-
-## 如何使用
-
-### 切换 Mock 数据开关
-
-在 `PlanModification.vue` 文件中，找到以下配置：
-
-```typescript
-// 开关：控制是否使用 mock 数据（开发时设置为 true，生产时设置为 false）
-const USE_MOCK_DATA = true
-```
-
-- **开发阶段**: 设置为 `true`，使用 mock 数据
-- **生产环境**: 设置为 `false`，调用真实 API
-
-### 数据转换流程
-
-```
-IApiStepTask (API 格式)
-    ↓ convertToIRawStepTask()
-IRawStepTask (内部格式)
-    ↓
-更新到 agentsStore.agentRawPlan.data
-    ↓
-Vue Flow 流程图自动更新
-```
-
-### Mock 数据特点
-
-1. **模拟网络延迟**: Mock 数据调用会模拟 800ms 的网络延迟
-2. **多方案选择**: 每个分支接口提供多个备选方案（目前默认使用第一个）
-3. **完整数据结构**: Mock 数据包含完整的字段，与真实 API 返回格式一致
-4. **类型安全**: 使用 TypeScript 类型定义，确保类型正确
-
-## 扩展 Mock 数据
-
-### 添加新的分支方案
-
-在对应的 mock 文件中添加新的数组元素：
-
-```typescript
-// branchPlanOutlineMock.ts
-const mockPlanBranchData: IApiStepTask[][] = [
-  // 现有方案...
-  [
-    // 新增方案
-    {
-      name: '新方案步骤1',
-      content: '...',
-      // ... 其他字段
-    },
-    {
-      name: '新方案步骤2',
-      content: '...',
-      // ... 其他字段
-    }
-  ]
-]
-```
-
-### 随机选择方案
-
-修改 `PlanModification.vue` 中的方案选择逻辑：
-
-```typescript
-// 当前：固定选择第一个
-const mockBranchTasks = branchPlanOutlineMock[0]
-
-// 改为：随机选择一个方案
-const randomIndex = Math.floor(Math.random() * branchPlanOutlineMock.length)
-const mockBranchTasks = branchPlanOutlineMock[randomIndex]
-```
-
-## 注意事项
-
-1. ⚠️ **生产环境**: 发布前务必将 `USE_MOCK_DATA` 设置为 `false`
-2. 🔍 **调试**: 查看控制台日志，以 `[Mock]` 开头的是 mock 数据相关日志
-3. 📝 **数据一致性**: 确保 mock 数据结构与真实 API 返回格式一致
-4. 🔄 **数据持久化**: Mock 数据仅存在于前端，刷新页面后会丢失
-
-## 相关文件
-
-- `PlanModification.vue`: 主要逻辑文件，包含分支添加和 mock 数据集成
-- `api/index.ts`: 真实 API 接口定义
-- `stores/modules/agents.ts`: 类型定义和数据存储

frontend/src/layout/components/Main/TaskTemplate/TaskSyllabus/Branch/mock/branchPlanOutlineMock.ts

@@ -1,246 +0,0 @@
-// branch_PlanOutline 接口的 Mock 数据和 Mock API
-// 模拟后端返回的原始数据格式（IRawPlanResponse）
-
-import type { IRawPlanResponse, IRawStepTask } from '@/stores'
-
-// 后端返回的数据格式
-export type BranchPlanOutlineResponse = IRawPlanResponse
-
-// Mock 数据：模拟后端返回的原始分支数据（不含 Collaboration_Brief_FrontEnd）
-// 注意：这里模拟的是 branch_PlanOutline 函数返回的数据，不是前端转换后的数据
-const mockBranchDataRaw: IRawStepTask[][] = [
-  // 第一个分支方案
-  [
-    {
-      StepName: '分析用户需求',
-      TaskContent: '分析用户需求，制定项目开发计划',
-      InputObject_List: ['腐蚀类型及成因列表'],
-      OutputObject: '项目开发计划书',
-      AgentSelection: ['腐蚀机理研究员', '实验材料学家'],
-      Collaboration_Brief_frontEnd: {
-        template: '',
-        data: {},
-      },
-      TaskProcess: [],
-    },
-    {
-      StepName: '系统设计与架构',
-      TaskContent: '设计系统架构和数据库结构',
-      InputObject_List: ['项目开发计划书'],
-      OutputObject: '系统设计文档',
-      AgentSelection: ['腐蚀机理研究员', '防护工程专家'],
-      Collaboration_Brief_frontEnd: {
-        template: '',
-        data: {},
-      },
-      TaskProcess: [],
-    },
-  ],
-  // 第二个分支方案（快速原型方案）
-  [
-    {
-      StepName: '需求快速原型',
-      TaskContent: '构建快速原型验证核心功能',
-      InputObject_List: ['腐蚀类型及成因列表'],
-      OutputObject: '原型系统',
-      AgentSelection: ['实验材料学家', '防护工程专家'],
-      Collaboration_Brief_frontEnd: {
-        template: '',
-        data: {},
-      },
-      TaskProcess: [],
-    },
-    {
-      StepName: '原型测试与优化',
-      TaskContent: '测试原型并根据反馈快速迭代',
-      InputObject_List: ['原型系统'],
-      OutputObject: '优化后的原型',
-      AgentSelection: ['腐蚀机理研究员', '实验材料学家'],
-      Collaboration_Brief_frontEnd: {
-        template: '',
-        data: {},
-      },
-      TaskProcess: [],
-    },
-  ],
-  // 第三个分支方案（质量优先方案）
-  [
-    {
-      StepName: '需求深度分析',
-      TaskContent: '深入分析用户需求和技术可行性',
-      InputObject_List: ['腐蚀类型及成因列表'],
-      OutputObject: '详细需求分析报告',
-      AgentSelection: ['腐蚀机理研究员', '防护工程专家'],
-      Collaboration_Brief_frontEnd: {
-        template: '',
-        data: {},
-      },
-      TaskProcess: [],
-    },
-    {
-      StepName: '质量保障设计',
-      TaskContent: '设计质量保障体系和测试方案',
-      InputObject_List: ['详细需求分析报告'],
-      OutputObject: '质量保障方案',
-      AgentSelection: ['实验材料学家', '防护工程专家'],
-      Collaboration_Brief_frontEnd: {
-        template: '',
-        data: {},
-      },
-      TaskProcess: [],
-    },
-    {
-      StepName: '系统开发与测试',
-      TaskContent: '按质量标准进行系统开发和测试',
-      InputObject_List: ['质量保障方案'],
-      OutputObject: '经过完整测试的系统',
-      AgentSelection: ['腐蚀机理研究员', '实验材料学家'],
-      Collaboration_Brief_frontEnd: {
-        template: '',
-        data: {},
-      },
-      TaskProcess: [],
-    },
-  ],
-  // 第四个分支方案（敏捷开发方案）
-  [
-    {
-      StepName: '迭代规划',
-      TaskContent: '制定敏捷开发迭代计划',
-      InputObject_List: ['腐蚀类型及成因列表'],
-      OutputObject: '迭代计划',
-      AgentSelection: ['防护工程专家', '腐蚀机理研究员'],
-      Collaboration_Brief_frontEnd: {
-        template: '',
-        data: {},
-      },
-      TaskProcess: [],
-    },
-    {
-      StepName: '快速开发与验证',
-      TaskContent: '快速开发并持续验证功能',
-      InputObject_List: ['迭代计划'],
-      OutputObject: '可运行的功能模块',
-      AgentSelection: ['实验材料学家', '防护工程专家'],
-      Collaboration_Brief_frontEnd: {
-        template: '',
-        data: {},
-      },
-      TaskProcess: [],
-    },
-  ],
-]
-
-/**
- * 模拟后端的 Add_Collaboration_Brief_FrontEnd 函数
- * 为每个任务添加前端协作简报（Collaboration_Brief_frontEnd）
- */
-function Add_Collaboration_Brief_FrontEnd(branchList: IRawStepTask[][]): IRawStepTask[][] {
-  return branchList.map((tasks) =>
-    tasks.map((task) => ({
-      ...task,
-      Collaboration_Brief_frontEnd: generateCollaborationBrief(task),
-    })),
-  )
-}
-
-/**
- * 生成协作简报（Collaboration_Brief_frontEnd）
- * 根据 StepName、TaskContent、AgentSelection 生成模板和数据
- */
-function generateCollaborationBrief(task: IRawStepTask): {
-  template: string
-  data: Record<string, any>
-} {
-  const agents = task.AgentSelection || []
-  const stepName = task.StepName || ''
-
-  // 为每个 agent 生成颜色
-  const colors = [
-    'hsl(210, 70%, 50%)',
-    'hsl(30, 70%, 50%)',
-    'hsl(120, 70%, 50%)',
-    'hsl(270, 70%, 50%)',
-  ]
-
-  // 生成 data 对象
-  const data: Record<string, any> = {}
-  agents.forEach((agent, index) => {
-    data[agent] = {
-      text: agent,
-      style: { background: colors[index % colors.length] },
-    }
-  })
-
-  // 生成 template（简化版本，实际应根据任务内容生成）
-  const template =
-    agents.length > 0
-      ? agents.map((agent, i) => `!<${agent}>!负责!<${stepName}-${i}>!`).join('，')
-      : ''
-
-  return {
-    template,
-    data,
-  }
-}
-
-/**
- * Mock API：模拟后端 branch_PlanOutline 接口调用
- *
- * @param branch_Number - 分支数量
- * @param Modification_Requirement - 修改需求
- * @param Existing_Steps - 现有步骤名称列表
- * @param Baseline_Completion - 基线完成度
- * @param InitialObject_List - 初始对象列表
- * @param General_Goal - 总体目标
- * @returns Promise<IRawPlanResponse> - 返回包含 'Collaboration Process' 的响应
- */
-export const mockBranchPlanOutlineAPI = async (params: {
-  branch_Number: number
-  Modification_Requirement: string
-  Existing_Steps: string[]
-  Baseline_Completion: number
-  InitialObject_List: string[]
-  General_Goal: string
-}): Promise<IRawPlanResponse> => {
-  // 模拟网络延迟 800ms
-  await new Promise((resolve) => setTimeout(resolve, 800))
-
-  console.log('[Mock API] branch_PlanOutline 调用参数:', params)
-
-  // 🆕 使用轮询方式选择分支方案（依次循环使用所有分支方案）
-  const totalBranches = mockBranchDataRaw.length
-  const sessionKey = `branch-plan-outline-index-${params.General_Goal || 'default'}`
-
-  // 获取上一次的选择索引
-  let lastIndex = parseInt(sessionStorage.getItem(sessionKey) || '0')
-
-  // 计算本次的选择索引（轮询到下一个分支）
-  const selectedBranchIndex = (lastIndex + 1) % totalBranches
-
-  // 保存本次的选择索引
-  sessionStorage.setItem(sessionKey, selectedBranchIndex.toString())
-
-  const rawBranchData = mockBranchDataRaw[selectedBranchIndex]
-
-  console.log(
-    '[Mock API] branch_PlanOutline 选择分支方案:',
-    selectedBranchIndex + 1,
-    '/',
-    totalBranches,
-  )
-
-  // 模拟后端处理：添加 Collaboration_Brief_FrontEnd
-  const processedBranches = Add_Collaboration_Brief_FrontEnd([rawBranchData])
-
-  // 构造响应数据（符合后端返回格式）
-  const response: IRawPlanResponse = {
-    'Collaboration Process': processedBranches[0] || [],
-  }
-
-  console.log('[Mock API] branch_PlanOutline 返回数据:', response)
-
-  return response
-}
-
-export default mockBranchPlanOutlineAPI

frontend/src/layout/components/Main/TaskTemplate/TaskSyllabus/Branch/mock/branchTaskProcessMock.ts

@@ -106,7 +106,7 @@ const mockBranchTaskProcessDataRaw: BranchAction[][] = [
  *
  * @param branch_Number - 分支数量
  * @param Modification_Requirement - 修改需求
- * @param Existing_Steps - 现有步骤名称列表
+ * @param Existing_Steps - 现有步骤列表（包含完整信息的对象数组）
  * @param Baseline_Completion - 基线完成度
  * @param stepTaskExisting - 现有任务
  * @param General_Goal - 总体目标
@@ -115,7 +115,7 @@ const mockBranchTaskProcessDataRaw: BranchAction[][] = [
 export const mockBranchTaskProcessAPI = async (params: {
   branch_Number: number
   Modification_Requirement: string
-  Existing_Steps: string[]
+  Existing_Steps: BranchAction[]
   Baseline_Completion: number
   stepTaskExisting: any
   General_Goal: string

frontend/src/layout/components/Main/TaskTemplate/TaskSyllabus/Branch/mock/fill-step-task-mock.ts

