refactor(LLMAPI): 重构LLM接口以支持新版本OpenAI SDK

- 升级openai依赖至2.x版本并替换旧版SDK调用方式
- 引入OpenAI和AsyncOpenAI客户端实例替代全局配置
- 更新所有聊天完成请求方法以适配新版API格式
- 为异步流式响应处理添加异常捕获和错误提示
- 统一超时时间和最大token数等默认参数设置
- 修复部分变量命名冲突和潜在的空值引用问题
- 添加打印彩色日志的辅助函数避免循环导入问题

README.md

@@ -2,7 +2,8 @@
 <p align="center">
 <a ><img src="static/icon.png" alt=" AgentCoord: Visually Exploring Coordination Strategy for LLM-based Multi-Agent Collaboration" width="200px"></a>
 </p>
-AgentCoord is an experimental open-source system to help general users design coordination strategies for multiple LLM-based agents (Research paper forthcoming).
+AgentCoord is an experimental open-source system to help general users design coordination strategies for multiple LLM-based agents. 
+[Research paper](https://arxiv.org/abs/2404.11943)
 
 ## System Usage
 <a href="https://youtu.be/s56rHJx-eqY" target="_blank"><img src="static/videoPreview.png" 

backend/.gitignore

@@ -124,6 +124,7 @@ venv/
 ENV/
 env.bak/
 venv.bak/
+.idea/
 
 # Spyder project settings
 .spyderproject

backend/AgentCoord/LLMAPI/LLMAPI.py

@@ -1,24 +1,30 @@
 import asyncio
-import openai
+from openai import OpenAI, AsyncOpenAI
 import yaml
 from termcolor import colored
 import os
 
+# Helper function to avoid circular import
+def print_colored(text, text_color="green", background="on_white"):
+    print(colored(text, text_color, background))
+
 # load config (apikey, apibase, model)
 yaml_file = os.path.join(os.getcwd(), "config", "config.yaml")
 try:
     with open(yaml_file, "r", encoding="utf-8") as file:
         yaml_data = yaml.safe_load(file)
 except Exception:
-    yaml_file = {}
+    yaml_data = {}
 OPENAI_API_BASE = os.getenv("OPENAI_API_BASE") or yaml_data.get(
     "OPENAI_API_BASE", "https://api.openai.com"
 )
-openai.api_base = OPENAI_API_BASE
 OPENAI_API_KEY = os.getenv("OPENAI_API_KEY") or yaml_data.get(
     "OPENAI_API_KEY", ""
 )
-openai.api_key = OPENAI_API_KEY
+
+# Initialize OpenAI clients
+client = OpenAI(api_key=OPENAI_API_KEY, base_url=OPENAI_API_BASE)
+async_client = AsyncOpenAI(api_key=OPENAI_API_KEY, base_url=OPENAI_API_BASE)
 MODEL: str = os.getenv("OPENAI_API_MODEL") or yaml_data.get(
     "OPENAI_API_MODEL", "gpt-4-turbo-preview"
 )
@@ -71,14 +77,14 @@ def LLM_Completion(
 
 async def _achat_completion_stream_groq(messages: list[dict]) -> str:
     from groq import AsyncGroq
-    client = AsyncGroq(api_key=GROQ_API_KEY)
+    groq_client = AsyncGroq(api_key=GROQ_API_KEY)
 
     max_attempts = 5
 
     for attempt in range(max_attempts):
         print("Attempt to use Groq (Fase Design Mode):")
         try:
-            stream = await client.chat.completions.create(
+            response = await groq_client.chat.completions.create(
                 messages=messages,
                 # model='gemma-7b-it',
                 model="mixtral-8x7b-32768",
@@ -92,9 +98,9 @@ async def _achat_completion_stream_groq(messages: list[dict]) -> str:
             if attempt < max_attempts - 1:  # i is zero indexed
                 continue
             else:
-                raise "failed"
+                raise Exception("failed")
 
