🐛 a better way to distinguish between executable and generated code #832

Author: Phinease

backend/prompts/managed_system_prompt_template.yaml

@@ -61,6 +61,8 @@ system_prompt: |-
      - 用简单的Python编写代码
      - 遵循python代码规范和python语法
      - 根据格式规范正确调用工具
+     - 考虑到代码执行与展示用户代码的区别，使用'代码：\n```<RUN>\n'开头，并以'```<END_CODE>'表达运行代码，使用'代码：\n```<DISPLAY:语言类型>\n'开头，并以'```<END_CODE>'表达展示代码
+     - 注意运行的代码不会被用户看到，所以如果用户需要看到代码，你需要使用'代码：\n```<DISPLAY:语言类型>\n'开头，并以'```<END_CODE>'表达展示代码。
 
   3. 观察结果：
      - 查看代码执行结果
@@ -99,7 +101,7 @@ system_prompt: |-
   {{ constraint }}
 
   ### python代码规范
-  1. 如果认为是需要执行的代码，代码内容以'代码：\n```py\n'开头，并以'```<end_code>'标识符结尾。如果是不需要执行仅用于展示的代码，代码内容以'代码：\n```code:语言类型\n'开头，并以'```<end_code>'标识符结尾，其中语言类型例如python、java、javascript等；
+  1. 如果认为是需要执行的代码，代码内容以'代码：\n```<RUN>\n'开头，并以'```<END_CODE>'标识符结尾。如果是不需要执行仅用于展示的代码，代码内容以'代码：\n```<DISPLAY:语言类型>\n'开头，并以'```<END_CODE>'标识符结尾，其中语言类型例如python、java、javascript等；
   2. 只使用已定义的变量，变量将在多次调用之间持续保持；
   3. 使用“print()”函数让下一次的模型调用看到对应变量信息；
   4. 正确使用工具的入参，使用关键字参数，不要用字典形式；

backend/prompts/managed_system_prompt_template_en.yaml

@@ -61,6 +61,8 @@ system_prompt: |-
      - Write code in simple Python
      - Follow Python coding standards and Python syntax
      - Call tools correctly according to format specifications
+     - To distinguish between code execution and displaying user code, use 'Code: \n```<RUN>\n' to start executing code and '```<END_CODE>' to indicate its completion. Use 'Code: \n```<DISPLAY:language_type>\n' to start displaying code and '```<END_CODE>' to indicate its completion.
+     - Note that executed code is not visible to users. If users need to see the code, use 'Code: \n```<DISPLAY:language_type>\n' as the start and '```<END_CODE>' to denote displayed code.
 
   3. Observe Results:
      - View code execution results
@@ -99,7 +101,7 @@ system_prompt: |-
   {{ constraint }}
 
   ### Python Code Specifications
-  1. If it is considered to be code that needs to be executed, the code content begins with 'code: \n```py\n' and ends with '```<end_code>'. If the code does not need to be executed for display only, the code content begins with 'code:\n```code:language_type\n', and ends with '```<end_code>', where language_type can be python, java, javascript, etc;
+  1. If it is considered to be code that needs to be executed, the code content begins with 'code: \n```<RUN>\n' and ends with '```<END_CODE>'. If the code does not need to be executed for display only, the code content begins with 'code:\n```<DISPLAY:language_type>\n', and ends with '```<END_CODE>', where language_type can be python, java, javascript, etc;
   2. Only use defined variables, variables will persist between multiple calls;
   3. Use "print()" function to let the next model call see corresponding variable information;
   4. Use tool input parameters correctly, use keyword arguments, not dictionary format;

backend/prompts/manager_system_prompt_template.yaml

@@ -62,6 +62,8 @@ system_prompt: |-
      - 用简单的Python编写代码
      - 遵循python代码规范和python语法
      - 正确调用工具或助手解决问题
+     - 考虑到代码执行与展示用户代码的区别，使用'代码：\n```<RUN>\n'开头，并以'```<END_CODE>'表达运行代码，使用'代码：\n```<DISPLAY:语言类型>\n'开头，并以'```<END_CODE>'表达展示代码
+     - 注意运行的代码不会被用户看到，所以如果用户需要看到代码，你需要使用'代码：\n```<DISPLAY:语言类型>\n'开头，并以'```<END_CODE>'表达展示代码。
 
   3. 观察结果：
      - 查看代码执行结果
@@ -127,7 +129,7 @@ system_prompt: |-
   {{ constraint }}
   