@@ -1,617 +0,0 @@
-// fill_stepTask 接口的 Mock API
-// 简化版本：硬编码多个流程卡片的填充内容（支持多个分支方案）
-
-import type { IRawStepTask } from '@/stores'
-
-// 后端返回的数据格式
-export type FillStepTaskResponse = IRawStepTask
-
-/**
- * Mock API：模拟后端 fill_stepTask 接口调用
- *
- * 根据 stepTask 的 StepName 返回对应的填充内容
- * 支持多个分支方案的轮询选择
- *
- * @param General_Goal - 总体目标
- * @param stepTask - 待填充的任务（包含基本字段）
- * @returns Promise<IRawStepTask> - 返回填充了 AgentSelection 和 TaskProcess 的完整任务
- */
-export const mockFillStepTaskAPI = async (params: {
-  General_Goal: string
-  stepTask: any
-}): Promise<IRawStepTask> => {
-  // 模拟网络延迟 300ms
-  await new Promise((resolve) => setTimeout(resolve, 300))
-
-  console.log('[Mock API] fill_stepTask 调用参数:', params)
-
-  const stepName = params.stepTask.StepName
-
-  // 🆕 使用轮询方式选择填充方案（依次循环使用所有分支方案）
-  const branchIndexKey = `fill-step-task-index-${params.stepTask?.Id || stepName || 'default'}`
-  let lastIndex = parseInt(sessionStorage.getItem(branchIndexKey) || '0')
-  const totalBranches = 4 // 总共4个分支方案
-  const selectedBranchIndex = (lastIndex + 1) % totalBranches
-  sessionStorage.setItem(branchIndexKey, selectedBranchIndex.toString())
-
-  console.log('[Mock API] fill_stepTask 选择分支方案:', selectedBranchIndex + 1, '/', totalBranches)
-
-  // ==================== 第一个任务组 ====================
-  if (stepName === '分析用户需求' || stepName === '需求快速原型' || stepName === '需求深度分析' || stepName === '迭代规划') {
-    let filledTask: IRawStepTask
-
-    if (selectedBranchIndex === 0) {
-      // 第一个分支方案：分析用户需求
-      filledTask = {
-        Id: params.stepTask.Id || 'task-1',
-        StepName: '分析用户需求',
-        TaskContent: '分析用户需求，制定项目开发计划',
-        InputObject_List: ['用户需求文档', '技术规范'],
-        OutputObject: '项目开发计划书',
-        AgentSelection: ['腐蚀机理研究员', '实验材料学家'],
-        Collaboration_Brief_frontEnd: {
-          template: '!腐蚀机理研究员!负责!腐蚀类型识别-0!，!实验材料学家!负责!腐蚀类型识别-1!',
-          data: {
-            腐蚀机理研究员: {
-              text: '腐蚀机理研究员',
-              style: { background: 'hsl(210, 70%, 50%)' },
-            },
-            实验材料学家: {
-              text: '实验材料学家',
-              style: { background: 'hsl(120, 70%, 50%)' },
-            },
-          },
-        },
-        TaskProcess: [
-          {
-            ID: 'action-1-1',
-            ActionType: 'Propose',
-            AgentName: '腐蚀机理研究员',
-            Description: '分析腐蚀环境和介质特征，识别潜在腐蚀类型',
-            ImportantInput: ['InputObject:腐蚀类型及成因列表'],
-          },
-          {
-            ID: 'action-1-2',
-            ActionType: 'Propose',
-            AgentName: '实验材料学家',
-            Description: '设计腐蚀实验方案，确定测试参数',
-            ImportantInput: ['ActionResult:action-1-1'],
-          },
-          {
-            ID: 'action-1-3',
-            ActionType: 'Critique',
-            AgentName: '腐蚀机理研究员',
-            Description: '评估实验方案的可行性',
-            ImportantInput: [
-              'ActionResult:action-1-2',
-              'ActionResult:action-1-1',
-              'InputObject:腐蚀类型及成因列表',
-            ],
-          },
-          {
-            ID: 'action-1-4',
-            ActionType: 'Improve',
-            AgentName: '实验材料学家',
-            Description: '优化实验设计，完善测试流程',
-            ImportantInput: ['ActionResult:action-1-3'],
-          },
-          {
-            ID: 'action-1-5',
-            ActionType: 'Finalize',
-            AgentName: '腐蚀机理研究员',
-            Description: '确认最终的腐蚀类型识别方案',
-            ImportantInput: ['ActionResult:action-1-4', 'ActionResult:action-1-3'],
-          },
-        ],
-      }
-    } else if (selectedBranchIndex === 1) {
-      // 第二个分支方案：需求快速原型
-      filledTask = {
-        Id: params.stepTask.Id || 'task-1',
-        StepName: '需求快速原型',
-        TaskContent: '构建快速原型验证核心功能',
-        InputObject_List: ['腐蚀类型及成因列表'],
-        OutputObject: '原型系统',
-        AgentSelection: ['实验材料学家', '防护工程专家'],
-        Collaboration_Brief_frontEnd: {
-          template: '!实验材料学家!负责!需求快速原型-0!，!防护工程专家!负责!需求快速原型-1!',
-          data: {
-            实验材料学家: {
-              text: '实验材料学家',
-              style: { background: 'hsl(120, 70%, 50%)' },
-            },
-            防护工程专家: {
-              text: '防护工程专家',
-              style: { background: 'hsl(30, 70%, 50%)' },
-            },
-          },
-        },
-        TaskProcess: [
-          {
-            ID: 'action-1-1-branch2',
-            ActionType: 'Propose',
-            AgentName: '实验材料学家',
-            Description: '基于腐蚀类型列表，设计快速原型验证方案',
-            ImportantInput: ['InputObject:腐蚀类型及成因列表'],
-          },
-          {
-            ID: 'action-1-2-branch2',
-            ActionType: 'Propose',
-            AgentName: '防护工程专家',
-            Description: '实现原型核心功能和用户界面',
-            ImportantInput: ['ActionResult:action-1-1-branch2'],
-          },
-          {
-            ID: 'action-1-3-branch2',
-            ActionType: 'Critique',
-            AgentName: '实验材料学家',
-            Description: '评估原型功能完整性和用户体验',
-            ImportantInput: ['ActionResult:action-1-2-branch2'],
-          },
-          {
-            ID: 'action-1-4-branch2',
-            ActionType: 'Improve',
-            AgentName: '防护工程专家',
-            Description: '根据反馈优化原型功能和界面',
-            ImportantInput: ['ActionResult:action-1-3-branch2'],
-          },
-          {
-            ID: 'action-1-5-branch2',
-            ActionType: 'Finalize',
-            AgentName: '实验材料学家',
-            Description: '确认最终版本的原型系统',
-            ImportantInput: ['ActionResult:action-1-4-branch2'],
-          },
-        ],
-      }
-    } else if (selectedBranchIndex === 2) {
-      // 第三个分支方案：需求深度分析
-      filledTask = {
-        Id: params.stepTask.Id || 'task-1',
-        StepName: '需求深度分析',
-        TaskContent: '深入分析用户需求和技术可行性',
-        InputObject_List: ['腐蚀类型及成因列表'],
-        OutputObject: '详细需求分析报告',
-        AgentSelection: ['腐蚀机理研究员', '防护工程专家'],
-        Collaboration_Brief_frontEnd: {
-          template: '!腐蚀机理研究员!负责!需求深度分析-0!，!防护工程专家!负责!需求深度分析-1!',
-          data: {
-            腐蚀机理研究员: {
-              text: '腐蚀机理研究员',
-              style: { background: 'hsl(210, 70%, 50%)' },
-            },
-            防护工程专家: {
-              text: '防护工程专家',
-              style: { background: 'hsl(30, 70%, 50%)' },
-            },
-          },
-        },
-        TaskProcess: [
-          {
-            ID: 'action-1-1-branch3',
-            ActionType: 'Propose',
-            AgentName: '腐蚀机理研究员',
-            Description: '全面分析腐蚀环境和技术约束条件',
-            ImportantInput: ['InputObject:腐蚀类型及成因列表'],
-          },
-          {
-            ID: 'action-1-2-branch3',
-            ActionType: 'Propose',
-            AgentName: '防护工程专家',
-            Description: '评估技术可行性和防护方案',
-            ImportantInput: ['ActionResult:action-1-1-branch3'],
-          },
-          {
-            ID: 'action-1-3-branch3',
-            ActionType: 'Critique',
-            AgentName: '腐蚀机理研究员',
-            Description: '评审需求分析的完整性和准确性',
-            ImportantInput: ['ActionResult:action-1-2-branch3', 'ActionResult:action-1-1-branch3'],
-          },
-          {
-            ID: 'action-1-4-branch3',
-            ActionType: 'Improve',
-            AgentName: '防护工程专家',
-            Description: '补充技术风险评估和应对策略',
-            ImportantInput: ['ActionResult:action-1-3-branch3'],
-          },
-          {
-            ID: 'action-1-5-branch3',
-            ActionType: 'Finalize',
-            AgentName: '腐蚀机理研究员',
-            Description: '确认详细需求分析报告',
-            ImportantInput: ['ActionResult:action-1-4-branch3', 'ActionResult:action-1-3-branch3'],
-          },
-        ],
-      }
-    } else {
-      // 第四个分支方案：迭代规划
-      filledTask = {
-        Id: params.stepTask.Id || 'task-1',
-        StepName: '迭代规划',
-        TaskContent: '制定敏捷开发迭代计划',
-        InputObject_List: ['腐蚀类型及成因列表'],
-        OutputObject: '迭代计划',
-        AgentSelection: ['防护工程专家', '腐蚀机理研究员'],
-        Collaboration_Brief_frontEnd: {
-          template: '!防护工程专家!负责!迭代规划-0!，!腐蚀机理研究员!负责!迭代规划-1!',
-          data: {
-            防护工程专家: {
-              text: '防护工程专家',
-              style: { background: 'hsl(30, 70%, 50%)' },
-            },
-            腐蚀机理研究员: {
-              text: '腐蚀机理研究员',
-              style: { background: 'hsl(210, 70%, 50%)' },
-            },
-          },
-        },
-        TaskProcess: [
-          {
-            ID: 'action-1-1-branch4',
-            ActionType: 'Propose',
-            AgentName: '防护工程专家',
-            Description: '制定敏捷开发总体框架和迭代周期',
-            ImportantInput: ['InputObject:腐蚀类型及成因列表'],
-          },
-          {
-            ID: 'action-1-2-branch4',
-            ActionType: 'Propose',
-            AgentName: '腐蚀机理研究员',
-            Description: '规划各迭代阶段的技术验证重点',
-            ImportantInput: ['ActionResult:action-1-1-branch4'],
-          },
-          {
-            ID: 'action-1-3-branch4',
-            ActionType: 'Critique',
-            AgentName: '防护工程专家',
-            Description: '评审迭代计划的合理性和可执行性',
-            ImportantInput: ['ActionResult:action-1-2-branch4'],
-          },
-          {
-            ID: 'action-1-4-branch4',
-            ActionType: 'Improve',
-            AgentName: '腐蚀机理研究员',
-            Description: '优化迭代节奏和里程碑设置',
-            ImportantInput: ['ActionResult:action-1-3-branch4'],
-          },
-          {
-            ID: 'action-1-5-branch4',
-            ActionType: 'Finalize',
-            AgentName: '防护工程专家',
-            Description: '确认最终迭代计划',
-            ImportantInput: ['ActionResult:action-1-4-branch4', 'ActionResult:action-1-3-branch4'],
-          },
-        ],
-      }
-    }
-
-    console.log('[Mock API] fill_stepTask 返回数据:', filledTask)
-    return filledTask
-  }
-  // ==================== 第二个任务组 ====================
-  else if (stepName === '系统设计与架构' || stepName === '原型测试与优化' || stepName === '质量保障设计' || stepName === '快速开发与验证') {
-    let filledTask: IRawStepTask
-
-    if (selectedBranchIndex === 0) {
-      // 第一个分支方案：系统设计与架构
-      filledTask = {
-        Id: params.stepTask.Id || 'task-2',
-        StepName: '系统设计与架构',
-        TaskContent: '设计系统架构和数据库结构',
-        InputObject_List: ['项目开发计划书'],
-        OutputObject: '系统设计文档',
-        AgentSelection: ['腐蚀机理研究员', '防护工程专家'],
-        Collaboration_Brief_frontEnd: {
-          template: '!腐蚀机理研究员!负责!系统设计与架构-0!，!防护工程专家!负责!系统设计与架构-1!',
-          data: {
-            腐蚀机理研究员: {
-              text: '腐蚀机理研究员',
-              style: { background: 'hsl(210, 70%, 50%)' },
-            },
-            防护工程专家: {
-              text: '防护工程专家',
-              style: { background: 'hsl(30, 70%, 50%)' },
-            },
-          },
-        },
-        TaskProcess: [
-          {
-            ID: 'action-2-1',
-            ActionType: 'Propose',
-            AgentName: '腐蚀机理研究员',
-            Description: '分析系统功能需求，提出整体架构方案',
-            ImportantInput: ['InputObject:项目开发计划书'],
-          },
-          {
-            ID: 'action-2-2',
-            ActionType: 'Propose',
-            AgentName: '防护工程专家',
-            Description: '设计防护系统架构和数据库模型',
-            ImportantInput: ['ActionResult:action-2-1'],
-          },
-          {
-            ID: 'action-2-3',
-            ActionType: 'Critique',
-            AgentName: '腐蚀机理研究员',
-            Description: '评审架构设计的合理性和可扩展性',
-            ImportantInput: ['ActionResult:action-2-2', 'InputObject:项目开发计划书'],
-          },
-          {
-            ID: 'action-2-4',
-            ActionType: 'Improve',
-            AgentName: '防护工程专家',
-            Description: '优化架构设计，补充性能和安全方案',
-            ImportantInput: ['ActionResult:action-2-3', 'ActionResult:action-2-2'],
-          },
-          {
-            ID: 'action-2-5',
-            ActionType: 'Finalize',
-            AgentName: '腐蚀机理研究员',
-            Description: '确认最终系统架构设计文档',
-            ImportantInput: ['ActionResult:action-2-4', 'ActionResult:action-2-3'],
-          },
-        ],
-      }
-    } else if (selectedBranchIndex === 1) {
-      // 第二个分支方案：原型测试与优化
-      filledTask = {
-        Id: params.stepTask.Id || 'task-2',
-        StepName: '原型测试与优化',
-        TaskContent: '测试原型并根据反馈快速迭代',
-        InputObject_List: ['原型系统'],
-        OutputObject: '优化后的原型',
-        AgentSelection: ['腐蚀机理研究员', '实验材料学家'],
-        Collaboration_Brief_frontEnd: {
-          template: '!腐蚀机理研究员!负责!原型测试与优化-0!，!实验材料学家!负责!原型测试与优化-1!',
-          data: {
-            腐蚀机理研究员: {
-              text: '腐蚀机理研究员',
-              style: { background: 'hsl(210, 70%, 50%)' },
-            },
-            实验材料学家: {
-              text: '实验材料学家',
-              style: { background: 'hsl(120, 70%, 50%)' },
-            },
-          },
-        },
-        TaskProcess: [
-          {
-            ID: 'action-2-1-branch2',
-            ActionType: 'Propose',
-            AgentName: '腐蚀机理研究员',
-            Description: '制定原型测试计划和验证标准',
-            ImportantInput: ['InputObject:原型系统'],
-          },
-          {
-            ID: 'action-2-2-branch2',
-            ActionType: 'Propose',
-            AgentName: '实验材料学家',
-            Description: '执行原型测试并收集性能数据',
-            ImportantInput: ['ActionResult:action-2-1-branch2'],
-          },
-          {
-            ID: 'action-2-3-branch2',
-            ActionType: 'Critique',
-            AgentName: '腐蚀机理研究员',
-            Description: '分析测试结果和用户反馈',
-            ImportantInput: ['ActionResult:action-2-2-branch2'],
-          },
-          {
-            ID: 'action-2-4-branch2',
-            ActionType: 'Improve',
-            AgentName: '实验材料学家',
-            Description: '根据测试结果优化原型功能和性能',
-            ImportantInput: ['ActionResult:action-2-3-branch2'],
-          },
-          {
-            ID: 'action-2-5-branch2',
-            ActionType: 'Finalize',
-            AgentName: '腐蚀机理研究员',
-            Description: '确认优化后的原型版本',
-            ImportantInput: ['ActionResult:action-2-4-branch2', 'ActionResult:action-2-3-branch2'],
-          },
-        ],
-      }
-    } else if (selectedBranchIndex === 2) {
-      // 第三个分支方案：质量保障设计
-      filledTask = {
-        Id: params.stepTask.Id || 'task-2',
-        StepName: '质量保障设计',
-        TaskContent: '设计质量保障体系和测试方案',
-        InputObject_List: ['详细需求分析报告'],
-        OutputObject: '质量保障方案',
-        AgentSelection: ['实验材料学家', '防护工程专家'],
-        Collaboration_Brief_frontEnd: {
-          template: '!实验材料学家!负责!质量保障设计-0!，!防护工程专家!负责!质量保障设计-1!',
-          data: {
-            实验材料学家: {
-              text: '实验材料学家',
-              style: { background: 'hsl(120, 70%, 50%)' },
-            },
-            防护工程专家: {
-              text: '防护工程专家',
-              style: { background: 'hsl(30, 70%, 50%)' },
-            },
-          },
-        },
-        TaskProcess: [
-          {
-            ID: 'action-2-1-branch3',
-            ActionType: 'Propose',
-            AgentName: '实验材料学家',
-            Description: '制定质量标准和测试指标体系',
-            ImportantInput: ['InputObject:详细需求分析报告'],
-          },
-          {
-            ID: 'action-2-2-branch3',
-            ActionType: 'Propose',
-            AgentName: '防护工程专家',
-            Description: '设计质量保障流程和验证方案',
-            ImportantInput: ['ActionResult:action-2-1-branch3'],
-          },
-          {
-            ID: 'action-2-3-branch3',
-            ActionType: 'Critique',
-            AgentName: '实验材料学家',
-            Description: '评审质量保障方案的完整性',
-            ImportantInput: ['ActionResult:action-2-2-branch3'],
-          },
-          {
-            ID: 'action-2-4-branch3',
-            ActionType: 'Improve',
-            AgentName: '防护工程专家',
-            Description: '完善质量保障方案，补充风险控制',
-            ImportantInput: ['ActionResult:action-2-3-branch3'],
-          },
-          {
-            ID: 'action-2-5-branch3',
-            ActionType: 'Finalize',
-            AgentName: '实验材料学家',
-            Description: '确认最终质量保障方案',
-            ImportantInput: ['ActionResult:action-2-4-branch3', 'ActionResult:action-2-3-branch3'],
-          },
-        ],
-      }
-    } else {
-      // 第四个分支方案：快速开发与验证
-      filledTask = {
-        Id: params.stepTask.Id || 'task-2',
-        StepName: '快速开发与验证',
-        TaskContent: '快速开发并持续验证功能',
-        InputObject_List: ['迭代计划'],
-        OutputObject: '可运行的功能模块',
-        AgentSelection: ['实验材料学家', '防护工程专家'],
-        Collaboration_Brief_frontEnd: {
-          template: '!实验材料学家!负责!快速开发与验证-0!，!防护工程专家!负责!快速开发与验证-1!',
-          data: {
-            实验材料学家: {
-              text: '实验材料学家',
-              style: { background: 'hsl(120, 70%, 50%)' },
-            },
-            防护工程专家: {
-              text: '防护工程专家',
-              style: { background: 'hsl(30, 70%, 50%)' },
-            },
-          },
-        },
-        TaskProcess: [
-          {
-            ID: 'action-2-1-branch4',
-            ActionType: 'Propose',
-            AgentName: '实验材料学家',
-            Description: '根据迭代计划快速开发功能模块',
-            ImportantInput: ['InputObject:迭代计划'],
-          },
-          {
-            ID: 'action-2-2-branch4',
-            ActionType: 'Propose',
-            AgentName: '防护工程专家',
-            Description: '实现功能防护机制和验证逻辑',
-            ImportantInput: ['ActionResult:action-2-1-branch4'],
-          },
-          {
-            ID: 'action-2-3-branch4',
-            ActionType: 'Critique',
-            AgentName: '实验材料学家',
-            Description: '验证功能模块的正确性和性能',
-            ImportantInput: ['ActionResult:action-2-2-branch4'],
-          },
-          {
-            ID: 'action-2-4-branch4',
-            ActionType: 'Improve',
-            AgentName: '防护工程专家',
-            Description: '根据验证结果优化功能实现',
-            ImportantInput: ['ActionResult:action-2-3-branch4'],
-          },
-          {
-            ID: 'action-2-5-branch4',
-            ActionType: 'Finalize',
-            AgentName: '实验材料学家',
-            Description: '确认可运行的功能模块',
-            ImportantInput: ['ActionResult:action-2-4-branch4', 'ActionResult:action-2-3-branch4'],
-          },
-        ],
-      }
-    }
-
-    console.log('[Mock API] fill_stepTask 返回数据:', filledTask)
-    return filledTask
-  }
-  // ==================== 第三个任务组（仅第三个分支方案） ====================
-  else if (stepName === '系统开发与测试') {
-    const filledTask: IRawStepTask = {
-      Id: params.stepTask.Id || 'task-3',
-      StepName: '系统开发与测试',
-      TaskContent: '按质量标准进行系统开发和测试',
-      InputObject_List: ['质量保障方案'],
-      OutputObject: '经过完整测试的系统',
-      AgentSelection: ['腐蚀机理研究员', '实验材料学家'],
-      Collaboration_Brief_frontEnd: {
-        template: '!腐蚀机理研究员!负责!系统开发与测试-0!，!实验材料学家!负责!系统开发与测试-1!',
-        data: {
-          腐蚀机理研究员: {
-            text: '腐蚀机理研究员',
-            style: { background: 'hsl(210, 70%, 50%)' },
-          },
-          实验材料学家: {
-            text: '实验材料学家',
-            style: { background: 'hsl(120, 70%, 50%)' },
-          },
-        },
-      },
-      TaskProcess: [
-        {
-          ID: 'action-3-1',
-          ActionType: 'Propose',
-          AgentName: '腐蚀机理研究员',
-          Description: '根据质量标准制定系统开发计划',
-          ImportantInput: ['InputObject:质量保障方案'],
-        },
-        {
-          ID: 'action-3-2',
-          ActionType: 'Propose',
-          AgentName: '实验材料学家',
-          Description: '按标准实施系统开发和单元测试',
-          ImportantInput: ['ActionResult:action-3-1'],
-        },
-        {
-          ID: 'action-3-3',
-          ActionType: 'Critique',
-          AgentName: '腐蚀机理研究员',
-          Description: '评审系统开发质量和测试覆盖率',
-          ImportantInput: ['ActionResult:action-3-2'],
-        },
-        {
-          ID: 'action-3-4',
-          ActionType: 'Improve',
-          AgentName: '实验材料学家',
-          Description: '根据评审结果完善系统功能和测试',
-          ImportantInput: ['ActionResult:action-3-3'],
-        },
-        {
-          ID: 'action-3-5',
-          ActionType: 'Finalize',
-          AgentName: '腐蚀机理研究员',
-          Description: '确认经过完整测试的系统版本',
-          ImportantInput: ['ActionResult:action-3-4', 'ActionResult:action-3-3'],
-        },
-      ],
-    }
-
-    console.log('[Mock API] fill_stepTask 返回数据:', filledTask)
-    return filledTask
-  }
-  // ==================== 其他任务 ====================
-  else {
-    // 其他任务，返回空 TaskProcess
-    console.warn('[Mock API] 未知的 StepName:', stepName)
-    return {
-      ...params.stepTask,
-      AgentSelection: [],
-      TaskProcess: [],
-    }
-  }
-}
-
-export default mockFillStepTaskAPI

frontend/src/layout/components/Main/TaskTemplate/TaskSyllabus/components/AgentAllocation.vue