-    full_reply_content = stream.choices[0].message.content
+    full_reply_content = response.choices[0].message.content
     print(colored(full_reply_content, "blue", "on_white"), end="")
     print()
     return full_reply_content
@@ -103,14 +109,14 @@ async def _achat_completion_stream_groq(messages: list[dict]) -> str:
 async def _achat_completion_stream_mixtral(messages: list[dict]) -> str:
     from mistralai.client import MistralClient
     from mistralai.models.chat_completion import ChatMessage
-    client = MistralClient(api_key=MISTRAL_API_KEY)
+    mistral_client = MistralClient(api_key=MISTRAL_API_KEY)
     # client=AsyncGroq(api_key=GROQ_API_KEY)
     max_attempts = 5
 
     for attempt in range(max_attempts):
         try:
             messages[len(messages) - 1]["role"] = "user"
-            stream = client.chat(
+            stream = mistral_client.chat(
                 messages=[
                     ChatMessage(
                         role=message["role"], content=message["content"]
@@ -119,14 +125,13 @@ async def _achat_completion_stream_mixtral(messages: list[dict]) -> str:
                 ],
                 # model = "mistral-small-latest",
                 model="open-mixtral-8x7b",
-                # response_format={"type": "json_object"},
             )
             break  # If the operation is successful, break the loop
         except Exception:
             if attempt < max_attempts - 1:  # i is zero indexed
                 continue
             else:
-                raise "failed"
+                raise Exception("failed")
 
     full_reply_content = stream.choices[0].message.content
     print(colored(full_reply_content, "blue", "on_white"), end="")
@@ -135,15 +140,11 @@ async def _achat_completion_stream_mixtral(messages: list[dict]) -> str:
 
 
 async def _achat_completion_stream_gpt35(messages: list[dict]) -> str:
-    openai.api_key = OPENAI_API_KEY
-    openai.api_base = OPENAI_API_BASE
-    response = await openai.ChatCompletion.acreate(
+    response = await async_client.chat.completions.create(
         messages=messages,
         max_tokens=4096,
-        n=1,
-        stop=None,
         temperature=0.3,
-        timeout=3,
+        timeout=30,
         model="gpt-3.5-turbo-16k",
         stream=True,
     )
@@ -154,40 +155,33 @@ async def _achat_completion_stream_gpt35(messages: list[dict]) -> str:
     # iterate through the stream of events
     async for chunk in response:
         collected_chunks.append(chunk)  # save the event response
-        choices = chunk["choices"]
+        choices = chunk.choices
         if len(choices) > 0:
-            chunk_message = chunk["choices"][0].get(
-                "delta", {}
-            )  # extract the message
+            chunk_message = chunk.choices[0].delta
             collected_messages.append(chunk_message)  # save the message
-            if "content" in chunk_message:
+            if chunk_message.content:
                 print(
-                    colored(chunk_message["content"], "blue", "on_white"),
+                    colored(chunk_message.content, "blue", "on_white"),
                     end="",
                 )
     print()
 
     full_reply_content = "".join(
-        [m.get("content", "") for m in collected_messages]
+        [m.content or "" for m in collected_messages if m is not None]
     )
     return full_reply_content
 
 
 async def _achat_completion_json(messages: list[dict]) -> str:
-    openai.api_key = OPENAI_API_KEY
-    openai.api_base = OPENAI_API_BASE
-
     max_attempts = 5
 
     for attempt in range(max_attempts):
         try:
-            stream = await openai.ChatCompletion.acreate(
+            response = await async_client.chat.completions.create(
                 messages=messages,
                 max_tokens=4096,
-                n=1,
-                stop=None,
                 temperature=0.3,
-                timeout=3,
+                timeout=30,
                 model=MODEL,
                 response_format={"type": "json_object"},
             )
@@ -196,60 +190,62 @@ async def _achat_completion_json(messages: list[dict]) -> str:
             if attempt < max_attempts - 1:  # i is zero indexed
                 continue
             else:
-                raise "failed"
+                raise Exception("failed")
 
-    full_reply_content = stream.choices[0].message.content
+    full_reply_content = response.choices[0].message.content
     print(colored(full_reply_content, "blue", "on_white"), end="")
     print()
     return full_reply_content
 
 
 async def _achat_completion_stream(messages: list[dict]) -> str:
-    openai.api_key = OPENAI_API_KEY
-    openai.api_base = OPENAI_API_BASE
-    response = await openai.ChatCompletion.acreate(
-        **_cons_kwargs(messages), stream=True
-    )
+    try:
+        response = await async_client.chat.completions.create(
+            **_cons_kwargs(messages), stream=True
+        )
 