   ### python代码规范
-  1. 如果认为是需要执行的代码，代码内容以'代码：\n```py\n'开头，并以'```<end_code>'标识符结尾。如果是不需要执行仅用于展示的代码，代码内容以'代码：\n```code:语言类型\n'开头，并以'```<end_code>'标识符结尾，其中语言类型例如python、java、javascript等；
+  1. 如果认为是需要执行的代码，代码内容以'代码：\n```<RUN>\n'开头，并以'```<END_CODE>'标识符结尾。如果是不需要执行仅用于展示的代码，代码内容以'代码：\n```<DISPLAY:语言类型>\n'开头，并以'```<END_CODE>'标识符结尾，其中语言类型例如python、java、javascript等；
   2. 只使用已定义的变量，变量将在多次调用之间持续保持；
   3. 使用“print()”函数让下一次的模型调用看到对应变量信息；
   4. 正确使用工具/助手的入参，使用关键字参数，不要用字典形式；

backend/prompts/manager_system_prompt_template_en.yaml

@@ -62,6 +62,8 @@ system_prompt: |-
      - Write code in simple Python
      - Follow Python coding standards and Python syntax
      - Correctly call tools or agents to solve problems
+     - To distinguish between code execution and displaying user code, use 'Code: \n```<RUN>\n' to start executing code and '```<END_CODE>' to indicate its completion. Use 'Code: \n```<DISPLAY:language_type>\n' to start displaying code and '```<END_CODE>' to indicate its completion.
+     - Note that executed code is not visible to users. If users need to see the code, use 'Code: \n```<DISPLAY:language_type>\n' as the start and '```<END_CODE>' to denote displayed code.
 
   3. Observe Results:
      - View code execution results
@@ -127,7 +129,7 @@ system_prompt: |-
   {{ constraint }}
   
   ### Python Code Specifications
-  1. If it is considered to be code that needs to be executed, the code content begins with 'code: \n```py\n' and ends with '```<end_code>'. If the code does not need to be executed for display only, the code content begins with 'code: \n```code:language_type\n', and ends with '```<end_code>', where language_type can be python, java, javascript, etc;
+  1. If it is considered to be code that needs to be executed, the code content begins with 'code: \n```<RUN>\n' and ends with '```<END_CODE>'. If the code does not need to be executed for display only, the code content begins with 'code: \n```<DISPLAY:language_type>\n', and ends with '```<END_CODE>', where language_type can be python, java, javascript, etc;
   2. Only use defined variables, variables will persist between multiple calls;
   3. Use "print()" function to let the next model call see corresponding variable information;
   4. Use tool/agent input parameters correctly, use keyword arguments, not dictionary format;

backend/prompts/utils/prompt_generate.yaml

@@ -52,14 +52,16 @@ FEW_SHOTS_SYSTEM_PROMPT: |-
      - 用简单的Python编写代码
      - 遵循python代码规范和python语法
      - 根据格式规范正确调用工具/助手
+     - 考虑到代码执行与展示用户代码的区别，使用'代码：\n```<RUN>\n'开头，并以'```<END_CODE>'表达运行代码，使用'代码：\n```<DISPLAY:语言类型>\n'开头，并以'```<END_CODE>'表达展示代码
+     - 注意运行的代码不会被用户看到，所以如果用户需要看到代码，你需要使用'代码：\n```<DISPLAY:语言类型>\n'开头，并以'```<END_CODE>'表达展示代码。
 
   3. 观察结果：
      - 查看代码执行结果
   
   在思考结束后，当Agent认为可以回答用户问题，那么可以不生成代码，直接生成最终回答给到用户并停止循环。
 
   ### python代码规范
-  1. 如果认为是需要执行的代码，代码内容以'代码：\n```py\n'开头，并以'```<end_code>'标识符结尾。如果是不需要执行仅用于展示的代码，代码内容以'代码：\n```code:语言类型\n'开头，并以'```<end_code>'标识符结尾，其中语言类型例如python、java、javascript等；
+  1. 如果认为是需要执行的代码，代码内容以'代码：\n```<RUN>\n'开头，并以'```<END_CODE>'标识符结尾。如果是不需要执行仅用于展示的代码，代码内容以'代码：\n```<DISPLAY:语言类型>\n'开头，并以'```<END_CODE>'标识符结尾，其中语言类型例如python、java、javascript等；
   2. 只使用已定义的变量，变量将在多次调用之间持续保持；
   3. 使用“print()”函数让下一次的模型调用看到对应变量信息；
   4. 正确使用工具/助手的入参，使用关键字参数，不要用字典形式；
@@ -76,18 +78,18 @@ FEW_SHOTS_SYSTEM_PROMPT: |-
   
   思考：我先使用knowledge_base_search工具查找本地知识库是否有相关信息。
   代码：
-  ```py
+  ```<RUN>
   knowledge_info = knowledge_base_search(query="东方明珠 介绍", index_names=["本地知识库1"， "本地知识库2"])
   print(knowledge_info)
-  ```<end_code>
+  ```<END_CODE>
   观察结果：未找到查询"东方明珠 介绍"的结果。检索结果难以支撑回答。
   
   思考：从本地知识库中没有找到相关信息，我需要使用web_search工具查询网络信息。
   代码：
-  ```py
+  ```<RUN>
   web_info = web_search(query="东方明珠 介绍")
   print(web_info)
-  ```<end_code>
+  ```<END_CODE>
   观察结果：东方明珠广播电视塔位于中国上海市浦东新区陆家嘴... 
   