@@ -1,11 +1,16 @@
 <script setup lang="ts">
 import { ref, computed, watch, nextTick } from 'vue'
 import { useAgentsStore, useSelectionStore } from '@/stores'
+import { convertToTaskProcess } from '@/stores/modules/agents'
 import { getAgentMapIcon } from '@/layout/components/config'
 import SvgIcon from '@/components/SvgIcon/index.vue'
 import api from '@/api'
 const agentsStore = useAgentsStore()
 const selectionStore = useSelectionStore()
+const agentsHeatmapData = ref<AgentHeatmapData[]>([])
+const selectedAgents = ref<Set<string>>(new Set())
+const selectedAssignmentGroup = ref<string[]>([])
+const selectedDimensions = ref<Set<string>>(new Set())
 
 // 获取当前选中的任务
 const currentTask = computed(() => {
@@ -17,7 +22,7 @@ const currentTaskAgents = computed(() => {
   return currentTask.value?.AgentSelection || []
 })
 
-// 所有agent列表（用于右侧展示）
+// 所有agent列表
 const allAgents = computed(() => {
   return agentsStore.agents
 })
@@ -32,19 +37,7 @@ interface AgentHeatmapData {
   scoreDetails?: Array<{ dimension: string; score: number; reason: string }> // 详细信息
 }
 
-// 所有agent的热力图数据
+// 确认的agent组合列表 - 从 store 读取
-const agentsHeatmapData = ref<AgentHeatmapData[]>([])
-
-// 选中的agent集合（用于Comparison部分）
-const selectedAgents = ref<Set<string>>(new Set())
-
-// 当前选中的Assignment组合的agents（用于Assignment卡片的边框显示）
-const selectedAssignmentGroup = ref<string[]>([])
-
-// 选中的维度集合
-const selectedDimensions = ref<Set<string>>(new Set())
-
-// 确认的agent组合列表 - 从 store 读取（按当前任务ID）
 const confirmedAgentGroups = computed(() => {
   const taskId = currentTask.value?.Id
   if (!taskId) return []
@@ -59,36 +52,44 @@ const isInitializing = ref(false)
 const isAddingDimension = ref(false)
 const isLoadingConfirm = ref(false) // 确认 agent 组合时的加载状态
 const isLoadingSelectGroup = ref(false) // 选择已保存组合时的加载状态
-const isLoadingInitialTask = ref(false) // 首次加载任务时的加载状态
 
+// 热力图排序：选中的 agent 排在前面，其他按平均分降序
+const sortHeatmapBySelection = () => {
+  if (agentsHeatmapData.value.length === 0) return
+  agentsHeatmapData.value.sort((a, b) => {
+    const aIsSelected = selectedAgents.value.has(a.agentName)
+    const bIsSelected = selectedAgents.value.has(b.agentName)
+
+    // 如果一个是选中的 agent，另一个不是，选中的 agent 排前面
+    if (aIsSelected && !bIsSelected) return -1
+    if (!aIsSelected && bIsSelected) return 1
+
+    // 如果都是或都不是选中的 agent，按选中的维度平均分降序排列
+    const selectedDimArray = Array.from(selectedDimensions.value)
+    const aAverage = calculateAgentAverage(a, selectedDimArray)
+    const bAverage = calculateAgentAverage(b, selectedDimArray)
+    return bAverage - aAverage
+  })
+}
 // 处理搜索提交
 const handleSubmit = async () => {
   if (!searchValue.value.trim() || isAddingDimension.value) return
 
   const newDimension = searchValue.value.trim()
-
   // 检查维度是否已存在
   if (scoreDimensions.value.includes(newDimension)) {
-    console.log(`维度"${newDimension}"已存在，跳过添加`)
     searchValue.value = ''
     return
   }
-
   try {
     isAddingDimension.value = true
-
+    const response = await api.agentSelectModifyAddAspect({
-    // 使用 Mock API 进行测试（开发阶段）
-    // 切换到真实API时，只需将 mockAgentSelectModifyAddAspect 改为 agentSelectModifyAddAspect
-    const response = await api.mockAgentSelectModifyAddAspect({
       aspectList: [...scoreDimensions.value, newDimension]
     })
 
-    console.log('Mock API - 添加新维度成功:', response)
-
-    // 1. 先添加新维度到本地维度列表（立即显示）
     scoreDimensions.value.push(response.aspectName)
 
-    // 2. 为每个agent添加新维度的评分（立即显示）
+    // 为每个agent添加新维度的评分（立即显示）
     agentsHeatmapData.value.forEach(agentData => {
       const scoreInfo = response.agentScores[agentData.agentName]
       if (scoreInfo) {
@@ -100,28 +101,42 @@ const handleSubmit = async () => {
         })
       }
     })
-
+    //异步更新 store（等前端显示完成后再更新，避免触发重新初始化）
-    console.log(`已添加新维度"${response.aspectName}"到热力图`)
-
-    // 3. 异步更新 store（等前端显示完成后再更新，避免触发重新初始化）
     await nextTick()
+    const taskId = currentTask.value?.Id
 
+    //更新按任务ID的存储
+    if (taskId) {
+      const existingTaskData = agentsStore.getTaskScoreData(taskId)
+      if (existingTaskData) {
+        // 检查维度是否已存在
+        if (!existingTaskData.aspectList.includes(response.aspectName)) {
+          existingTaskData.aspectList.push(response.aspectName)
+        }
+        // 更新评分数据
+        for (const [agentName, scoreInfo] of Object.entries(response.agentScores)) {
+          if (!existingTaskData.agentScores[agentName]) {
+            existingTaskData.agentScores[agentName] = {}
+          }
+          existingTaskData.agentScores[agentName][response.aspectName] = scoreInfo
+        }
+        // 保存到 store
+        agentsStore.setTaskScoreData(taskId, existingTaskData)
+      }
+    }
+
+    // 兼容旧版本：更新全局存储
     const currentStoreData = agentsStore.IAgentSelectModifyAddRequest
     if (currentStoreData && currentStoreData.aspectList) {
-      // 检查维度是否已存在（防止store中已存在）
       if (!currentStoreData.aspectList.includes(response.aspectName)) {
         currentStoreData.aspectList.push(response.aspectName)
       }
-      // 更新评分数据
-      // response.agentScores: { agentName: { score, reason } }
-      // 转换为 agentScores[agentName][aspectName] = { score, reason }
       for (const [agentName, scoreInfo] of Object.entries(response.agentScores)) {
         if (!currentStoreData.agentScores[agentName]) {
           currentStoreData.agentScores[agentName] = {}
         }
         currentStoreData.agentScores[agentName][response.aspectName] = scoreInfo
       }
-      console.log(`已更新 store`)
     }
 
     // 清空输入框
@@ -133,7 +148,7 @@ const handleSubmit = async () => {
   }
 }
 
-// 判断两个agent组合是否相同（不考虑顺序）
+// 判断两个agent组合是否相同
 const areAgentGroupsEqual = (group1: string[], group2: string[]): boolean => {
   if (group1.length !== group2.length) return false
   const sorted1 = [...group1].sort()
@@ -145,7 +160,6 @@ const areAgentGroupsEqual = (group1: string[], group2: string[]): boolean => {
 const confirmAgentSelection = async () => {
   if (selectedAgents.value.size > 0 && currentTask.value?.Id) {
     const agentArray = Array.from(selectedAgents.value)
-
     // 检查该agent组合是否已存在（统一检查，包括初始组合和用户添加的组合）
     const existingGroups = agentsStore.getConfirmedAgentGroups(currentTask.value.Id)
     const existingDuplicateIndex = existingGroups.findIndex(group =>
@@ -153,64 +167,30 @@ const confirmAgentSelection = async () => {
     )
 
     if (existingDuplicateIndex !== -1) {
-      // 找到重复的组合，只选中该卡片，不添加重复
-      console.log('该agent组合已存在，只选中卡片，不添加重复')
       const existingGroup = existingGroups[existingDuplicateIndex]!
       selectedAssignmentGroup.value = [...existingGroup]
       selectedAgents.value = new Set(existingGroup)
       return
     }
-
-    // 该组合不存在，添加新组合
-    // 使用 store 的方法添加，按任务ID存储（不持久化到localStorage）
     agentsStore.addConfirmedAgentGroup(currentTask.value.Id, agentArray)
-    console.log(`已确认 agent 组合 [任务 ${currentTask.value.Id}]:`, agentArray)
-
-    // 调用 Mock API 填充任务流程
     try {
       isLoadingConfirm.value = true
-      console.log('=== 开始调用 mockFillStepTaskTaskProcess Mock API ===')
+      const stepTaskForApi = agentsStore.createStepTaskForApi(agentArray)
-      console.log('1. 当前任务数据 (IRawStepTask 格式):', currentTask.value)
+      const goal = agentsStore.agentRawPlan.data?.['General Goal'] || '开发智能协作系统'
-      console.log('2. 选中的 agents:', agentArray)
 
-      // 将 IRawStepTask 转换为 IApiStepTask 格式
+      if (agentsStore.isStopping || agentsStore.hasStoppedFilling) {
-      const stepTaskForApi = {
+        isLoadingConfirm.value = false
-        name: currentTask.value.StepName || '',
+        return
-        content: currentTask.value.TaskContent || '',
-        inputs: currentTask.value.InputObject_List || [],
-        output: currentTask.value.OutputObject || '',
-        agents: agentArray,
-        brief: currentTask.value.Collaboration_Brief_frontEnd || {
-          template: '',
-          data: {}
-        },
-        process: []
       }
 
-      console.log('3. 转换后的 API 请求参数 (IApiStepTask 格式):', stepTaskForApi)
+      const filledTask = await api.fillStepTaskTaskProcess({
-
-      // 获取 goal（从 agentRawPlan 中获取）
-      const goal = agentsStore.agentRawPlan.data?.['General Goal'] || '开发智能协作系统'
-      console.log('4. General Goal:', goal)
-
-      // 调用 Mock API
-      const filledTask = await api.mockFillStepTaskTaskProcess({
         goal,
         stepTask: stepTaskForApi,
         agents: agentArray
       })
-
-      console.log('=== Mock API 返回成功 ===')
-      console.log('5. 返回的完整数据 (IApiStepTask 格式):', filledTask)
-      console.log('6. TaskProcess 流程数量:', filledTask.process?.length || 0)
-      console.log('7. TaskProcess 详情:', filledTask.process)
-      console.log('8. Collaboration Brief:', filledTask.brief)
-
-      // 存储到 selectionStore
       selectionStore.setAgentTaskProcess(currentTask.value.Id, agentArray, filledTask)
-      console.log('✅ Mock API 调用成功，数据已存储到 selectionStore')
     } catch (error) {
-      console.error('❌ Mock API 调用失败:', error)
+      console.error('❌ API 调用失败:', error)
     } finally {
       isLoadingConfirm.value = false
     }
@@ -219,84 +199,35 @@ const confirmAgentSelection = async () => {
 
 // 点击 Assignment 部分的 agent 组合卡片，更新 Comparison 部分的选中状态
 const selectAgentGroup = async (agentNames: string[]) => {
-  // 更新Assignment边框状态
   selectedAssignmentGroup.value = [...agentNames]
-
-  // 更新Comparison部分的选中状态
   selectedAgents.value = new Set(agentNames)
-  // 触发响应式更新
   selectedAgents.value = new Set(selectedAgents.value)
 
-  // 保存到 store（持久化选择状态）
+  // 保存到 store
   if (currentTask.value?.Id) {
     agentsStore.setSelectedAgentGroup(currentTask.value.Id, agentNames)
   }
 
-  // 🆕 联动更新：更新 currentTask 的 AgentSelection 和 TaskProcess
+  // 更新 currentTask 的 AgentSelection 和 TaskProcess
   if (currentTask.value?.Id && agentNames.length > 0) {
-    console.log('🔄 开始联动更新 currentTask 的 agent 组合:', agentNames)
-
-    // 从 selectionStore 获取该 agent 组合对应的 TaskProcess 数据
     let taskProcessData = selectionStore.getAgentTaskProcess(currentTask.value.Id, agentNames)
 
-    // 🆕 如果 selectionStore 中没有数据，自动调用 API 加载
     if (!taskProcessData) {
-      console.log('⚠️ selectionStore 中没有该组合的 TaskProcess 数据，开始加载...')
-      console.log('📋 当前任务信息:', {
-        taskId: currentTask.value.Id,
-        taskName: currentTask.value.StepName,
-        agents: agentNames
-      })
-
       try {
         isLoadingSelectGroup.value = true
-        // 将 IRawStepTask 转换为 IApiStepTask 格式
+        const stepTaskForApi = agentsStore.createStepTaskForApi(agentNames)
-        const stepTaskForApi = {
-          name: currentTask.value.StepName || '',
-          content: currentTask.value.TaskContent || '',
-          inputs: currentTask.value.InputObject_List || [],
-          output: currentTask.value.OutputObject || '',
-          agents: agentNames,
-          brief: currentTask.value.Collaboration_Brief_frontEnd || {
-            template: '',
-            data: {}
-          },
-          process: []
-        }
-
         const goal = agentsStore.agentRawPlan.data?.['General Goal'] || '开发智能协作系统'
 
-        console.log('📤 准备调用 API:', {
+        if (agentsStore.isStopping || agentsStore.hasStoppedFilling) {
+          return
+        }
+
+        const filledTask = await api.fillStepTaskTaskProcess({
           goal,
           stepTask: stepTaskForApi,
           agents: agentNames
         })
-
-        // 调用 Mock API 加载数据
-        const filledTask = await api.mockFillStepTaskTaskProcess({
-          goal,
-          stepTask: stepTaskForApi,
-          agents: agentNames
-        })
-
-        console.log('🔍 API 返回的完整响应:', filledTask)
-        console.log('🔍 process 数据:', filledTask.process)
-        console.log('🔍 process 长度:', filledTask.process?.length)
-        console.log('✅ TaskProcess 数据加载成功:', {
-          agents: agentNames,
-          processCount: filledTask.process?.length || 0,
-          processDetail: filledTask.process,
-          briefDetail: filledTask.brief
-        })
-
-        // 存储到 selectionStore
         selectionStore.setAgentTaskProcess(currentTask.value.Id, agentNames, filledTask)
-
-        console.log('💾 数据已存储到 selectionStore，验证存储是否成功...')
-        // 验证存储
-        const storedData = selectionStore.getAgentTaskProcess(currentTask.value.Id, agentNames)
-        console.log('📦 验证存储结果:', storedData ? '成功' : '失败', storedData)
-
         taskProcessData = filledTask
       } catch (error) {
         console.error('❌ 加载 TaskProcess 数据失败:', error)
@@ -307,62 +238,17 @@ const selectAgentGroup = async (agentNames: string[]) => {
     }
 
     if (taskProcessData) {
-      console.log('✅ 找到 TaskProcess 数据，开始更新 currentTask:', {
+      const convertedTaskProcess = convertToTaskProcess(taskProcessData.process || [])
-        taskId: currentTask.value.Id,
-        agents: agentNames,
-        processCount: taskProcessData.process?.length || 0,
-        processDetail: taskProcessData.process,
-        briefDetail: taskProcessData.brief
-      })
-
-      // 🆕 数据格式转换：IApiAgentAction[] → TaskProcess[]
-      const convertedTaskProcess = (taskProcessData.process || []).map(action => ({
-        ID: action.id,
-        ActionType: action.type,
-        AgentName: action.agent,
-        Description: action.description,
-        ImportantInput: action.inputs || []
-      }))
-
-      console.log('🔄 数据格式转换完成:', {
-        original: taskProcessData.process,
-        converted: convertedTaskProcess
-      })
-
-      // 更新 currentTask 的 AgentSelection 和 TaskProcess
-      // 使用 updateCurrentAgentSelection 强制更新，避免被 setCurrentTask 的"智能保留"逻辑阻止
       agentsStore.updateCurrentAgentSelection(
         [...agentNames],
         convertedTaskProcess,
         taskProcessData.brief || currentTask.value.Collaboration_Brief_frontEnd
       )
-
-      console.log('✅ currentTask 已更新，新的 AgentSelection:', agentNames)
-      console.log('📋 验证更新结果:', agentsStore.currentTask)
-    }
     }
 