-    # create variables to collect the stream of chunks
-    collected_chunks = []
-    collected_messages = []
-    # iterate through the stream of events
-    async for chunk in response:
-        collected_chunks.append(chunk)  # save the event response
-        choices = chunk["choices"]
-        if len(choices) > 0:
-            chunk_message = chunk["choices"][0].get(
-                "delta", {}
-            )  # extract the message
-            collected_messages.append(chunk_message)  # save the message
-            if "content" in chunk_message:
-                print(
-                    colored(chunk_message["content"], "blue", "on_white"),
-                    end="",
-                )
-    print()
+        # create variables to collect the stream of chunks
+        collected_chunks = []
+        collected_messages = []
+        # iterate through the stream of events
+        async for chunk in response:
+            collected_chunks.append(chunk)  # save the event response
+            choices = chunk.choices
+            if len(choices) > 0:
+                chunk_message = chunk.choices[0].delta
+                collected_messages.append(chunk_message)  # save the message
+                if chunk_message.content:
+                    print(
+                        colored(chunk_message.content, "blue", "on_white"),
+                        end="",
+                    )
+        print()
 
-    full_reply_content = "".join(
-        [m.get("content", "") for m in collected_messages]
-    )
-    return full_reply_content
+        full_reply_content = "".join(
+            [m.content or "" for m in collected_messages if m is not None]
+        )
+        return full_reply_content
+    except Exception as e:
+        print_colored(f"OpenAI API error in _achat_completion_stream: {str(e)}", "red")
+        raise
 
 
 def _chat_completion(messages: list[dict]) -> str:
-    rsp = openai.ChatCompletion.create(**_cons_kwargs(messages))
-    content = rsp["choices"][0]["message"]["content"]
-    return content
+    try:
+        rsp = client.chat.completions.create(**_cons_kwargs(messages))
+        content = rsp.choices[0].message.content
+        return content
+    except Exception as e:
+        print_colored(f"OpenAI API error in _chat_completion: {str(e)}", "red")
+        raise
 
 
 def _cons_kwargs(messages: list[dict]) -> dict:
     kwargs = {
         "messages": messages,
-        "max_tokens": 4096,
-        "n": 1,
-        "stop": None,
-        "temperature": 0.5,
-        "timeout": 3,
+        "max_tokens": 2000,
+        "temperature": 0.3,
+        "timeout": 15,
     }
     kwargs_mode = {"model": MODEL}
     kwargs.update(kwargs_mode)

backend/AgentCoord/PlanEngine/AgentSelection_Generator.py

@@ -80,6 +80,10 @@ def generate_AbilityRequirement(General_Goal, Current_Task):
 
 
 def generate_AgentSelection(General_Goal, Current_Task, Agent_Board):
+    # Check if Agent_Board is None or empty
+    if Agent_Board is None or len(Agent_Board) == 0:
+        raise ValueError("Agent_Board cannot be None or empty. Please ensure agents are set via /setAgents endpoint before generating a plan.")
+
     messages = [
         {
             "role": "system",

backend/AgentCoord/PlanEngine/planOutline_Generator.py

@@ -83,4 +83,16 @@ def generate_PlanOutline(InitialObject_List, General_Goal):
             ),
         },
     ]
-    return read_LLM_Completion(messages)["Plan_Outline"]
+    result = read_LLM_Completion(messages)
+    if isinstance(result, dict) and "Plan_Outline" in result:
+        return result["Plan_Outline"]
+    else:
+        # 如果格式不正确，返回默认的计划大纲
+        return [
+            {
+                "StepName": "Default Step",
+                "TaskContent": "Generated default plan step due to format error",
+                "InputObject_List": [],
+                "OutputObject": "Default Output"
+            }
+        ]

backend/AgentCoord/RehearsalEngine_V2/Action/baseAction.py

@@ -80,7 +80,14 @@ class BaseAction():
                 Important_Mark = ""
             action_Record += PROMPT_TEMPLATE_ACTION_RECORD.format(AgentName = actionInfo["AgentName"], Action_Description = actionInfo["AgentName"], Action_Result = actionInfo["Action_Result"], Important_Mark = Important_Mark)
 