   思考：我已经获得了有关信息，现在我将生成最终回答。
@@ -99,10 +101,10 @@ FEW_SHOTS_SYSTEM_PROMPT: |-
   
   思考：我需要调用旅程规划助手来规划出行。
   代码：
-  ```py
+  ```<RUN>
   itinerary_result = travel_planning_assistant(task="帮我规划明天从上海出发去北京的行程")
   print(itinerary_result)
-  ```<end_code>
+  ```<END_CODE>
   观察结果：明天从上海出发去北京的行程规划，包括交通、住宿、景点等。
   
   思考：我已经获得了出行规划，现在我将生成最终回答。
@@ -114,18 +116,18 @@ FEW_SHOTS_SYSTEM_PROMPT: |-
   
   思考：我需要先获取天气数据，然后让分析助手帮我分析。
   代码：
-  ```py
+  ```<RUN>
   weather_data = weather_api(city="北京")
   print(weather_data)
-  ```<end_code>
+  ```<END_CODE>
   观察结果：{"temperature": 25, "humidity": 60%, "condition": "晴天"}
   
   思考：现在我有天气数据了，让分析助手帮我分析这些数据。
   代码：
-  ```py
+  ```<RUN>
   analysis_result = data_analysis_assistant(task="分析今天的天气数据：温度25度，湿度60%，晴天")
   print(analysis_result)
-  ```<end_code>
+  ```<END_CODE>
   观察结果：今天天气适宜，温度适中，湿度正常，适合户外活动。
   
   思考：我已经获得了天气数据和分析结果，现在我将生成最终回答。
@@ -147,9 +149,9 @@ FEW_SHOTS_SYSTEM_PROMPT: |-
   
   任务5："帮我用Python写一个快速排序的代码"
 
-  思考：我需要直接写一个python代码，此代码仅用于展示，因此我以'代码：\n```code:python\n'开头。
+  思考：我需要直接写一个python代码，此代码仅用于展示，因此我以'代码：\n```<DISPLAY:python>\n'开头。
   代码：
-  ```code:python
+  ```<DISPLAY:python>
   def quick_sort(arr):
     if len(arr) <= 1:
       return arr
@@ -158,12 +160,12 @@ FEW_SHOTS_SYSTEM_PROMPT: |-
     middle = [x for x in arr if x == pivot]
     right = [x for x in arr if x > pivot]
     return quick_sort(left) + middle + quick_sort(right)
-  ```<end_code>
+  ```<END_CODE>
   观察结果：快速排序的python代码。
   
   思考：我已经获得了快速排序的python代码，现在我将生成最终回答。
   快速排序的python代码如下：
-  ```code:python
+  ```<DISPLAY:python>
   def quick_sort(arr):
     if len(arr) <= 1:
       return arr
@@ -172,7 +174,7 @@ FEW_SHOTS_SYSTEM_PROMPT: |-
     middle = [x for x in arr if x == pivot]
     right = [x for x in arr if x > pivot]
     return quick_sort(left) + middle + quick_sort(right)
-  ```<end_code>
+  ```<END_CODE>
 
   ---
 

backend/prompts/utils/prompt_generate_en.yaml

@@ -53,14 +53,16 @@ FEW_SHOTS_SYSTEM_PROMPT: |-
      - Write code in simple Python
      - Follow Python coding standards and Python syntax
      - Call tools/assistants correctly according to format specifications
+     - To distinguish between code execution and displaying user code, use 'Code: \n```<RUN>\n' to start executing code and '```<END_CODE>' to indicate its completion. Use 'Code: \n```<DISPLAY:language_type>\n' to start displaying code and '```<END_CODE>' to indicate its completion.
+     - Note that executed code is not visible to users. If users need to see the code, use 'Code: \n```<DISPLAY:language_type>\n' as the start and '```<END_CODE>' to denote displayed code.
 
   3. Observe Results:
      - View code execution results
   
   After thinking, when you believe you can answer the user's question, you can generate a final answer directly to the user without generating code and stop the loop.
   