-  // 重新排序：被选中的agent排在前面，其他agent按平均分降序
+    sortHeatmapBySelection()
-  if (agentsHeatmapData.value.length > 0) {
-    agentsHeatmapData.value.sort((a, b) => {
-      const aIsSelected = selectedAgents.value.has(a.agentName)
-      const bIsSelected = selectedAgents.value.has(b.agentName)
-
-      // 如果一个是选中的agent，另一个不是，选中的agent排前面
-      if (aIsSelected && !bIsSelected) return -1
-      if (!aIsSelected && bIsSelected) return 1
-
-      // 如果都是或都不是选中的agent，按选中的维度平均分降序排列
-      const selectedDimArray = Array.from(selectedDimensions.value)
-      const aAverage = calculateAgentAverage(a, selectedDimArray)
-      const bAverage = calculateAgentAverage(b, selectedDimArray)
-      return bAverage - aAverage
-    })
   }
-
-  console.log('选中 agent 组合:', Array.from(selectedAgents.value))
 }
-
 // 切换维度选中状态
 const toggleDimensionSelection = (dimension: string) => {
   if (selectedDimensions.value.has(dimension)) {
@@ -370,26 +256,8 @@ const toggleDimensionSelection = (dimension: string) => {
   } else {
     selectedDimensions.value.add(dimension)
   }
-  // 触发响应式更新
   selectedDimensions.value = new Set(selectedDimensions.value)
-
+  sortHeatmapBySelection()
-  // 重新排序agent
-  if (agentsHeatmapData.value.length > 0) {
-    agentsHeatmapData.value.sort((a, b) => {
-      const aIsSelected = selectedAgents.value.has(a.agentName)
-      const bIsSelected = selectedAgents.value.has(b.agentName)
-
-      // 如果一个是选中的agent，另一个不是，选中的agent排前面
-      if (aIsSelected && !bIsSelected) return -1
-      if (!aIsSelected && bIsSelected) return 1
-
-      // 如果都是或都不是选中的agent，按选中的维度平均分降序排列
-      const selectedDimArray = Array.from(selectedDimensions.value)
-      const aAverage = calculateAgentAverage(a, selectedDimArray)
-      const bAverage = calculateAgentAverage(b, selectedDimArray)
-      return bAverage - aAverage
-    })
-  }
 }
 
 // 判断维度是否被选中
@@ -404,31 +272,8 @@ const toggleAgentSelection = (agentName: string) => {
   } else {
     selectedAgents.value.add(agentName)
   }
-  // 触发响应式更新
   selectedAgents.value = new Set(selectedAgents.value)
-
+  sortHeatmapBySelection()
-  // 重新排序：被选中的agent排在前面，其他agent按平均分降序
-  if (agentsHeatmapData.value.length > 0) {
-    agentsHeatmapData.value.sort((a, b) => {
-      const aIsSelected = selectedAgents.value.has(a.agentName)
-      const bIsSelected = selectedAgents.value.has(b.agentName)
-
-      // 如果一个是选中的agent，另一个不是，选中的agent排前面
-      if (aIsSelected && !bIsSelected) return -1
-      if (!aIsSelected && bIsSelected) return 1
-
-      // 如果都是或都不是选中的agent，按选中的维度平均分降序排列
-      const selectedDimArray = Array.from(selectedDimensions.value)
-      const aAverage = calculateAgentAverage(a, selectedDimArray)
-      const bAverage = calculateAgentAverage(b, selectedDimArray)
-      return bAverage - aAverage
-    })
-  }
-}
-
-// 判断agent是否被选中
-const isAgentSelected = (agentName: string) => {
-  return selectedAgents.value.has(agentName)
 }
 
 // 判断agent组合是否被选中（用于Assignment卡片的选中效果）
@@ -498,33 +343,46 @@ const calculateAgentAverage = (agentData: AgentHeatmapData, selectedDimensions?:
   return count > 0 ? total / count : 0
 }
 
-// 模拟API调用 - 获取智能体评分数据
+// API调用 - 获取智能体评分数据
 const fetchAgentScores = async () => {
-  // 先检查 store 中是否已有数据
+  const taskId = currentTask.value?.Id
-  const storedData = agentsStore.IAgentSelectModifyAddRequest
+  if (!taskId) {
+    console.warn('⚠️ fetchAgentScores: 当前任务没有 Id')
+    return null
+  }
+  //先检查 store 中是否有该任务的评分数据
+  const storedData = agentsStore.getTaskScoreData(taskId)
   if (storedData && storedData.aspectList && storedData.aspectList.length > 0) {
-    console.log('使用 store 中的评分数据')
+    return storedData
-    return {
+  }
-      aspectList: storedData.aspectList,
+
-      agentScores: storedData.agentScores
+  // 如果正在预加载中，等待预加载完成
+  if (agentsStore.isTaskPreloading(taskId)) {
+    // 等待一小段时间，让预加载完成
+    await new Promise(resolve => setTimeout(resolve, 1000))
+    // 再次尝试获取
+    const cachedAfterPreload = agentsStore.getTaskScoreData(taskId)
+    if (cachedAfterPreload) {
+      return cachedAfterPreload
     }
   }
 
-  // TODO: 切换到真实API时，取消注释下面这行
+  // 检查是否已停止
-  // const agentScores = await api.agentSelectModifyInit({
+  if (agentsStore.isStopping || agentsStore.hasStoppedFilling) {
-  //   goal: agentsStore.agentRawPlan.data?.['General Goal'] || '',
+    console.log('检测到停止信号，跳过获取智能体评分')
-  //   stepTask: currentTask.value
+    return null
-  // })
+  }
 
-  // 🆕 使用 Mock API（开发阶段），传递当前任务的 AgentSelection 以获取对应的维度
+  // 调用 API 获取评分（如果没有缓存或预加载）
-  const goal = agentsStore.agentRawPlan.data?.['General Goal'] || '开发智能协作系统'
+  const agentScores = await api.agentSelectModifyInit({
-  console.log('📤 调用 mockAgentSelectModifyInit，参数:', {
+    goal: agentsStore.agentRawPlan.data?.['General Goal'] || '',
-    goal,
+    stepTask: currentTask.value
-    stepTask: currentTask.value,
-    agentSelection: currentTask.value?.AgentSelection
   })
 
-  const agentScores = await api.mockAgentSelectModifyInit()
+  // 再次检查是否已停止（API 调用后）
+  if (agentsStore.isStopping || agentsStore.hasStoppedFilling) {
+    return null
+  }
 
   // 从转换后的数据中提取维度列表（第一个agent的所有维度）
   const firstAgent = Object.keys(agentScores)[0]
@@ -532,8 +390,8 @@ const fetchAgentScores = async () => {
 
   console.log('✅ 获取到的维度列表:', aspectList)
 
-  // 保存到 store
+  // 🆕 保存到 store（按任务ID存储）
-  agentsStore.setAgentScoreData({
+  agentsStore.setTaskScoreData(taskId, {
     aspectList,
     agentScores
   })
@@ -548,9 +406,7 @@ const fetchAgentScores = async () => {
 const initHeatmapData = async () => {
   const agents = allAgents.value
   if (agents.length === 0 || isInitializing.value || isAddingDimension.value) return
-
   isInitializing.value = true
-
   // 先创建一个只包含agent基本信息的临时数据，让agent头像先显示
   const tempData = agents.map(agent => ({
     agentName: agent.Name,
@@ -562,16 +418,12 @@ const initHeatmapData = async () => {
   tempData.sort((a, b) => {
     const aIsAssigned = currentTaskAgents.value.includes(a.agentName)
     const bIsAssigned = currentTaskAgents.value.includes(b.agentName)
-
     if (aIsAssigned && !bIsAssigned) return -1
     if (!aIsAssigned && bIsAssigned) return 1
-
     return a.agentName.localeCompare(b.agentName, 'zh-CN')
   })
-
   agentsHeatmapData.value = tempData
-
+  //默认选中Assignment部分的agent（即currentTaskAgents）
-  // 初始化选中状态：默认选中Assignment部分的agent（即currentTaskAgents）
   if (currentTaskAgents.value.length > 0) {
     selectedAgents.value = new Set(currentTaskAgents.value)
     selectedAssignmentGroup.value = [...currentTaskAgents.value]
@@ -584,32 +436,33 @@ const initHeatmapData = async () => {
 
   try {
     // 获取接口数据
-    const { aspectList, agentScores } = await fetchAgentScores()
+    const scoreData = await fetchAgentScores()
+    if (!scoreData) {
+      console.warn('⚠️ initHeatmapData: 没有获取到评分数据')
+      return
+    }
+
+    const { aspectList, agentScores } = scoreData
 
-    // 设置评分维度
     scoreDimensions.value = aspectList
-
-    // 转换数据结构：为每个agent生成评分数据
     const heatmapData = agents.map(agent => {
       const agentName = agent.Name
       const scores: number[] = []
       const scoreDetails: Array<{ dimension: string; score: number; reason: string }> = []
 
       // 按照维度顺序获取评分
-      aspectList.forEach(dimension => {
+      aspectList.forEach((dimension: string) => {
-        // 数据结构：agentScores[agentName][dimension]
         const agentScoreData = agentScores[agentName]
-        const scoreData = agentScoreData ? agentScoreData[dimension] : null
+        const scoreItem = agentScoreData ? agentScoreData[dimension] : null
 
-        if (scoreData) {
+        if (scoreItem) {
-          scores.push(scoreData.score)
+          scores.push(scoreItem.score)
           scoreDetails.push({
             dimension,
-            score: scoreData.score,
+            score: scoreItem.score,
-            reason: scoreData.reason
+            reason: scoreItem.reason
           })
         } else {
-          // 如果没有评分数据，默认为1
           scores.push(1)
         }
       })
@@ -620,9 +473,8 @@ const initHeatmapData = async () => {
         scoreDetails
       }
     })
-
     // 排序逻辑：
-    // 1. Assignment部分的agent（currentTaskAgents）排在前面
+    // 1. Assignment部分的agent排在前面
     // 2. 其他agent按照平均分降序排列
     // 3. 同一组内也按平均分降序排列
     heatmapData.sort((a, b) => {
@@ -644,27 +496,16 @@ const initHeatmapData = async () => {
     agentsHeatmapData.value = heatmapData
   } catch (error) {
     console.error('获取智能体评分数据失败:', error)
-    // 降级处理：使用默认维度列表和随机数据
-    const defaultAspects = ['能力', '可用性', '专业性', '效率', '准确性', '协作性']
-    scoreDimensions.value = defaultAspects
-    agentsHeatmapData.value = tempData.map(item => ({
-      ...item,
-      scores: Array.from(
-        { length: defaultAspects.length },
-        () => Math.floor(Math.random() * 5) + 1
-      ),
-      scoreDetails: []
-    }))
   } finally {
     isInitializing.value = false
   }
 }
 
-// 监听agent列表变化，初始化热力矩阵
+//监听agent列表变化或任务切换，初始化热力矩阵
 watch(
-  () => allAgents.value.length,
+  [() => allAgents.value.length, () => currentTask.value?.Id],
-  newLength => {
+  ([agentLength, taskId]) => {
-    if (newLength > 0) {
+    if (agentLength > 0 && taskId) {
       initHeatmapData()
     }
   },
@@ -709,9 +550,6 @@ watch(
 )
 
 // 初始化和任务切换处理
-// 1. 首次加载时，自动将初始 agent 组合添加到 confirmedAgentGroups
-// 2. 恢复之前的选择状态
-// 3. 加载 TaskProcess 数据
 watch(
   () => currentTask.value,
   async newTask => {
@@ -723,75 +561,37 @@ watch(
     const existingGroups = agentsStore.getConfirmedAgentGroups(newTask.Id)
     const hasInitialGroup = existingGroups.length > 0
 
-    // 首次初始化：自动添加初始组合（相当于系统自动执行了 confirmAgentSelection）
+    // 首次初始化：自动添加初始组合到 confirmedAgentGroups
     if (!hasInitialGroup) {
-      console.log('🎯 首次初始化，自动添加初始 agent 组合到 Assignment')
       agentsStore.addConfirmedAgentGroup(newTask.Id, [...newTask.AgentSelection])
 
-      // 调用 API 获取初始 agent 组合的 TaskProcess 数据
+      // 初始化时 TaskProcess 一定存在，直接存储
-      console.log('🔄 开始加载初始 agent 组合的 TaskProcess...')
+      selectionStore.setAgentTaskProcess(newTask.Id, newTask.AgentSelection, {
-
-      // 检查是否已有数据
-      if (!selectionStore.hasAgentTaskProcess(newTask.Id, newTask.AgentSelection)) {
-        try {
-          isLoadingInitialTask.value = true
-          console.log('=== 开始加载初始 agent 组合的 TaskProcess ===')
-          console.log('1. 任务ID:', newTask.Id)
-          console.log('2. 初始 agents:', newTask.AgentSelection)
-
-          // 将 IRawStepTask 转换为 IApiStepTask 格式
-          const stepTaskForApi = {
         name: newTask.StepName || '',
         content: newTask.TaskContent || '',
         inputs: newTask.InputObject_List || [],
         output: newTask.OutputObject || '',
         agents: newTask.AgentSelection,
-            brief: newTask.Collaboration_Brief_frontEnd || {
+        brief: newTask.Collaboration_Brief_frontEnd || { template: '', data: {} },
-              template: '',
+        process: newTask.TaskProcess.map(action => ({
-              data: {}
+          id: action.ID,
-            },
+          type: action.ActionType,
-            process: []
+          agent: action.AgentName,
-          }
+          description: action.Description,
-
+          inputs: action.ImportantInput
-          console.log('3. 转换后的 API 请求参数:', stepTaskForApi)
+        }))
-
-          const goal = agentsStore.agentRawPlan.data?.['General Goal'] || '开发智能协作系统'
-          console.log('4. General Goal:', goal)
-
-          // 调用 Mock API 获取 TaskProcess
-          const filledTask = await api.mockFillStepTaskTaskProcess({
-            goal,
-            stepTask: stepTaskForApi,
-            agents: newTask.AgentSelection
       })
-
-          console.log('=== 初始 agent 组合 TaskProcess 加载成功 ===')
-          console.log('5. TaskProcess 流程数量:', filledTask.process?.length || 0)
-
-          // 存储到 selectionStore
-          selectionStore.setAgentTaskProcess(newTask.Id, newTask.AgentSelection, filledTask)
-          console.log('✅ 初始 agent 组合的 TaskProcess 已加载并存储')
-        } catch (error) {
-          console.error('❌ 加载初始 agent 组合的 TaskProcess 失败:', error)
-        } finally {
-          isLoadingInitialTask.value = false
-        }
-      } else {
-        console.log('ℹ️ 初始 agent 组合已有 TaskProcess 数据，跳过加载')
-      }
     }
 
-    // 从 store 中恢复之前选择的 agent 组合（如果有）
+    // 从 store 中恢复之前选择的 agent 组合
     const savedAgentGroup = agentsStore.getSelectedAgentGroup(newTask.Id)
     if (savedAgentGroup) {
-      console.log('📂 恢复之前选择的 agent 组合:', savedAgentGroup)
       selectedAssignmentGroup.value = [...savedAgentGroup]
       selectedAgents.value = new Set(savedAgentGroup)
     } else {
-      // 没有保存的选择，默认选中第一个组合（即初始组合）
+      // 没有保存的选择，默认选中初始组合
       const allGroups = agentsStore.getConfirmedAgentGroups(newTask.Id)
       if (allGroups.length > 0 && allGroups[0]) {
-        console.log('🔄 默认选中第一个组合（初始组合）:', allGroups[0])
         selectedAssignmentGroup.value = [...allGroups[0]]
         selectedAgents.value = new Set(allGroups[0])
       }
@@ -799,7 +599,6 @@ watch(
   },
   { immediate: true } // 立即执行一次
 )
-
 // 获取热力图颜色（评分1-5，颜色从浅到深）
 const getHeatmapColor = (score: number) => {
   const colors = [
@@ -818,7 +617,7 @@ const getHeatmapColor = (score: number) => {
     <!-- 左侧区域 - 20% -->
     <div class="allocation-left">
       <div class="section-card" v-loading="isLoadingSelectGroup">
-        <div class="section-title">Assignment</div>
+        <div class="section-title">智能体</div>
         <!-- 所有 agent 组合卡片（包括初始组合和用户创建的组合） -->
         <div
           v-for="(group, groupIndex) in confirmedAgentGroups"
@@ -853,7 +652,6 @@ const getHeatmapColor = (score: number) => {
             </div>
           </el-tooltip>
         </div>
-        <div v-if="!currentTask" class="placeholder-text">请先在任务大纲中选择一个任务</div>
       </div>
     </div>
 
@@ -863,11 +661,11 @@ const getHeatmapColor = (score: number) => {
     <!-- 右侧区域 - 79% -->
     <div class="allocation-right">
       <div class="section-card">
-        <div class="section-title">Comparison</div>
+        <div class="section-title">智能体对比</div>
         <div
           v-if="allAgents.length > 0"
           class="comparison-content"
-          v-loading="isInitializing || isLoadingInitialTask || isLoadingConfirm"
+          v-loading="isInitializing || isLoadingConfirm"
         >
           <!-- 热力矩阵图 -->
           <div v-if="agentsHeatmapData.length > 0" class="heatmap-container">

frontend/src/layout/components/Main/TaskTemplate/TaskSyllabus/components/AssignmentButton.vue

@@ -14,14 +14,14 @@ const isDisabled = computed(() => {
   return !agentsStore.currentTask
 })
 