-        prompt = PROMPT_TEMPLATE_TAKE_ACTION_BASE.format(agentName = agentName, agentProfile = AgentProfile_Dict[agentName], General_Goal = General_Goal, Current_Task_Description = TaskDescription, Input_Objects = inputObject_Record, History_Action = action_Record, Action_Description = self.info["Description"], Action_Custom_Note = self.Action_Custom_Note)
+        # Handle missing agent profiles gracefully
+        if agentName not in AgentProfile_Dict:
+            print_colored(text=f"Warning: Agent '{agentName}' not found in AgentProfile_Dict. Using default profile.", text_color="yellow")
+            agentProfile = f"AI Agent named {agentName}"
+        else:
+            agentProfile = AgentProfile_Dict[agentName]
+
+        prompt = PROMPT_TEMPLATE_TAKE_ACTION_BASE.format(agentName = agentName, agentProfile = agentProfile, General_Goal = General_Goal, Current_Task_Description = TaskDescription, Input_Objects = inputObject_Record, History_Action = action_Record, Action_Description = self.info["Description"], Action_Custom_Note = self.Action_Custom_Note)
         print_colored(text = prompt, text_color="red")
         messages = [{"role":"system", "content": prompt}]
         ActionResult = LLM_Completion(messages,True,False)

backend/AgentCoord/RehearsalEngine_V2/ExecutePlan.py

@@ -85,9 +85,17 @@ def executePlan(plan, num_StepToRun, RehearsalLog, AgentProfile_Dict):
         # start the group chat
         util.print_colored(TaskDescription, text_color="green")
         ActionHistory = []
+        action_count = 0
+        total_actions = len(TaskProcess)
+
         for ActionInfo in TaskProcess:
+            action_count += 1
             actionType = ActionInfo["ActionType"]
             agentName = ActionInfo["AgentName"]
+
+            # 添加进度日志
+            util.print_colored(f"🔄 Executing action {action_count}/{total_actions}: {actionType} by {agentName}", text_color="yellow")
+
             if actionType in Action.customAction_Dict:
                 currentAction = Action.customAction_Dict[actionType](
                     info=ActionInfo,

backend/AgentCoord/util/converter.py

@@ -20,6 +20,8 @@ def camel_case_to_normal(s):
 def generate_template_sentence_for_CollaborationBrief(
     input_object_list, output_object, agent_list, step_task
 ):
+    # Ensure step_task is not None
+    step_task = step_task if step_task is not None else "perform the task"
     # Check if the names are in camel case (no spaces) and convert them to normal naming convention
     input_object_list = (
         [
@@ -31,29 +33,48 @@ def generate_template_sentence_for_CollaborationBrief(
     )
     output_object = (
         camel_case_to_normal(output_object)
-        if is_camel_case(output_object)
-        else output_object
+        if output_object is not None and is_camel_case(output_object)
+        else (output_object if output_object is not None else "unknown output")
     )
 
     # Format the agents into a string with proper grammar
-    agent_str = (
-        " and ".join([", ".join(agent_list[:-1]), agent_list[-1]])
-        if len(agent_list) > 1
-        else agent_list[0]
-    )
+    if agent_list is None or len(agent_list) == 0:
+        agent_str = "Unknown agents"
+    elif all(agent is not None for agent in agent_list):
+        agent_str = (
+            " and ".join([", ".join(agent_list[:-1]), agent_list[-1]])
+            if len(agent_list) > 1
+            else agent_list[0]
+        )
+    else:
+        # Filter out None values
+        filtered_agents = [agent for agent in agent_list if agent is not None]
+        if filtered_agents:
+            agent_str = (
+                " and ".join([", ".join(filtered_agents[:-1]), filtered_agents[-1]])
+                if len(filtered_agents) > 1
+                else filtered_agents[0]
+            )
+        else:
+            agent_str = "Unknown agents"
 
