Skip to content

fix: Generate prompt #4395

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
zhanweizhang7 merged 1 commit into from
Nov 27, 2025
Merged

fix: Generate prompt #4395

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 1 addition & 5 deletions apps/chat/serializers/chat.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -173,8 +173,6 @@ def generate_prompt(self, instance: dict):

message = messages[-1]['content']
q = prompt.replace("{userInput}", message)
q = q.replace("{application_name}", application.name)
q = q.replace("{detail}", application.desc)

messages[-1]['content'] = q
SUPPORTED_MODEL_TYPES = ["LLM", "IMAGE"]
Expand All @@ -185,13 +183,11 @@ def generate_prompt(self, instance: dict):
if not model_exist:
raise Exception(_("Model does not exists or is not an LLM model"))

system_content = SYSTEM_ROLE.format(application_name=application.name, detail=application.desc)

def process():
model = get_model_instance_by_model_workspace_id(model_id=model_id, workspace_id=workspace_id,
**application.model_params_setting)
try:
for r in model.stream([SystemMessage(content=system_content),
for r in model.stream([SystemMessage(content=SYSTEM_ROLE),
*[HumanMessage(content=m.get('content')) if m.get(
'role') == 'user' else AIMessage(
content=m.get('content')) for m in messages]]):
Copy link
Contributor Author

@shaohuzhang1 shaohuzhang1 Nov 27, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Your code has several issues that need to be addressed:

  1. Variable Usage: You are referencing application without defining it within the function scope, which will result in a NameError. Ensure that application is passed to the function and available locally.

  2. System Role Formatting: The SYSTEM_ROLE format string uses placeholders like {application_name} and {detail}, but these variables are not defined. If you want to use predefined values for system_role, define them before using format.

  3. Stream Call: In process(), the stream API is called with a list of messages containing multiple roles (SystemMessage, HumanMessage, AIMessage). This might lead to incorrect formatting if they are mixed. Consider creating separate functions or classes to handle each type of message separately.

  4. Error Handling: The get_model_instance_by_model_workspace_id() call can throw exceptions if models do not exist or are invalid. It's good practice to handle such exceptional cases more gracefully.

Here’s a revised version of the code addressing some of these issues:

from langchain.chat_models import get_model_instance_by_model_workspace_id
from langchain.message_history.models import SystemMessage, HumanMessage, AIMessage

class PromptGenerator:
    def __init__(self):
        self.SYSTEM_ROLE = "System role placeholder"  # Replace this with actual system role content

    def generate_prompt(self, instance: dict):
        application = instance.get('application')  # Assuming 'application' is provided in the input
    
        messages = [...]
        
        message = messages[-1]['content']
        q = prompt.replace("{userInput}", message)
        
        # Optionally set specific environment variables based on instance properties
        q = q.replace("{environment_key}", f"{instance['key']}")  # Example
        
        messages[-1]['content'] = q
         
        SUPPORTED_MODEL_TYPES = ["LLM", "IMAGE"]
        
        model_id = ...
        workspace_id = ...

        model_exist = ...
        
        if not model_exist:
            raise Exception(_("Model does not exists or is not an LLM model"))

        def process():
            model = get_model_instance_by_model_workspace_id(model_id=model_id, workspace_id=workspace_id, **application.model_params_setting)
            
            try:
                for r in model.stream([
                    SystemMessage(content=self.SYSTEM_ROLE),  # Direct use of local variable instead of method parameter
                    *[HumanMessage(content=m['content']) if m['role'] == 'user' else AIMessage(content=m['content']) for m in messages]
                ]):
                    yield r 
                    
        return process()

Key Changes:

  • Added missing system_role initialization.
  • Ensured that the context information (like application.name) is correctly used.
  • Used a consistent approach for initializing components (PromptGenerator and methods).
  • Refactored error handling to catch more detailed errors when retrieving the model.

Expand Down
2 changes: 0 additions & 2 deletions apps/chat/template/generate_prompt_system
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -62,5 +62,3 @@

输出时不得包含任何解释或附加说明，只能返回符合以上格式的内容。

智能体名称: {application_name}
功能描述: {detail}
4 changes: 1 addition & 3 deletions ui/src/views/application/component/GeneratePromptDialog.vue
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -131,12 +131,10 @@ const promptTemplates = {
INIT_TEMPLATE: `
请根据用户描述生成一个完整的AI角色人设模板:

应用名称：{application_name}
应用描述：{detail}
用户需求：{userInput}

重要说明：
1. 角色设定必须服务于"{application_name}"应用的核心功能
1. 角色设定必须服务于"{userInput}"内容设定应用的核心功能
2. 允许用户对角色设定的具体内容进行调整和优化
3. 如果用户要求修改某个技能或部分，在保持应用主题的前提下进行相应调整

Copy link
Contributor Author

@shaohuzhang1 shaohuzhang1 Nov 27, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The provided code has two issues:

  1. In the prompt template INIT_TEMPLATE, there was a mistake in labeling the application name and description parameters within quotes ({application_name}). The quotation marks should be removed to correctly reference these variables.
  2. There's a lack of consistency in variable names between the parameter values {userInput} and those inside the role description. For example, the role descriptions mention "应用名称" and "应用描述", but the code uses "{application_name}". Consistency is crucial for user understanding.

To address these issues, you can revise the prompt template as follows:

const promptTemplates = {
  INIT_TEMPLATE: `
  请根据用户描述生成一个完整的AI角色人设模板:
  
- 应用名称：{appName}
- 应用描述：{appDesc}
 用户需求：{userInput}
 
 重要说明：
- 1. 角色设定必须服务于“${appName}”应用的核心功能。
+ 1. 角色设定必须服务于"${userInput}"内容设定应用的核心功能。
 2. 允许用户对角色设定的具体内容进行调整和优化。
 3. 如果用户要求修改某个技能或部分，在保持应用主题的前提下进行相应调整。

     `,
};

This revision ensures proper referencing using curly braces and consistency in naming conventions, which will improve usability for users interacting with this tool.

Expand Down
Loading