-// 获取agent组合卡片数量 - 当前任务agents(1) + 已确认的agent组合数量
+// 获取agent组合卡片数量
 const agentGroupCount = computed(() => {
   if (!agentsStore.currentTask?.Id) return 1
 
   // 获取该任务的已确认agent组合
   const confirmedGroups = agentsStore.getConfirmedAgentGroups(agentsStore.currentTask.Id)
 
-  // 当前任务agents(1) + 已确认的agent组合数量
+  // 当前任务agents数量 + 已确认的agent组合数量
   return confirmedGroups.length || 1
 })
 

frontend/src/layout/components/Main/TaskTemplate/TaskSyllabus/components/BranchButton.vue

@@ -8,25 +8,21 @@ const emit = defineEmits<{
   (e: 'click'): void
 }>()
 
-// 获取分支数量 - 主分支(1) + 额外分支数量
+// 获取分支数量
 const branchCount = computed(() => {
-  // flowBranches 包含所有通过 Branch 创建的分支
   const extraBranches = selectionStore.flowBranches?.length || 1
-  // 始终至少有1个主分支
   return extraBranches
 })
 
 const handleClick = () => {
-  // console.log('点击分支按钮')
   emit('click')
-  // 触发打开分支窗口
   agentsStore.openPlanModification()
 }
 </script>
 <template>
   <div
     class="branch-button"
-    :class="{ 'has-branches': branchCount > 0 }"
+    :class="{ 'has-branches': branchCount > 1 }"
     @click="handleClick"
     :title="`${branchCount} 个分支`"
   >

frontend/src/layout/components/Main/TaskTemplate/TaskSyllabus/components/FloatWindow.vue

@@ -39,7 +39,6 @@
 
 <script setup lang="ts">
 import { ref, computed, onMounted, onUnmounted, nextTick } from 'vue'
-import { debounce } from 'lodash-es'
 export interface IFloatingWindowProps {
   title?: string | any
   onClose?: () => void
@@ -262,24 +261,20 @@ const handleResizeMove = (e: MouseEvent) => {
   emit('update:position', position.value)
 }
 
-const handleResizeEnd = debounce(() => {
+const handleResizeEnd = () => {
   isResizing.value = false
   resizeDirection.value = null
   document.removeEventListener('mousemove', handleResizeMove)
   document.removeEventListener('mouseup', handleResizeEnd)
 
   props.onResize?.()
-}, 50)
+}
 
 const setResizeable = (value: boolean) => {
   resizeable.value = value
 }
 
-const handleClose = () => {
+const handleClose = () => props.onClose?.()
-  if (props.onClose) {
-    props.onClose()
-  }
-}
 
 // 初始化
 onMounted(() => {

frontend/src/layout/components/Main/TaskTemplate/TaskSyllabus/index.vue

@@ -3,11 +3,15 @@ import SvgIcon from '@/components/SvgIcon/index.vue'
 import { getAgentMapIcon } from '@/layout/components/config.ts'
 import { type ConnectArg, Jsplumb } from '@/layout/components/Main/TaskTemplate/utils.ts'
 import { type IRawStepTask, useAgentsStore } from '@/stores'
-import { computed, ref, nextTick, watch, onMounted } from 'vue'
+import { computed, nextTick, watch, onMounted } from 'vue'
 import { AnchorLocations } from '@jsplumb/browser-ui'
+import { Loading } from '@element-plus/icons-vue'
 import MultiLineTooltip from '@/components/MultiLineTooltip/index.vue'
 import Bg from './Bg.vue'
 import BranchButton from './components/BranchButton.vue'
+import Notification from '@/components/Notification/Notification.vue'
+import { useNotification } from '@/composables/useNotification'
+import TaskContentEditor from '@/components/TaskContentEditor/index.vue'
 
 // 判断计划是否就绪
 const planReady = computed(() => {
@@ -21,8 +25,6 @@ const openPlanModification = () => {
 const emit = defineEmits<{
   (el: 'resetAgentRepoLine'): void
   (el: 'setCurrentTask', task: IRawStepTask): void
-  (el: 'add-output', outputName: string): void
-  (el: 'click-branch'): void
 }>()
 
 const jsplumb = new Jsplumb('task-syllabus')
@@ -37,119 +39,85 @@ const collaborationProcess = computed(() => {
   return agentsStore.agentRawPlan.data?.['Collaboration Process'] ?? []
 })
 
-// 编辑状态管理
+// 检测是否正在填充详情
-const editingTaskId = ref<string | null>(null)
+const isFillingDetails = computed(() => {
-const editingContent = ref('')
+  const process = agentsStore.agentRawPlan.data?.['Collaboration Process'] || []
-
+  return (
-// 添加新产物状态管理
+    process.length > 0 &&
-const isAddingOutput = ref(false)
+    process.some(step => !step.AgentSelection || step.AgentSelection.length === 0)
-const newOutputInputRef = ref<HTMLElement>()
+  )
-const newOutputName = ref('')
-
-// 处理加号点击
-const handleAddOutputClick = () => {
-  isAddingOutput.value = true
-  newOutputName.value = ''
-
-  nextTick(() => {
-    setTimeout(() => {
-      if (newOutputInputRef.value) {
-        newOutputInputRef.value?.focus()
-      }
-      jsplumb.instance.repaintEverything()
-    }, 50)
 })
-}
 
-// 保存新产物
+// 计算填充进度
-const saveNewOutput = () => {
+const completedSteps = computed(() => {
-  if (newOutputName.value.trim()) {
+  const process = agentsStore.agentRawPlan.data?.['Collaboration Process'] || []
-    const outputName = newOutputName.value.trim()
+  return process.filter(step => step.AgentSelection && step.AgentSelection.length > 0).length
-    const success = agentsStore.addNewOutput(outputName)
-    if (success) {
-      emit('add-output', outputName)
-      isAddingOutput.value = false
-      newOutputName.value = ''
-      nextTick(() => {
-        setTimeout(() => {
-          jsplumb.instance.repaintEverything()
-        }, 50)
 })
-      console.log('添加新产物成功', outputName)
-    } else {
-      // 退出编辑状态
-      isAddingOutput.value = false
-      newOutputName.value = ''
-    }
-  }
-}
 
-// 取消添加产物
+const totalSteps = computed(() => {
-const cancelAddOutput = () => {
+  return agentsStore.agentRawPlan.data?.['Collaboration Process']?.length || 0
-  isAddingOutput.value = false
-  newOutputName.value = ''
-
-  nextTick(() => {
-    setTimeout(() => {
-      jsplumb.instance.repaintEverything()
-    }, 50)
 })
-}
 
-// 处理新产物的键盘事件
+// Notification 通知
-const handleNewOutputKeydown = (event: KeyboardEvent) => {
+const {
-  if (event.key === 'Enter') {
+  notifications,
-    event.preventDefault()
+  progress: showProgress,
-    saveNewOutput()
+  updateProgressDetail,
-  } else if (event.key === 'Escape') {
+  removeNotification
-    cancelAddOutput()
+} = useNotification()
-  }
+const fillingProgressNotificationId = ref<string | null>(null)
-}
 
-// 新产物输入框失去焦点处理
+// 监听填充进度，显示通知
-const handleNewOutputBlur = () => {
+watch(
-  setTimeout(() => {
+  [isFillingDetails, completedSteps, totalSteps, () => agentsStore.hasStoppedFilling],
-    if (newOutputName.value.trim() === '') {
+  ([filling, completed, total, hasStopped]) => {
-      cancelAddOutput()
+    // 如果用户已停止，关闭进度通知
-    }
+    if (hasStopped && fillingProgressNotificationId.value) {
-  }, 100)
+      removeNotification(fillingProgressNotificationId.value)
-}
+      fillingProgressNotificationId.value = null
-
-// 开始编辑
-const startEditing = (task: IRawStepTask) => {
-  if (!task.Id) {
-    console.warn('Task ID is missing, cannot start editing')
       return
     }
-  editingTaskId.value = task.Id
-  editingContent.value = task.TaskContent || ''
-}
 
-// 保存编辑
+    if (filling && total > 0) {
-const saveEditing = () => {
+      if (!fillingProgressNotificationId.value) {
-  if (editingTaskId.value && editingContent.value.trim()) {
+        // 创建进度通知
-    const taskToUpdate = collaborationProcess.value.find(item => item.Id === editingTaskId.value)
+        fillingProgressNotificationId.value = showProgress('生成协作流程', completed, total)
-    if (taskToUpdate) {
+        updateProgressDetail(
-      taskToUpdate.TaskContent = editingContent.value.trim()
+          fillingProgressNotificationId.value,
+          `${completed}/${total}`,
+          '正在分配智能体...',
+          completed,
+          total
+        )
+      } else {
+        // 更新进度通知
+        updateProgressDetail(
+          fillingProgressNotificationId.value,
+          `${completed}/${total}`,
+          '正在分配智能体...',
+          completed,
+          total
+        )
       }
+    } else if (fillingProgressNotificationId.value && !filling) {
+      // 填充完成，关闭进度通知
+      removeNotification(fillingProgressNotificationId.value)
+      fillingProgressNotificationId.value = null
     }
-  editingTaskId.value = null
+  },
-  editingContent.value = ''
+  { immediate: true }
-}
+)
 
-// 取消编辑
+// 保存编辑内容
-const cancelEditing = () => {
+const handleContentSave = (taskId: string, content: string) => {
-  editingTaskId.value = null
+  const taskToUpdate = collaborationProcess.value.find(item => item.Id === taskId)
-  editingContent.value = ''
+  if (taskToUpdate && content !== taskToUpdate.TaskContent) {
+    taskToUpdate.TaskContent = content
+    // 记录修改过的步骤索引到store用于重新执行
+    const stepIndex = collaborationProcess.value.findIndex(item => item.Id === taskId)
+    if (stepIndex >= 0) {
+      agentsStore.addModifiedStep(stepIndex)
     }
-
-// 处理键盘事件
-const handleKeydown = (event: KeyboardEvent) => {
-  if (event.key === 'Enter') {
-    event.preventDefault()
-    saveEditing()
-  } else if (event.key === 'Escape') {
-    cancelEditing()
   }
 }
 
@@ -201,44 +169,32 @@ function clear() {
   jsplumb.reset()
 }
 
-// 🆕 封装连线重绘方法
+//  封装连线重绘方法
 const redrawConnections = () => {
-  console.log('🔄 重新绘制 jsplumb 连线')
-
-  // 等待 DOM 更新完成
   nextTick(() => {
-    // 清除旧连线
     jsplumb.reset()
-
-    // 等待 DOM 稳定后重新绘制
     setTimeout(() => {
       const arr: ConnectArg[] = []
       const currentTaskId = agentsStore.currentTask?.Id
-
-      // 重新绘制所有连线
       collaborationProcess.value.forEach(item => {
         arr.push(...handleCurrentTask(item, item.Id !== currentTaskId))
       })
-
       jsplumb.connects(arr)
-      console.log('✅ jsplumb 连线重绘完成，任务数:', collaborationProcess.value.length)
     }, 100)
   })
 }
 
-// 🆕 监听 collaborationProcess 变化，自动重绘连线
+// 监听 collaborationProcess 变化，自动重绘连线
 watch(
   () => collaborationProcess,
   () => {
-    console.log('🔍 collaborationProcess 发生变化，触发重绘')
     redrawConnections()
   },
   { deep: true }
 )
 
-// 🆕 组件挂载后初始化连线
+// 组件挂载后初始化连线
 onMounted(() => {
-  // 初始化时绘制连线
   nextTick(() => {
     setTimeout(() => {
       const arr: ConnectArg[] = []
@@ -246,7 +202,6 @@ onMounted(() => {
         arr.push(...handleCurrentTask(item, true))
       })
       jsplumb.connects(arr)
-      console.log('✅ 初始化 jsplumb 连线完成')
     }, 100)
   })
 })
@@ -259,6 +214,9 @@ defineExpose({
 
 <template>
   <div class="h-full flex flex-col">
+    <!-- Notification 通知 -->
+    <Notification :notifications="notifications" @close="id => removeNotification(id)" />
+
     <div class="text-[18px] font-bold mb-[18px] text-[var(--color-text-title-header)]">
       任务大纲
     </div>
@@ -273,8 +231,7 @@ defineExpose({
         class="w-full relative min-h-full"
         id="task-syllabus"
       >
-        <Bg :is-adding="isAddingOutput" @start-add-output="handleAddOutputClick" />
+        <Bg />
-
         <div class="w-full flex items-center gap-[14%] mb-[35px]">
           <div class="flex-1 flex justify-center">
             <div
@@ -292,47 +249,6 @@ defineExpose({
             </div>
           </div>
         </div>
-        <!-- 添加新产物卡片 -->
-        <div
-          v-if="isAddingOutput"
-          class="card-it w-full flex items-center gap-[14%] bg-[var(--color-card-bg)] add-output-form mb-[100px]"
-        >
-          <!-- 左侧空白的流程卡片占位 -->
-          <div class="w-[43%] relative z-99" style="height: 20px"></div>
-
-          <!-- 右侧可编辑的产物卡片 -->
-          <el-card
-            class="w-[43%] relative task-syllabus-output-object-card border-dashed border-2 border-[var(--color-primary)]"
-          >
-            <div class="h-full flex items-center justify-center">
-              <!-- 输入框 -->
-              <el-input
-                ref="newOutputInputRef"
-                v-model="newOutputName"
-                placeholder="Enter保存，ESC取消"
-                @keydown="handleNewOutputKeydown"
-                @blur="handleNewOutputBlur"
-                size="large"
-                class="w-full"
-              />
-            </div>
-          </el-card>
-        </div>
-
-        <!-- 显示临时产物卡片 -->
-        <div
-          v-for="output in agentsStore.additionalOutputs"
-          :key="output"
-          class="card-it w-full flex items-center gap-[14%] bg-[var(--color-card-bg)] mb-[100px]"
-        >
-          <!-- 左侧空白的流程卡片占位 -->
-          <div class="w-[43%] relative z-99" style="height: 100px"></div>
-
-          <!-- 右侧产物卡片 -->
-          <el-card class="w-[43%] relative task-syllabus-output-object-card" :shadow="true">
-            <div class="text-[18px] font-bold text-center">{{ output }}</div>
-          </el-card>
-        </div>
         <div
           v-for="item in collaborationProcess"
           :key="item.Id"
@@ -352,54 +268,48 @@ defineExpose({
             <div class="h-[1px] w-full bg-[var(--color-border-separate)] my-[8px]"></div>
 
             <!-- 任务内容区域 - 支持双击编辑 -->
-            <div v-if="editingTaskId === item.Id" class="w-full">
+            <TaskContentEditor :task="item" @save="handleContentSave">
-              <div class="flex flex-col gap-3">
+              <template #display>
-                <el-input
-                  v-model="editingContent"
-                  type="textarea"
-                  :autosize="{ minRows: 2, maxRows: 4 }"
-                  placeholder="请输入任务内容"
-                  @keydown="handleKeydown"
-                  class="task-content-editor"
-                  size="small"
-                />
-                <div class="flex justify-end">
-                  <svg-icon
-                    icon-class="Check"
-                    size="20px"
-                    color="#328621"
-                    class="cursor-pointer mr-4"
-                    @click="saveEditing"
-                    title="保存"
-                  />
-                  <svg-icon
-                    icon-class="Cancel"
-                    size="20px"
-                    color="#8e0707"
-                    class="cursor-pointer mr-1"
-                    @click="cancelEditing"
-                    title="取消"
-                  />
-                </div>
-              </div>
-            </div>
-            <div v-else @dblclick="startEditing(item)" class="w-full cursor-pointer">
                 <MultiLineTooltip placement="right" :text="item.TaskContent" :lines="3">
                   <div class="text-[14px] text-[var(--color-text-secondary)] task-content-display">
                     {{ item.TaskContent }}
                   </div>
                 </MultiLineTooltip>
-            </div>
+              </template>
+            </TaskContentEditor>
 
             <div class="h-[1px] w-full bg-[var(--color-border-separate)] my-[8px]"></div>
+
             <div
-              class="flex items-center gap-2 overflow-y-auto flex-wrap relative w-full max-h-[72px]"
+              class="flex items-center gap-2 flex-wrap relative w-full"
+              :class="
+                !item.AgentSelection || item.AgentSelection.length === 0
+                  ? 'min-h-[40px]'
+                  : 'overflow-y-auto max-h-[72px]'
+              "
             >
               <!--  连接到智能体库的连接点 -->
               <div
                 class="absolute left-[-10px] top-1/2 transform -translate-y-1/2"
                 :id="`task-syllabus-flow-agents-${item.Id}`"
               ></div>
+
+              <!-- 未填充智能体时显示Loading -->
+              <div
+                v-if="
+                  (!item.AgentSelection || item.AgentSelection.length === 0) &&
+                  !agentsStore.hasStoppedFilling
+                "
+                class="flex items-center gap-2 text-[var(--color-text-secondary)] text-[14px]"
+              >
+                <el-icon class="is-loading" :size="20">
+                  <Loading />
+                </el-icon>
+                <span>正在分配智能体<span class="loading-dots"></span></span>
+              </div>
+
+              <!-- 已填充智能体时显示智能体列表 -->
+              <template v-else>
                 <el-tooltip
                   v-for="agentSelection in item.AgentSelection"
                   :key="agentSelection"
@@ -411,7 +321,9 @@ defineExpose({
                       <div class="text-[18px] font-bold">{{ agentSelection }}</div>
                       <div class="h-[1px] w-full bg-[#494B51] my-[8px]"></div>
                       <div>
-                      {{ item.TaskProcess.find(i => i.AgentName === agentSelection)?.Description }}
+                        {{
+                          item.TaskProcess.find(i => i.AgentName === agentSelection)?.Description
+                        }}
                       </div>
                     </div>
                   </template>
@@ -426,6 +338,7 @@ defineExpose({
                     />
                   </div>
                 </el-tooltip>
+              </template>
             </div>
           </el-card>
           <!-- 产物卡片 -->
@@ -440,7 +353,7 @@ defineExpose({
         </div>
       </div>
     </div>
-    <BranchButton v-dev-only v-if="planReady" @click="openPlanModification" />
+    <BranchButton v-if="planReady" @click="openPlanModification" />
   </div>
 </template>
 <style lang="scss" scoped>
@@ -501,66 +414,6 @@ defineExpose({
   white-space: pre-wrap;
 }
 