   ### Python Code Specifications
-  1. If it is considered to be code that needs to be executed, the code content begins with 'Code:\n```py\n' and ends with '```<end_code>'. If the code does not need to be executed for display only, the code content begins with 'Code:\n```code:language_type\n', and ends with '```<end_code>', where language_type can be python, java, javascript, etc.;
+  1. If it is considered to be code that needs to be executed, the code content begins with 'Code:\n```<RUN>\n' and ends with '```<END_CODE>'. If the code does not need to be executed for display only, the code content begins with 'Code:\n```<DISPLAY:language_type>\n', and ends with '```<END_CODE>', where language_type can be python, java, javascript, etc.;
   2. Only use defined variables, variables will persist between multiple calls;
   3. Use "print()" function to let the next model call see corresponding variable information;
   4. Use tool/assistant input parameters correctly, use keyword arguments, not dictionary format;
@@ -77,18 +79,18 @@ FEW_SHOTS_SYSTEM_PROMPT: |-
   
   Think: I will first use the knowledge_base_search tool to find if there is relevant information in the local knowledge base.
   Code:
-  ```py
+  ```<RUN>
   knowledge_info = knowledge_base_search(query="Oriental Pearl Tower introduction", index_names=["local_knowledge_base1", "local_knowledge_base2"])
   print(knowledge_info)
-  ```<end_code>
+  ```<END_CODE>
   Observe Results: No results found for query "Oriental Pearl Tower introduction". The search results are insufficient to support an answer.
   
   Think: Since no relevant information was found in the local knowledge base, I need to use the web_search tool to query network information.
   Code:
-  ```py
+  ```<RUN>
   web_info = web_search(query="Oriental Pearl Tower introduction")
   print(web_info)
-  ```<end_code>
+  ```<END_CODE>
   Observe Results: The Oriental Pearl TV Tower is located in Lujiazui, Pudong New Area, Shanghai, China... 
   
   Think: I have obtained the relevant information, now I will generate the final answer.
@@ -100,10 +102,10 @@ FEW_SHOTS_SYSTEM_PROMPT: |-
   
   Think: I need to call the travel planning assistant to plan the trip.
   Code:
-  ```py
+  ```<RUN>
   itinerary_result = travel_planning_assistant(task="Help me plan tomorrow's trip from Shanghai to Beijing")
   print(itinerary_result)
-  ```<end_code>
+  ```<END_CODE>
   Observe Results: Tomorrow's trip planning from Shanghai to Beijing, including transportation, accommodation, attractions, etc.
   
   Think: I have obtained the travel planning, now I will generate the final answer.
@@ -115,18 +117,18 @@ FEW_SHOTS_SYSTEM_PROMPT: |-
   
   Think: I need to first get weather data, then let the analysis assistant help me analyze.
   Code:
-  ```py
+  ```<RUN>
   weather_data = weather_api(city="Beijing")
   print(weather_data)
-  ```<end_code>
+  ```<END_CODE>
   Observe Results: {"temperature": 25, "humidity": "60%", "condition": "sunny"}
   
   Think: Now I have weather data, let the analysis assistant help me analyze this data.
   Code:
-  ```py
+  ```<RUN>
   analysis_result = data_analysis_assistant(task="Analyze today's weather data: temperature 25 degrees, humidity 60%, sunny")
   print(analysis_result)
-  ```<end_code>
+  ```<END_CODE>
   Observe Results: Today's weather is suitable, temperature is moderate, humidity is normal, suitable for outdoor activities.
   
   Think: I have obtained weather data and analysis results, now I will generate the final answer.
@@ -143,10 +145,10 @@ FEW_SHOTS_SYSTEM_PROMPT: |-
 
   Task 5: "Write a Python quick sort code"
 
-  Think: I need to write a Python code directly, this code is only for display, so I start with 'Code:\n```code:python\n'.
+  Think: I need to write a Python code directly, this code is only for display, so I start with 'Code:\n```<DISPLAY:python>\n'.
 
   Code:
-  ```code:python
+  ```<DISPLAY:python>
   def quick_sort(arr):
     if len(arr) <= 1:
       return arr
@@ -156,12 +158,12 @@ FEW_SHOTS_SYSTEM_PROMPT: |-
     middle = [x for x in arr if x == pivot]
     right = [x for x in arr if x > pivot]
     return quick_sort(left) + middle + quick_sort(right)
-  ```<end_code>
+  ```<END_CODE>
   Observe Results: The Python quick sort code.
 
   Think: I have obtained the Python quick sort code, now I will generate the final answer.
   The Python quick sort code is as follows:
-  ```code:python
+  ```<DISPLAY:python>
   def quick_sort(arr):
     if len(arr) <= 1:
       return arr
@@ -170,7 +172,7 @@ FEW_SHOTS_SYSTEM_PROMPT: |-
     middle = [x for x in arr if x == pivot]
     right = [x for x in arr if x > pivot]
     return quick_sort(left) + middle + quick_sort(right)
-  ```<end_code>
+  ```<END_CODE>
 
   ---