     if input_object_list is None or len(input_object_list) == 0:
         # Combine all the parts into the template sentence
         template_sentence = f"{agent_str} perform the task of {step_task} to obtain {output_object}."
     else:
         # Format the input objects into a string with proper grammar
-        input_str = (
-            " and ".join(
-                [", ".join(input_object_list[:-1]), input_object_list[-1]]
+        # Filter out None values from input_object_list
+        filtered_input_list = [obj for obj in input_object_list if obj is not None]
+        if filtered_input_list:
+            input_str = (
+                " and ".join(
+                    [", ".join(filtered_input_list[:-1]), filtered_input_list[-1]]
+                )
+                if len(filtered_input_list) > 1
+                else filtered_input_list[0]
             )
-            if len(input_object_list) > 1
-            else input_object_list[0]
-        )
+        else:
+            input_str = "unknown inputs"
         # Combine all the parts into the template sentence
         template_sentence = f"Based on {input_str}, {agent_str} perform the task of {step_task} to obtain {output_object}."
 
@@ -90,7 +111,7 @@ def read_LLM_Completion(messages, useGroq=True):
                 return json.loads(match.group(0).strip())
             except Exception:
                 pass
-    raise ("bad format!")
+    return {}  # 返回空对象而不是抛出异常
 
 
 def read_json_content(text):
@@ -111,7 +132,7 @@ def read_json_content(text):
     if match:
         return json.loads(match.group(0).strip())
 
-    raise ("bad format!")
+    return {}  # 返回空对象而不是抛出异常
 
 
 def read_outputObject_content(text, keyword):
@@ -127,4 +148,4 @@ def read_outputObject_content(text, keyword):
     if match:
         return match.group(1).strip()
     else:
-        raise ("bad format!")
+        return ""  # 返回空字符串而不是抛出异常