-.add-output-btn {
-  opacity: 0.8;
-  transition: opacity 0.2s ease;
-
-  &:hover {
-    opacity: 1;
-  }
-
-  button {
-    background: transparent;
-    cursor: pointer;
-
-    &:hover {
-      background-color: rgba(59, 130, 246, 0.05);
-      transform: translateY(-1px);
-      box-shadow: 0 2px 8px rgba(0, 0, 0, 0.1);
-    }
-
-    &:active {
-      transform: translateY(0);
-    }
-  }
-}
-
-.add-output-form {
-  animation: slideDown 0.3s ease-out;
-  :deep(.el-card__body) {
-    padding: 16px;
-    display: flex;
-    align-items: center;
-    justify-content: center;
-  }
-  :deep(.el-input__wrapper) {
-    border: 1px solid var(--color-text);
-    background: transparent;
-    box-shadow: none;
-
-    &.is-focus {
-      border-color: var(--color-text);
-      box-shadow: 0 0 0 1px var(--color-primary-light);
-    }
-    :deep(.el-input__inner) {
-      font-size: 14px;
-      text-align: center;
-      font-weight: bold;
-    }
-  }
-}
-
-@keyframes slideDown {
-  from {
-    opacity: 0;
-    transform: translateY(-10px);
-  }
-  to {
-    opacity: 1;
-    transform: translateY(0);
-  }
-}
-
 // 输入框样式
 :deep(.el-input__wrapper) {
   background: transparent;
@@ -624,4 +477,33 @@ defineExpose({
     }
   }
 }
+
+// 加载动画省略号
+.loading-dots {
+  display: inline-block;
+  width: 1.2em;
+  text-align: left;
+
+  &::after {
+    content: '';
+    animation: dots 1.5s steps(4, end) infinite;
+  }
+}
+
+@keyframes dots {
+  0%,
+  20% {
+    content: '';
+  }
+  40% {
+    content: '.';
+  }
+  60% {
+    content: '..';
+  }
+  80%,
+  100% {
+    content: '...';
+  }
+}
 </style>

frontend/src/layout/components/Main/TaskTemplate/index.vue

@@ -47,10 +47,10 @@ function handleTaskSyllabusCurrentTask(task: IRawStepTask) {
   agentsStore.setCurrentTask(task)
 }
 
+// 更新任务大纲内部的线
 function handleTaskResultCurrentTask(task: IRawStepTask) {
   scrollToElementTop(`task-syllabus-flow-${task.Id}`)
   agentsStore.setCurrentTask(task)
-  // 更新任务大纲内部的线
   taskSyllabusRef.value?.changeTask(task, false)
 }
 
@@ -71,11 +71,6 @@ function clear() {
   taskResultJsplumb.repaintEverything()
 }
 
-const additionalOutputs = ref<string[]>([])
-const handleAddOutput = (outputName: string) => {
-  additionalOutputs.value.unshift(outputName)
-}
-
 defineExpose({
   changeTask,
   resetAgentRepoLine,
@@ -99,7 +94,6 @@ defineExpose({
         ref="taskSyllabusRef"
         @resetAgentRepoLine="resetAgentRepoLine"
         @set-current-task="handleTaskSyllabusCurrentTask"
-        @add-output="handleAddOutput"
       />
     </div>
     <!--  执行结果 -->
@@ -108,7 +102,6 @@ defineExpose({
         ref="taskResultRef"
         @refresh-line="taskResultJsplumb.repaintEverything"
         @set-current-task="handleTaskResultCurrentTask"
-        :additional-outputs="additionalOutputs"
       />
     </div>
   </div>

frontend/src/main.ts

@@ -7,6 +7,7 @@ import './styles/tailwindcss.css'
 import 'element-plus/theme-chalk/dark/css-vars.css'
 import 'virtual:svg-icons-register'
 import { initService } from '@/utils/request.ts'
+import websocket from '@/utils/websocket'
 import { setupStore, useConfigStore } from '@/stores'
 import { setupDirective } from '@/ directive'
 import * as ElementPlusIconsVue from '@element-plus/icons-vue'
@@ -20,6 +21,24 @@ async function init() {
   setupStore(app)
   setupDirective(app)
   initService()
+
+  // 初始化WebSocket连接
+  try {
+    // WebSocket需要直接连接到后端，不能通过代理
+    const apiBaseUrl = configStore.config.apiBaseUrl || `${import.meta.env.BASE_URL || '/'}api`
+    // 移除 /api 后缀，如果是相对路径则构造完整URL
+    let wsUrl = apiBaseUrl.replace(/\/api$/, '')
+    // 如果是相对路径（以/开头），使用当前host
+    if (wsUrl.startsWith('/')) {
+      wsUrl = `${window.location.protocol}//${window.location.host}${wsUrl}`
+    }
+    console.log('🔌 Connecting to WebSocket at:', wsUrl)
+    await websocket.connect(wsUrl)
+    console.log('✅ WebSocket initialized successfully')
+  } catch (error) {
+    console.warn('⚠️  WebSocket connection failed, will use REST API fallback:', error)
+  }
+
   document.title = configStore.config.centerTitle
   app.use(router)
   app.mount('#app')

frontend/src/stores/modules/agents.ts

@@ -73,6 +73,13 @@ export interface IScoreItem {
   reason: string
 }
 
+// 单个任务的评分数据结构
+export interface ITaskScoreData {
+  aspectList: string[] // 维度列表
+  agentScores: Record<string, Record<string, IScoreItem>> // agent -> 维度 -> 评分
+  timestamp: number // 缓存时间戳
+}
+
 export interface IAgentSelectModifyAddRequest {
   aspectList: string[] // 维度列表
   agentScores: Record<string, Record<string, IScoreItem>> // agent -> 维度 -> 评分（与后端返回格式一致）
@@ -89,6 +96,23 @@ export interface IRawPlanResponse {
   'Collaboration Process'?: IRawStepTask[]
 }
 
+/**
+ * IApiAgentAction[] → TaskProcess[] 格式转换
+ * @param actions IApiAgentAction 数组
+ * @returns TaskProcess 数组
+ */
+export function convertToTaskProcess(
+  actions: { id: string; type: string; agent: string; description: string; inputs?: string[] }[]
+): TaskProcess[] {
+  return actions.map(action => ({
+    ID: action.id,
+    ActionType: action.type,
+    AgentName: action.agent,
+    Description: action.description,
+    ImportantInput: action.inputs || []
+  }))
+}
+
 const storageKey = '$agents' as const
 
 // 清除所有以 storageKey 开头的 localStorage
@@ -105,13 +129,104 @@ export const useAgentsStore = defineStore('agents', () => {
     agents.value = agent
   }
 
-  // 智能体评分数据存储
+  // 🆕 新的按任务ID存储的评分数据
+  const taskScoreDataMap = useStorage<Record<string, ITaskScoreData>>(
+    `${storageKey}-task-score-data`,
+    {},
+  )
+
+  // 🆕 预加载状态追踪（用于避免重复预加载）
+  const preloadingTaskIds = ref<Set<string>>(new Set())
+
+  // 🆕 获取指定任务的评分数据（按任务ID获取）
+  function getTaskScoreData(taskId: string): IAgentSelectModifyAddRequest | null {
+    if (!taskId) {
+      console.warn('⚠️ getTaskScoreData: taskId 为空')
+      return null
+    }
+
+    const taskScoreData = taskScoreDataMap.value[taskId]
+    if (taskScoreData) {
+      console.log(`✅ 使用任务 ${taskId} 的缓存评分数据，维度数: ${taskScoreData.aspectList.length}`)
+      return {
+        aspectList: taskScoreData.aspectList,
+        agentScores: taskScoreData.agentScores,
+      }
+    }
+
+    console.log(`ℹ️ 任务 ${taskId} 没有缓存的评分数据`)
+    return null
+  }
+
+  // 🆕 设置指定任务的评分数据
+  function setTaskScoreData(taskId: string, data: IAgentSelectModifyAddRequest) {
+    if (!taskId) {
+      console.warn('⚠️ setTaskScoreData: taskId 为空，无法存储')
+      return
+    }
+    taskScoreDataMap.value = {
+      ...taskScoreDataMap.value,
+      [taskId]: {
+        ...data,
+        timestamp: Date.now(),
+      },
+    }
+    console.log(`✅ 任务 ${taskId} 的评分数据已存储`, {
+      aspectCount: data.aspectList.length,
+      agentCount: Object.keys(data.agentScores).length,
+    })
+  }
+
+  // 🆕 检查指定任务是否有评分数据
+  function hasTaskScoreData(taskId: string): boolean {
+    if (!taskId) return false
+    return !!taskScoreDataMap.value[taskId]?.aspectList?.length
+  }
+
+  // 🆕 获取所有已预加载的任务ID列表
+  function getPreloadedTaskIds(): string[] {
+    return Object.keys(taskScoreDataMap.value)
+  }
+
+  // 🆕 清除指定任务的评分数据
+  function clearTaskScoreData(taskId: string) {
+    if (taskId && taskScoreDataMap.value[taskId]) {
+      const newMap = { ...taskScoreDataMap.value }
+      delete newMap[taskId]
+      taskScoreDataMap.value = newMap
+      console.log(`✅ 任务 ${taskId} 的评分数据已清除`)
+    }
+  }
+
+  // 🆕 清除所有任务的评分数据
+  function clearAllTaskScoreData() {
+    taskScoreDataMap.value = {}
+    preloadingTaskIds.value.clear()
+    console.log('✅ 所有任务的评分数据已清除')
+  }
+
+  // 🆕 标记任务正在预加载中
+  function markTaskPreloading(taskId: string) {
+    preloadingTaskIds.value.add(taskId)
+  }
+
+  // 🆕 标记任务预加载完成
+  function markTaskPreloadComplete(taskId: string) {
+    preloadingTaskIds.value.delete(taskId)
+  }
+
+  // 🆕 检查任务是否正在预加载
+  function isTaskPreloading(taskId: string): boolean {
+    return preloadingTaskIds.value.has(taskId)
+  }
+
+  // 兼容旧版本：全局评分数据存储（单任务模式，已废弃，保留用于兼容）
   const IAgentSelectModifyAddRequest = useStorage<IAgentSelectModifyAddRequest | null>(
     `${storageKey}-score-data`,
     null,
   )
 
-  // 设置智能体评分数据
+  // 设置智能体评分数据（兼容旧版本）
   function setAgentScoreData(data: IAgentSelectModifyAddRequest) {
     IAgentSelectModifyAddRequest.value = data
   }
@@ -131,8 +246,6 @@ export const useAgentsStore = defineStore('agents', () => {
     }
 
     // 添加该维度的评分数据
-    // scores: { agentName: { score, reason } }
-    // 转换为 agentScores[agentName][aspectName] = { score, reason }
     for (const [agentName, scoreItem] of Object.entries(scores)) {
       if (!IAgentSelectModifyAddRequest.value.agentScores[agentName]) {
         IAgentSelectModifyAddRequest.value.agentScores[agentName] = {}
@@ -141,7 +254,7 @@ export const useAgentsStore = defineStore('agents', () => {
     }
   }
 
-  // 清除智能体评分数据
+  // 清除智能体评分数据（兼容旧版本）
   function clearAgentScoreData() {
     IAgentSelectModifyAddRequest.value = null
   }
@@ -221,6 +334,19 @@ export const useAgentsStore = defineStore('agents', () => {
 
   // 当前的展示的任务流程
   const currentTask = ref<IRawStepTask>()
+
+  // 记录用户修改过的步骤索引（用于重新执行）
+  const modifiedSteps = ref<Set<number>>(new Set())
+  function addModifiedStep(stepIndex: number) {
+    modifiedSteps.value.add(stepIndex)
+  }
+  function clearModifiedSteps() {
+    modifiedSteps.value.clear()
+  }
+  function hasModifiedSteps() {
+    return modifiedSteps.value.size > 0
+  }
+
   function setCurrentTask(task: IRawStepTask) {
     const existingTask = currentTask.value
 
@@ -359,11 +485,9 @@ export const useAgentsStore = defineStore('agents', () => {
     }
     currentTask.value = undefined
     executePlan.value = []
+    hasStoppedFilling.value = false
   }
 
-  // 额外的产物列表
-  const additionalOutputs = ref<string[]>([])
-
   // 用户在 Assignment 中选择的 agent 组合（按任务ID分别存储）
   const selectedAgentGroupMap = ref<Map<string, string[]>>(new Map())
 
@@ -388,31 +512,40 @@ export const useAgentsStore = defineStore('agents', () => {
     selectedAgentGroupMap.value.clear()
   }
 
-  // 添加新产物
+  // 标记是否用户已停止智能体分配过程
-  function addNewOutput(outputObject: string) {
+  const hasStoppedFilling = ref(false)
-    if (!outputObject.trim()) return false
 
-    const trimmedOutput = outputObject.trim()
+  // 标记是否正在停止中（全局停止状态）
-    if (!additionalOutputs.value.includes(trimmedOutput)) {
+  const isStopping = ref(false)
-      additionalOutputs.value.unshift(trimmedOutput)
+
-      return true
+  // 设置停止状态
-    }
+  function setHasStoppedFilling(value: boolean) {
-    return false
+    hasStoppedFilling.value = value
   }
 
-  // 删除额外产物
+  // 设置正在停止状态
-  function removeAdditionalOutput(outputObject: string) {
+  function setIsStopping(value: boolean) {
-    const index = additionalOutputs.value.indexOf(outputObject)
+    isStopping.value = value
-    if (index > -1) {
-      additionalOutputs.value.splice(index, 1)
-      return true
-    }
-    return false
   }
 
-  // 清除额外产物
+  /**
-  function clearAdditionalOutputs() {
+   * 构建 IApiStepTask 对象（用于 API 调用）
-    additionalOutputs.value = []
+   * @param agents 选中的 agent 列表
+   * @returns IApiStepTask 格式的对象
+   */
+  function createStepTaskForApi(agents: string[]): IApiStepTask {
+    return {
+      name: currentTask.value?.StepName || '',
+      content: currentTask.value?.TaskContent || '',
+      inputs: currentTask.value?.InputObject_List || [],
+      output: currentTask.value?.OutputObject || '',
+      agents,
+      brief: currentTask.value?.Collaboration_Brief_frontEnd || {
+        template: '',
+        data: {}
+      },
+      process: []
+    }
   }
 
   return {
@@ -425,15 +558,12 @@ export const useAgentsStore = defineStore('agents', () => {
     setCurrentTaskProcess, // 🆕 设置当前任务的 TaskProcess
     updateCurrentAgentSelection, // 🆕 强制更新 AgentSelection（用于 AgentAllocation 切换组合）
     syncCurrentTaskToMainProcess, // 🆕 同步 currentTask 到主流程
+    createStepTaskForApi, // 🆕 构建 IApiStepTask 对象
     agentRawPlan,
     setAgentRawPlan,
     executePlan,
     setExecutePlan,
     resetAgent,
-    additionalOutputs,
-    addNewOutput,
-    removeAdditionalOutput,
-    clearAdditionalOutputs,
     // 用户选择的 agent 组合
     selectedAgentGroupMap,
     setSelectedAgentGroup,
@@ -454,12 +584,35 @@ export const useAgentsStore = defineStore('agents', () => {
     setAgentScoreData,
     addAgentScoreAspect,
     clearAgentScoreData,
+    // 🆕 按任务ID存储的评分数据
+    taskScoreDataMap,
+    getTaskScoreData,
+    setTaskScoreData,
+    hasTaskScoreData,
+    getPreloadedTaskIds,
+    clearTaskScoreData,
+    clearAllTaskScoreData,
+    preloadingTaskIds,
+    markTaskPreloading,
+    markTaskPreloadComplete,
+    isTaskPreloading,
     // 确认的agent组合列表（按任务ID分别存储）
     confirmedAgentGroupsMap,
     getConfirmedAgentGroups,
     addConfirmedAgentGroup,
     clearConfirmedAgentGroups,
     clearAllConfirmedAgentGroups,
+    // 停止填充状态
+    hasStoppedFilling,
+    setHasStoppedFilling,
+    // 正在停止状态
+    isStopping,
+    setIsStopping,
+    // 重新执行相关
+    modifiedSteps,
+    addModifiedStep,
+    clearModifiedSteps,
+    hasModifiedSteps,
   }
 })
 

frontend/src/stores/modules/selection.ts

@@ -5,25 +5,40 @@ import { store } from '../index'
 import type { IRawStepTask, IApiStepTask } from './agents'
 import type { Node, Edge } from '@vue-flow/core'
 
-// 分支节点数据接口 - 用于存储流程图中的节点和边
+/**
+ * 分支数据接口
+ * @description 用于存储流程图中的分支节点和边数据
+ */
 export interface IBranchData {
-  id: string // 分支唯一ID
+  /** 分支唯一 ID */
-  parentNodeId: string // 父节点ID（根节点或任务节点）
+  id: string
-  branchContent: string // 分支需求内容
+  /** 父节点 ID（根节点或任务节点） */
-  branchType: 'root' | 'task' // 分支类型
+  parentNodeId: string
-  nodes: Node[] // 分支包含的所有节点
+  /** 分支需求内容 */
-  edges: Edge[] // 分支包含的所有边
+  branchContent: string
-  tasks: IRawStepTask[] // 分支的任务数据
+  /** 分支类型 */
-  createdAt: number // 创建时间
+  branchType: 'root' | 'task'
+  /** 分支包含的所有节点 */
+  nodes: Node[]
+  /** 分支包含的所有边 */
+  edges: Edge[]
+  /** 分支的任务数据 */
+  tasks: IRawStepTask[]
+  /** 创建时间 */
+  createdAt: number
 }
 
 export const useSelectionStore = defineStore('selection', () => {
   // ==================== 任务大纲探索分支数据存储 ====================
+
+  /** 流程图分支列表 */
   const flowBranches = ref<IBranchData[]>([])
 
-  // 任务大纲分支管理
+  /**
-
+   * 添加流程图分支
-  // 添加流程图分支
+   * @param data 分支数据
+   * @returns 分支 ID
+   */
   function addFlowBranch(data: {
     parentNodeId: string
     branchContent: string
@@ -44,51 +59,73 @@ export const useSelectionStore = defineStore('selection', () => {
       createdAt: Date.now(),
     }
     flowBranches.value.push(newBranch)
-    console.log('📂 保存流程图分支到 store:', newBranch)
     return branchId
   }
 
-  // 获取所有流程图分支
+  /**
+   * 获取所有流程图分支
+   * @returns 所有流程图分支数据
+   */
   function getAllFlowBranches(): IBranchData[] {
     return flowBranches.value
   }
 
-  // 根据父节点ID获取流程图分支
+  /**
+   * 根据父节点 ID 获取流程图分支
+   * @param parentNodeId 父节点 ID
+   * @returns 匹配的流程图分支列表
+   */
   function getFlowBranchesByParent(parentNodeId: string): IBranchData[] {
     return flowBranches.value.filter((branch) => branch.parentNodeId === parentNodeId)
   }
 
-  // 删除流程图分支
+  /**
+   * 删除流程图分支
+   * @param branchId 分支 ID
+   * @returns 是否删除成功
+   */
   function removeFlowBranch(branchId: string): boolean {
     const index = flowBranches.value.findIndex((branch) => branch.id === branchId)
     if (index > -1) {
       flowBranches.value.splice(index, 1)
-      console.log('🗑️ 删除流程图分支:', branchId)
       return true
     }
     return false
   }
 
-  // 清除所有流程图分支
+  /** 清除所有流程图分支 */
   function clearFlowBranches() {
     flowBranches.value = []
-    console.log('🗑️ 清除所有流程图分支')
   }
 
-  // 根据父节点ID清除流程图分支
+  /**
+   * 根据父节点 ID 清除流程图分支
+   * @param parentNodeId 父节点 ID
+   */
   function clearFlowBranchesByParent(parentNodeId: string) {
     flowBranches.value = flowBranches.value.filter((branch) => branch.parentNodeId !== parentNodeId)
-    console.log('🗑️ 清除父节点为', parentNodeId, '的流程图分支')
   }
 
   // ==================== 任务过程探索分支数据存储 ====================
-  // 用于存储任务过程探索中的分支数据
+  /**
-  // 结构: Map<taskStepId, IBranchData[]>
+   * 任务过程分支数据映射
-  const taskProcessBranchesMap = ref<Map<string, IBranchData[]>>(new Map())
+   * @description 存储结构: Map<taskStepId, Map<agentGroupKey, IBranchData[]>>
+   * - taskStepId: 任务步骤 ID
+   * - agentGroupKey: agent 组合的唯一标识（排序后的 JSON 字符串）
+   * - IBranchData[]: 该 agent 组合在该任务下的所有分支数据
+   */
+  const taskProcessBranchesMap = ref<Map<string, Map<string, IBranchData[]>>>(new Map())
 
-  // 添加任务过程分支
+  /**
+   * 添加任务过程分支
+   * @param taskStepId 任务步骤 ID
+   * @param agents Agent 列表
+   * @param data 分支数据
+   * @returns 分支 ID
+   */
   function addTaskProcessBranch(
     taskStepId: string,
+    agents: string[],
     data: {
       parentNodeId: string
       branchContent: string
@@ -99,6 +136,8 @@ export const useSelectionStore = defineStore('selection', () => {
     },
   ): string {
     const branchId = `task-process-branch-${uuidv4()}`
+    const agentGroupKey = getAgentGroupKey(agents)
+
     const newBranch: IBranchData = {
       id: branchId,
       parentNodeId: data.parentNodeId,
@@ -110,66 +149,131 @@ export const useSelectionStore = defineStore('selection', () => {
       createdAt: Date.now(),
     }
 
-    // 获取或创建该任务步骤的分支列表
+    // 获取或创建该任务步骤的 Map
     if (!taskProcessBranchesMap.value.has(taskStepId)) {
-      taskProcessBranchesMap.value.set(taskStepId, [])
+      taskProcessBranchesMap.value.set(taskStepId, new Map())
     }
-    taskProcessBranchesMap.value.get(taskStepId)!.push(newBranch)
 
-    console.log('📂 保存任务过程分支到 store:', { taskStepId, branch: newBranch })
+    // 获取或创建该 agent 组合的分支列表
+    const agentMap = taskProcessBranchesMap.value.get(taskStepId)!
+    if (!agentMap.has(agentGroupKey)) {
+      agentMap.set(agentGroupKey, [])
+    }
+    agentMap.get(agentGroupKey)!.push(newBranch)
+
     return branchId
   }
 
-  // 获取指定任务步骤的所有分支
+  /**
-  function getTaskProcessBranches(taskStepId: string): IBranchData[] {
+   * 获取指定任务步骤和 agent 组合的所有分支
-    return taskProcessBranchesMap.value.get(taskStepId) || []
+   * @param taskStepId 任务步骤 ID
+   * @param agents Agent 列表
+   * @returns 匹配的任务过程分支列表
+   */
+  function getTaskProcessBranches(taskStepId: string, agents: string[]): IBranchData[] {
+    const agentGroupKey = getAgentGroupKey(agents)
+    return taskProcessBranchesMap.value.get(taskStepId)?.get(agentGroupKey) || []
   }
 
-  // 获取所有任务过程分支
+  /**
-  function getAllTaskProcessBranches(): Map<string, IBranchData[]> {
+   * 获取所有任务过程分支
+   * @returns 所有任务过程分支数据映射
+   */
+  function getAllTaskProcessBranches(): Map<string, Map<string, IBranchData[]>> {
     return taskProcessBranchesMap.value
   }
 
-  // 根据父节点ID获取任务过程分支
+  /**
-  function getTaskProcessBranchesByParent(taskStepId: string, parentNodeId: string): IBranchData[] {
+   * 根据父节点 ID 获取任务过程分支
-    const branches = taskProcessBranchesMap.value.get(taskStepId) || []
+   * @param taskStepId 任务步骤 ID
+   * @param agents Agent 列表
+   * @param parentNodeId 父节点 ID
+   * @returns 匹配的任务过程分支列表
+   */
+  function getTaskProcessBranchesByParent(
+    taskStepId: string,
+    agents: string[],
+    parentNodeId: string,
+  ): IBranchData[] {
+    const agentGroupKey = getAgentGroupKey(agents)
+    const branches = taskProcessBranchesMap.value.get(taskStepId)?.get(agentGroupKey) || []
     return branches.filter((branch) => branch.parentNodeId === parentNodeId)
   }
 
-  // 删除任务过程分支
+  /**
-  function removeTaskProcessBranch(taskStepId: string, branchId: string): boolean {
+   * 删除任务过程分支
-    const branches = taskProcessBranchesMap.value.get(taskStepId)
+   * @param taskStepId 任务步骤 ID
+   * @param agents Agent 列表
+   * @param branchId 分支 ID
+   * @returns 是否删除成功
+   */
+  function removeTaskProcessBranch(
+    taskStepId: string,
+    agents: string[],
+    branchId: string,
+  ): boolean {
+    const agentGroupKey = getAgentGroupKey(agents)
+    const branches = taskProcessBranchesMap.value.get(taskStepId)?.get(agentGroupKey)
     if (branches) {
       const index = branches.findIndex((branch) => branch.id === branchId)
       if (index > -1) {
         branches.splice(index, 1)
-        console.log('🗑️ 删除任务过程分支:', { taskStepId, branchId })
         return true
       }
     }
     return false
   }
 
-  // 清除指定任务步骤的所有分支
+  /**
-  function clearTaskProcessBranches(taskStepId: string) {
+   * 清除指定任务步骤和 agent 组合的所有分支
+   * @param taskStepId 任务步骤 ID
+   * @param agents Agent 列表（可选，不传则清除该任务步骤的所有分支）
+   */
+  function clearTaskProcessBranches(taskStepId: string, agents?: string[]) {
+    if (agents) {
+      // 清除指定 agent 组合的分支
+      const agentGroupKey = getAgentGroupKey(agents)
+      const agentMap = taskProcessBranchesMap.value.get(taskStepId)
+      if (agentMap) {
+        agentMap.delete(agentGroupKey)
+      }
+    } else {
+      // 清除该任务步骤的所有分支（所有 agent 组合）
       taskProcessBranchesMap.value.delete(taskStepId)
-    console.log('🗑️ 清除任务步骤的所有分支:', taskStepId)
+    }
   }
 
   // ==================== Agent 组合 TaskProcess 数据存储 ====================
-  // 用于存储 fill_stepTask_TaskProcess 接口返回的数据
+  /**
-  // 结构: Map<taskId, Map<agentGroupKey, IApiStepTask>>
+   * Agent 组合 TaskProcess 数据映射
-  // - taskId: 任务ID
+   * @description 用于存储 fill_stepTask_TaskProcess 接口返回的数据
-  // - agentGroupKey: agent 组合的唯一标识（排序后的 JSON 字符串）
+   * 存储结构: Map<taskId, Map<agentGroupKey, IApiStepTask>>
-  // - IApiStepTask: 该 agent 组合的完整 TaskProcess 数据
+   * - taskId: 任务 ID
+   * - agentGroupKey: agent 组合的唯一标识（排序后的 JSON 字符串）
+   * - IApiStepTask: 该 agent 组合的完整 TaskProcess 数据
+   */
   const agentTaskProcessMap = ref<Map<string, Map<string, IApiStepTask>>>(new Map())
 
-  // 生成 agent 组合的唯一 key（排序后保证一致性）
+  /**
+   * 生成 agent 组合的唯一 key
+   * @description 排序后保证一致性
+   * @param agents Agent 列表
+   * @returns Agent 组合的唯一标识
+   */
   function getAgentGroupKey(agents: string[]): string {
-    return JSON.stringify(agents.sort())
+    // 处理 undefined 或 null 的情况
+    if (!agents || !Array.isArray(agents)) {
+      return JSON.stringify([])
+    }
+    return JSON.stringify([...agents].sort())
   }
 
-  // 存储 agent 组合的 TaskProcess 数据
+  /**
+   * 存储 agent 组合的 TaskProcess 数据
+   * @param taskId 任务 ID
+   * @param agents Agent 列表
+   * @param taskProcess TaskProcess 数据
+   */
   function setAgentTaskProcess(taskId: string, agents: string[], taskProcess: IApiStepTask) {
     const groupKey = getAgentGroupKey(agents)
 
@@ -180,79 +284,144 @@ export const useSelectionStore = defineStore('selection', () => {
 
     // 存储该 agent 组合的 TaskProcess 数据
     agentTaskProcessMap.value.get(taskId)!.set(groupKey, taskProcess)
-    console.log(`📦 存储 agent 组合 TaskProcess [任务: ${taskId}, agents: ${agents.join(', ')}]`)
   }
 
-  // 获取 agent 组合的 TaskProcess 数据
+  /**
+   * 获取 agent 组合的 TaskProcess 数据
+   * @param taskId 任务 ID
+   * @param agents Agent 列表
+   * @returns TaskProcess 数据
+   */
   function getAgentTaskProcess(taskId: string, agents: string[]): IApiStepTask | undefined {
     const groupKey = getAgentGroupKey(agents)
     return agentTaskProcessMap.value.get(taskId)?.get(groupKey)
   }
 
-  // 检查 agent 组合是否已有 TaskProcess 数据
+  /**
+   * 检查 agent 组合是否已有 TaskProcess 数据
+   * @param taskId 任务 ID
+   * @param agents Agent 列表
+   * @returns 是否存在数据
+   */
   function hasAgentTaskProcess(taskId: string, agents: string[]): boolean {
     const groupKey = getAgentGroupKey(agents)
     return agentTaskProcessMap.value.get(taskId)?.has(groupKey) || false
   }
 
-  // 清除指定任务的所有 agent 组合 TaskProcess 数据
+  /**
+   * 清除指定任务的所有 agent 组合 TaskProcess 数据
+   * @param taskId 任务 ID
+   */
   function clearAgentTaskProcess(taskId: string) {
     agentTaskProcessMap.value.delete(taskId)
-    console.log(`🗑️ 清除任务的 agent 组合 TaskProcess 数据: ${taskId}`)
   }
 
-  // 清除所有任务的 agent 组合 TaskProcess 数据
+  /** 清除所有任务的 agent 组合 TaskProcess 数据 */
   function clearAllAgentTaskProcess() {
     agentTaskProcessMap.value.clear()
-    console.log('🗑️ 清除所有任务的 agent 组合 TaskProcess 数据')
   }
 
   // ==================== 当前生效的任务过程分支 ====================
-  // 记录每个任务步骤当前生效的分支 ID（持久化选中状态）
+  /**
-  // 结构: Map<taskStepId, branchId>
+   * 当前生效的任务过程分支映射
-  const activeTaskProcessBranchMap = ref<Map<string, string>>(new Map())
+   * @description 记录每个任务步骤和 agent 组合当前生效的分支 ID（持久化选中状态）
+   * 存储结构: Map<taskStepId, Map<agentGroupKey, branchId>>
+   * - taskStepId: 任务步骤 ID
+   * - agentGroupKey: agent 组合的唯一标识
+   * - branchId: 当前选中的分支 ID
+   */
+  const activeTaskProcessBranchMap = ref<Map<string, Map<string, string>>>(new Map())
 
-  // 🆕 当前生效的 TaskProcess 数据（用于外部组件显示职责分配）
+  /**
-  // 结构: Map<taskStepId, TaskProcess[]>
+   * 当前生效的 TaskProcess 数据映射
-  const activeTaskProcessDataMap = ref<Map<string, any[]>>(new Map())
+   * @description 用于外部组件显示职责分配
+   * 存储结构: Map<taskStepId, Map<agentGroupKey, TaskProcess[]>>
+   */
+  const activeTaskProcessDataMap = ref<Map<string, Map<string, any[]>>>(new Map())
 
-  // 设置当前生效的分支
+  /**
-  function setActiveTaskProcessBranch(taskStepId: string, branchId: string) {
+   * 设置当前生效的分支
-    activeTaskProcessBranchMap.value.set(taskStepId, branchId)
+   * @param taskStepId 任务步骤 ID
-    console.log('✅ 设置当前生效分支:', { taskStepId, branchId })
+   * @param agents Agent 列表
+   * @param branchId 分支 ID
+   */
+  function setActiveTaskProcessBranch(taskStepId: string, agents: string[], branchId: string) {
+    const agentGroupKey = getAgentGroupKey(agents)
+
+    // 获取或创建该任务步骤的 Map
+    if (!activeTaskProcessBranchMap.value.has(taskStepId)) {
+      activeTaskProcessBranchMap.value.set(taskStepId, new Map())
     }
 
-  // 🆕 设置当前生效的 TaskProcess 数据
+    activeTaskProcessBranchMap.value.get(taskStepId)!.set(agentGroupKey, branchId)
-  function setActiveTaskProcessData(taskStepId: string, taskProcess: any[]) {
-    activeTaskProcessDataMap.value.set(taskStepId, taskProcess)
-    console.log('✅ 设置当前生效的 TaskProcess 数据:', { taskStepId, taskProcess })
   }
 
-  // 获取当前生效的分支 ID
+  /**
-  function getActiveTaskProcessBranch(taskStepId: string): string | undefined {
+   * 设置当前生效的 TaskProcess 数据
-    return activeTaskProcessBranchMap.value.get(taskStepId)
+   * @param taskStepId 任务步骤 ID
+   * @param agents Agent 列表
+   * @param taskProcess TaskProcess 数据
+   */
+  function setActiveTaskProcessData(taskStepId: string, agents: string[], taskProcess: any[]) {
+    const agentGroupKey = getAgentGroupKey(agents)
+
+    // 获取或创建该任务步骤的 Map
+    if (!activeTaskProcessDataMap.value.has(taskStepId)) {
+      activeTaskProcessDataMap.value.set(taskStepId, new Map())
     }
 
-  // 🆕 获取当前生效的 TaskProcess 数据
+    activeTaskProcessDataMap.value.get(taskStepId)!.set(agentGroupKey, taskProcess)
-  function getActiveTaskProcessData(taskStepId: string): any[] | undefined {
-    return activeTaskProcessDataMap.value.get(taskStepId)
   }
 