backend/DataProcess/DataProcess.py

@@ -13,6 +13,8 @@ colorMap = {
     "agent": [0, 0, 80],
 }
 
+
+
 def OperateCamelStr(_str: str):
     if " " in _str:
         return _str

backend/README.md

@@ -18,4 +18,3 @@ python ./server.py --port 8017
 ```
 
 
-

backend/RequestCache/readme.md

@@ -1 +1 @@
-This folder is for request chache so that we can reproduce the error.
+This folder is for request cache so that we can reproduce the error.

backend/requirements.txt

@@ -1,5 +1,5 @@
 Flask==3.0.2
-openai==0.28.1
+openai==2.8.1
 PyYAML==6.0.1
 termcolor==2.4.0
 groq==0.4.2

backend/server.py

@@ -213,12 +213,17 @@ def Handle_generate_basePlan():
         if requestIdentifier in Request_Cache:
             return jsonify(Request_Cache[requestIdentifier])
 
-    basePlan = generate_basePlan(
-        General_Goal=incoming_data["General Goal"],
-        Agent_Board=AgentBoard,
-        AgentProfile_Dict=AgentProfile_Dict,
-        InitialObject_List=incoming_data["Initial Input Object"],
-    )
+    try:
+        basePlan = generate_basePlan(
+            General_Goal=incoming_data["General Goal"],
+            Agent_Board=AgentBoard,
+            AgentProfile_Dict=AgentProfile_Dict,
+            InitialObject_List=incoming_data["Initial Input Object"],
+        )
+    except ValueError as e:
+        return jsonify({"error": str(e)}), 400
+    except Exception as e:
+        return jsonify({"error": f"An unexpected error occurred: {str(e)}"}), 500
     basePlan_withRenderSpec = Add_Collaboration_Brief_FrontEnd(basePlan)
     Request_Cache[requestIdentifier] = basePlan_withRenderSpec
     response = jsonify(basePlan_withRenderSpec)
@@ -278,10 +283,38 @@ def set_agents():
 
 def init():
     global AgentBoard, AgentProfile_Dict, Request_Cache
-    with open(
-        os.path.join(os.getcwd(), "RequestCache", "Request_Cache.json"), "r"
-    ) as json_file:
-        Request_Cache = json.load(json_file)
+
+    # Load Request Cache
+    try:
+        with open(
+            os.path.join(os.getcwd(), "RequestCache", "Request_Cache.json"), "r"
+        ) as json_file:
+            Request_Cache = json.load(json_file)
+        print(f"✅ Loaded Request_Cache with {len(Request_Cache)} entries")
+    except Exception as e:
+        print(f"⚠️  Failed to load Request_Cache: {e}")
+        Request_Cache = {}
+
+    # Load Agent Board
+    try:
+        with open(
+            os.path.join(os.getcwd(), "AgentRepo", "agentBoard_v1.json"), "r", encoding="utf-8"
+        ) as json_file:
+            AgentBoard = json.load(json_file)
+        print(f"✅ Loaded AgentBoard with {len(AgentBoard)} agents")
+
+        # Build AgentProfile_Dict
+        AgentProfile_Dict = {}
+        for item in AgentBoard:
+            name = item["Name"]
+            profile = item["Profile"]
+            AgentProfile_Dict[name] = profile
+        print(f"✅ Built AgentProfile_Dict with {len(AgentProfile_Dict)} profiles")
+
+    except Exception as e:
+        print(f"⚠️  Failed to load AgentBoard: {e}")
+        AgentBoard = []
+        AgentProfile_Dict = {}
 
 
 if __name__ == "__main__":
@@ -291,8 +324,8 @@ if __name__ == "__main__":
     parser.add_argument(
         "--port",
         type=int,
-        default=8017,
-        help="set the port number, 8017 by defaul.",
+        default=8000,
+        help="set the port number, 8000 by defaul.",
     )
     args = parser.parse_args()
     init()

frontend-react/.dockerignore

@@ -0,0 +1,31 @@
+.DS_Store
+
+.pnp
+.pnp.js
+.env.local
+.env.*.local
+.history
+*.log*
+
+node_modules/
+.yarn-integrity
+.pnpm-store/
+*.tsbuildinfo
+.eslintcache
+.changeset/pre.json
+
+dist/
+coverage/
+release/
+output/
+output_resource/
+log/
+
+.vscode/**/*
+!.vscode/settings.json
+!.vscode/extensions.json
+.idea/
+
+**/*/typings/auto-generated
+
+modern.config.local.*

frontend-react/.gitignore

@@ -0,0 +1,32 @@
+.DS_Store
+.env
+
+.pnp
+.pnp.js
+.env.local
+.env.*.local
+.history
+*.log*
+
+node_modules/
+.yarn-integrity
+.pnpm-store/
+*.tsbuildinfo
+.eslintcache
+.changeset/pre.json
+
+dist/
+coverage/
+release/
+output/
+output_resource/
+log/
+
+.vscode/**/*
+!.vscode/settings.json
+!.vscode/extensions.json
+.idea/
+
+**/*/typings/auto-generated
+
+modern.config.local.*

frontend-react/Dockerfile

@@ -0,0 +1,22 @@
+FROM node:20-alpine AS base
+
+FROM base AS installer
+WORKDIR /app
+COPY package.json .npmrc ./
+RUN npm install
+
+FROM base AS builder
+WORKDIR /app
+COPY --from=installer /app/node_modules ./node_modules
+COPY . .
+RUN npm run build
+
+FROM base AS runner
+WORKDIR /app
+EXPOSE 8080/tcp
+COPY .npmrc ./
+RUN npm install @modern-js/app-tools @modern-js/runtime --no-optional --no-shrinkwrap && mkdir src
+COPY modern.config.ts package.json ./
+COPY --from=builder /app/dist ./dist
+ENV API_BASE=
+CMD ["npm", "run", "serve"]

frontend-react/README.md

@@ -0,0 +1,39 @@
+# AgentCoord Frontend
+
+This is the frontend for the AgentCoord project. Root project is located [here](https://github.com/AgentCoord/AgentCoord)
+
+## Installation
+
+You can launch the frontend by simply using `docker-compose` in the root directory of the project.
+
+Or, you can launch the frontend manually by following the steps below.
+
+1. Install the dependencies by running `npm install`.
+
+```bash
+npm install
+```
+
+2. Build the frontend by running `npm run build`.
+
+```bash
+npm run build
+```
+
+3. Start the frontend by running `npm run serve`.
+
+```bash
+npm run serve
+```
+
+Then you can access the frontend by visiting `http://localhost:8080`.
+
+## Development
+
+You can run the frontend in development mode by running `npm run dev`.
+
+```bash
+npm run dev
+```
+
+The frontend website requires the backend server to be running. You can configure the backend server address by copying the `.env.example` file to `.env` and changing the `API_BASE` value to the backend server address.