-  // 清除生效分支
+  /**
-  function clearActiveTaskProcessBranch(taskStepId: string) {
+   * 获取当前生效的分支 ID
+   * @param taskStepId 任务步骤 ID
+   * @param agents Agent 列表
+   * @returns 分支 ID
+   */
+  function getActiveTaskProcessBranch(taskStepId: string, agents: string[]): string | undefined {
+    const agentGroupKey = getAgentGroupKey(agents)
+    return activeTaskProcessBranchMap.value.get(taskStepId)?.get(agentGroupKey)
+  }
+
+  /**
+   * 获取当前生效的 TaskProcess 数据
+   * @param taskStepId 任务步骤 ID
+   * @param agents Agent 列表
+   * @returns TaskProcess 数据
+   */
+  function getActiveTaskProcessData(taskStepId: string, agents: string[]): any[] | undefined {
+    const agentGroupKey = getAgentGroupKey(agents)
+    return activeTaskProcessDataMap.value.get(taskStepId)?.get(agentGroupKey)
+  }
+
+  /**
+   * 清除生效分支
+   * @param taskStepId 任务步骤 ID
+   * @param agents Agent 列表（可选，不传则清除该任务步骤的所有生效分支）
+   */
+  function clearActiveTaskProcessBranch(taskStepId: string, agents?: string[]) {
+    if (agents) {
+      // 清除指定 agent 组合的生效分支
+      const agentGroupKey = getAgentGroupKey(agents)
+      activeTaskProcessBranchMap.value.get(taskStepId)?.delete(agentGroupKey)
+      activeTaskProcessDataMap.value.get(taskStepId)?.delete(agentGroupKey)
+    } else {
+      // 清除该任务步骤的所有生效分支（所有 agent 组合）
       activeTaskProcessBranchMap.value.delete(taskStepId)
       activeTaskProcessDataMap.value.delete(taskStepId)
     }
+  }
 
   return {
-    // 状态
+    // ==================== 状态 ====================
     flowBranches,
     taskProcessBranchesMap,
     agentTaskProcessMap,
     activeTaskProcessBranchMap,
-    activeTaskProcessDataMap, // 🆕 新增
+    activeTaskProcessDataMap,
 
-    // 任务大纲分支管理方法
+    // ==================== 任务大纲分支管理方法 ====================
     addFlowBranch,
     getAllFlowBranches,
     getFlowBranchesByParent,
@@ -260,7 +429,7 @@ export const useSelectionStore = defineStore('selection', () => {
     clearFlowBranches,
     clearFlowBranchesByParent,
 
-    // 任务过程分支管理方法
+    // ==================== 任务过程分支管理方法 ====================
     addTaskProcessBranch,
     getTaskProcessBranches,
     getAllTaskProcessBranches,
@@ -268,14 +437,14 @@ export const useSelectionStore = defineStore('selection', () => {
     removeTaskProcessBranch,
     clearTaskProcessBranches,
 
-    // 🆕 任务过程分支生效状态管理方法
+    // ==================== 任务过程分支生效状态管理方法 ====================
     setActiveTaskProcessBranch,
-    setActiveTaskProcessData, // 🆕 新增
+    setActiveTaskProcessData,
     getActiveTaskProcessBranch,
-    getActiveTaskProcessData, // 🆕 新增
+    getActiveTaskProcessData,
     clearActiveTaskProcessBranch,
 
-    // Agent 组合 TaskProcess 数据管理方法
+    // ==================== Agent 组合 TaskProcess 数据管理方法 ====================
     getAgentGroupKey,
     setAgentTaskProcess,
     getAgentTaskProcess,
@@ -287,6 +456,7 @@ export const useSelectionStore = defineStore('selection', () => {
 
 /**
  * 用于在组件外部使用 Selection Store
+ * @returns Selection Store 实例
  */
 export function useSelectionStoreHook() {
   return useSelectionStore(store)

frontend/src/utils/retry.ts

@@ -0,0 +1,167 @@
+/**
+ * 重试工具函数
+ * @description 提供通用的重试机制，支持指数退避和自定义重试条件
+ */
+
+/**
+ * 重试选项配置
+ */
+export interface RetryOptions {
+  /** 最大重试次数，默认 3 */
+  maxRetries?: number
+  /** 初始延迟时间（毫秒），默认 2000 */
+  initialDelayMs?: number
+  /** 是否使用指数退避，默认 true */
+  useExponentialBackoff?: boolean
+  /** 最大延迟时间（毫秒），默认 30000 */
+  maxDelayMs?: number
+  /** 自定义重试条件函数，返回 true 表示应该重试 */
+  shouldRetry?: (error: any, attempt: number) => boolean
+  /** 重试前的回调函数 */
+  onRetry?: (error: any, attempt: number, delay: number) => void
+}
+
+/**
+ * 默认重试条件 - 检查是否是 rate limiting 错误
+ * @param error 错误对象
+ * @returns 是否应该重试
+ */
+export function defaultShouldRetry(error: any): boolean {
+  if (!error) return false
+  const message = error?.message || String(error)
+  // 检查 406 Not Acceptable、429 Too Many Requests 等 rate limiting 错误
+  const isRateLimit =
+    message.includes('406') ||
+    message.includes('429') ||
+    message.includes('Not Acceptable') ||
+    message.includes('Too Many Requests') ||
+    message.includes('rate limit')
+  return isRateLimit
+}
+
+/**
+ * 带重试机制的异步函数执行器
+ *
+ * @example
+ * ```typescript
+ * // 基本用法
+ * const result = await withRetry(() => api.someRequest())
+ *
+ * // 自定义重试次数和延迟
+ * const result = await withRetry(() => api.request(), {
+ *   maxRetries: 5,
+ *   initialDelayMs: 1000,
+ *   useExponentialBackoff: true
+ * })
+ *
+ * // 自定义重试条件
+ * const result = await withRetry(() => api.request(), {
+ *   shouldRetry: (error) => error?.code === 'NETWORK_ERROR'
+ * })
+ *
+ * // 带重试回调
+ * const result = await withRetry(() => api.request(), {
+ *   onRetry: (error, attempt, delay) => {
+ *     console.log(`第 ${attempt} 次重试，等待 ${delay}ms`, error.message)
+ *   }
+ * })
+ * ```
+ *
+ * @param fn - 要执行的异步函数
+ * @param options - 重试选项配置
+ * @returns Promise resolves with the result of the function
+ * @throws 如果重试次数用尽，抛出最后一次的错误
+ */
+export async function withRetry<T>(
+  fn: () => Promise<T>,
+  options: RetryOptions = {},
+): Promise<T> {
+  const {
+    maxRetries = 3,
+    initialDelayMs = 2000,
+    useExponentialBackoff = true,
+    maxDelayMs = 30000,
+    shouldRetry = defaultShouldRetry,
+    onRetry,
+  } = options
+
+  let lastError: any = null
+  let currentDelay = initialDelayMs
+
+  for (let attempt = 1; attempt <= maxRetries + 1; attempt++) {
+    try {
+      // 执行目标函数
+      return await fn()
+    } catch (error: any) {
+      lastError = error
+
+      // 判断是否应该重试
+      const shouldRetryAttempt = shouldRetry(error, attempt)
+
+      // 如果是最后一次尝试 或者 不应该重试，则抛出错误
+      if (attempt > maxRetries || !shouldRetryAttempt) {
+        if (attempt > maxRetries) {
+          console.error(
+            `❌ [withRetry] 已达到最大重试次数 (${maxRetries})，放弃请求`,
+          )
+        }
+        throw error
+      }
+
+      // 执行重试回调
+      if (onRetry) {
+        onRetry(error, attempt, currentDelay)
+      }
+
+      // 打印日志
+      console.log(
+        `⏳ [withRetry] 第 ${attempt} 次重试，等待 ${currentDelay}ms...`,
+        error?.message || String(error),
+      )
+
+      // 等待延迟时间
+      await new Promise((resolve) => setTimeout(resolve, currentDelay))
+
+      // 计算下一次延迟时间（指数退避）
+      if (useExponentialBackoff) {
+        currentDelay = Math.min(currentDelay * 2, maxDelayMs)
+      }
+    }
+  }
+
+  //理论上不会到达这里，因为循环内会 throw
+  throw lastError
+}
+
+/**
+ * 简化的重试装饰器 - 适用于类方法
+ *
+ * @example
+ * ```typescript
+ * class MyService {
+ *   @retryable({ maxRetries: 3, initialDelayMs: 1000 })
+ *   async fetchData() {
+ *     return await api.request()
+ *   }
+ * }
+ * ```
+ *
+ * @param options - 重试选项配置
+ * @returns 装饰器函数
+ */
+export function retryable(options: RetryOptions = {}) {
+  return function <T extends (...args: any[]) => Promise<any>>(
+    _target: any,
+    _propertyKey: string,
+    descriptor: TypedPropertyDescriptor<T>,
+  ) {
+    const originalMethod = descriptor.value
+    if (!originalMethod) return descriptor
+
+    descriptor.value = function (this: any, ...args: Parameters<T>) {
+      return withRetry(() => originalMethod.apply(this, args), options)
+    } as T
+
+    return descriptor
+  } as MethodDecorator
+}

frontend/src/utils/websocket.ts

@@ -0,0 +1,300 @@
+/**
+ * WebSocket 客户端封装
+ * 基于 socket.io-client 实现
+ */
+
+import { io, Socket } from 'socket.io-client'
+
+interface WebSocketConfig {
+  url?: string
+  reconnectionAttempts?: number
+  reconnectionDelay?: number
+  timeout?: number
+}
+
+interface RequestMessage {
+  id: string
+  action: string
+  data: any
+}
+
+interface ResponseMessage {
+  id: string
+  status: 'success' | 'error' | 'streaming' | 'complete'
+  data?: any
+  error?: string
+  stage?: string
+  message?: string
+  [key: string]: any
+}
+
+interface StreamProgressCallback {
+  (data: any): void
+}
+
+type RequestHandler = {
+  resolve: (value: any) => void
+  reject: (error: Error) => void
+  timer?: ReturnType<typeof setTimeout>
+  onProgress?: StreamProgressCallback
+}
+
+class WebSocketClient {
+  private socket: Socket | null = null
+  private requestHandlers = new Map<string, RequestHandler>()
+  private streamHandlers = new Map<string, StreamProgressCallback>()
+  private config: Required<WebSocketConfig>
+  private isConnected = false
+
+  constructor() {
+    this.config = {
+      url: '',
+      reconnectionAttempts: 5,
+      reconnectionDelay: 1000,
+      timeout: 300000, // 5分钟超时
+    }
+  }
+
+  /**
+   * 连接到WebSocket服务器
+   */
+  connect(url?: string): Promise<void> {
+    return new Promise((resolve, reject) => {
+      const wsUrl = url || this.config.url || window.location.origin
+
+      this.socket = io(wsUrl, {
+        transports: ['websocket', 'polling'],
+        reconnection: true,
+        reconnectionAttempts: this.config.reconnectionAttempts,
+        reconnectionDelay: this.config.reconnectionDelay,
+      })
+
+      this.socket.on('connect', () => {
+        this.isConnected = true
+        resolve()
+      })
+
+      this.socket.on('connect_error', (error) => {
+        reject(error)
+      })
+
+      this.socket.on('disconnect', () => {
+        this.isConnected = false
+      })
+
+      this.socket.on('connected', () => {
+        // Server connected message
+      })
+
+      // 监听响应消息
+      this.socket.on('response', (response: ResponseMessage) => {
+        const { id, status, data, error, generation_id, execution_id } = response
+        const handler = this.requestHandlers.get(id)
+
+        if (handler) {
+          // 清除超时定时器
+          if (handler.timer) {
+            clearTimeout(handler.timer)
+          }
+
+          if (status === 'success') {
+            // 返回完整响应，包含 data、generation_id、execution_id 等
+            // 注意：需要检查 data 是否为 null，因为 typeof null === 'object'
+            // generation_id/execution_id 可能放在 data 中，需要兼容处理
+            // 注意：如果 data 是数组，不能展开，否则会破坏数组结构
+            const resolvedGenerationId = generation_id || (data && typeof data === 'object' && !Array.isArray(data) && data.generation_id)
+            const resolvedExecutionId = execution_id || (data && typeof data === 'object' && !Array.isArray(data) && data.execution_id)
+
+            // 直接返回 data，保持原始数据结构（数组或对象）
+            handler.resolve({
+              data,
+              generation_id: resolvedGenerationId,
+              execution_id: resolvedExecutionId
+            })
+          } else {
+            handler.reject(new Error(error || 'Unknown error'))
+          }
+
+          // 删除处理器
+          this.requestHandlers.delete(id)
+        }
+      })
+
+      // 监听流式进度消息
+      this.socket.on('progress', (response: ResponseMessage) => {
+        const { id, status, data, error } = response
+
+        // 首先检查是否有对应的流式处理器
+        const streamCallback = this.streamHandlers.get(id)
+        if (streamCallback) {
+          if (status === 'streaming') {
+            // 解析 data 字段（JSON 字符串）并传递给回调
+            try {
+              const parsedData = typeof data === 'string' ? JSON.parse(data) : data
+              streamCallback(parsedData)
+            } catch (e) {
+              // Failed to parse progress data
+            }
+          } else if (status === 'complete') {
+            this.streamHandlers.delete(id)
+            streamCallback({ type: 'complete' })
+          } else if (status === 'error') {
+            this.streamHandlers.delete(id)
+            streamCallback({ type: 'error', error })
+          }
+          return
+        }
+
+        // 检查是否有对应的普通请求处理器（支持send()方法的进度回调）
+        const requestHandler = this.requestHandlers.get(id)
+        if (requestHandler && requestHandler.onProgress) {
+          // 解析 data 字段并传递给进度回调
+          try {
+            const parsedData = typeof data === 'string' ? JSON.parse(data) : data
+            requestHandler.onProgress(parsedData)
+          } catch (e) {
+            // Failed to parse progress data
+          }
+        }
+      })
+
+      // 心跳检测
+      this.socket.on('pong', () => {
+        // Pong received
+      })
+    })
+  }
+
+  /**
+   * 发送请求（双向通信，支持可选的进度回调）
+   */
+  send(action: string, data: any, timeout?: number, onProgress?: StreamProgressCallback): Promise<any> {
+    if (!this.socket || !this.isConnected) {
+      return Promise.reject(new Error('WebSocket未连接'))
+    }
+
+    return new Promise((resolve, reject) => {
+      const requestId = `${action}-${Date.now()}-${Math.random().toString(36).substr(2, 9)}`
+
+      // 设置超时
+      const timeoutMs = timeout || this.config.timeout
+      const timer = setTimeout(() => {
+        if (this.requestHandlers.has(requestId)) {
+          this.requestHandlers.delete(requestId)
+          reject(new Error(`Request timeout: ${action}`))
+        }
+      }, timeoutMs)
+
+      // 保存处理器（包含可选的进度回调）
+      this.requestHandlers.set(requestId, { resolve, reject, timer, onProgress })
+
+      // 发送消息
+      this.socket!.emit(action, {
+        id: requestId,
+        action,
+        data,
+      } as RequestMessage)
+    })
+  }
+
+  /**
+   * 订阅流式数据
+   */
+  subscribe(
+    action: string,
+    data: any,
+    onProgress: StreamProgressCallback,
+    onComplete?: () => void,
+    onError?: (error: Error) => void,
+  ): void {
+    if (!this.socket || !this.isConnected) {
+      onError?.(new Error('WebSocket未连接'))
+      return
+    }
+
+    const requestId = `${action}-${Date.now()}-${Math.random().toString(36).substr(2, 9)}`
+
+    // 保存流式处理器
+    const wrappedCallback = (progressData: any) => {
+      if (progressData?.type === 'complete') {
+        this.streamHandlers.delete(requestId)
+        onComplete?.()
+      } else {
+        onProgress(progressData)
+      }
+    }
+
+    this.streamHandlers.set(requestId, wrappedCallback)
+
+    // 发送订阅请求
+    this.socket.emit(action, {
+      id: requestId,
+      action,
+      data,
+    } as RequestMessage)
+  }
+
+  /**
+   * 发送心跳
+   */
+  ping(): void {
+    if (this.socket && this.isConnected) {
+      this.socket.emit('ping')
+    }
+  }
+
+  /**
+   * 断开连接
+   */
+  disconnect(): void {
+    if (this.socket) {
+      // 清理所有处理器
+      this.requestHandlers.forEach((handler) => {
+        if (handler.timer) {
+          clearTimeout(handler.timer)
+        }
+      })
+      this.requestHandlers.clear()
+      this.streamHandlers.clear()
+
+      this.socket.disconnect()
+      this.socket = null
+      this.isConnected = false
+    }
+  }
+
+  /**
+   * 获取连接状态
+   */
+  get connected(): boolean {
+    return this.isConnected && this.socket?.connected === true
+  }
+
+  /**
+   * 获取Socket ID
+   */
+  get id(): string | undefined {
+    return this.socket?.id
+  }
+
+  /**
+   * 监听事件
+   */
+  on(event: string, callback: (...args: any[]) => void): void {
+    if (this.socket) {
+      this.socket.on(event, callback)
+    }
+  }
+
+  /**
+   * 取消监听事件
+   */
+  off(event: string, callback?: (...args: any[]) => void): void {
+    if (this.socket) {
+      this.socket.off(event, callback)
+    }
+  }
+}
+
+// 导出单例
+export default new WebSocketClient()

frontend/vite.config.ts

@@ -47,7 +47,8 @@ export default defineConfig({
       '/api': {
         changeOrigin: true,
         // 接口地址
-        target: 'http://82.157.183.212:21092',
+        // target: 'http://82.157.183.212:21092',
+        target: 'http://82.157.183.212:21097',
         // target: 'http://localhost:8000',
         // rewrite: (path: string) => path.replace(/^\/api/, ''),
         // configure: (proxy, options) => {