Skip to content

feat: Improve prompt #4027

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
Sep 12, 2025
Merged

feat: Improve prompt #4027

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
13 changes: 9 additions & 4 deletions ui/src/api/application/application.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -206,16 +206,21 @@ const open: (application_id: string, loading?: Ref<boolean>) => Promise<Result<s
}

/**
* 生成优化提示词
*
* 生成提示词
* @param workspace_id
* @param model_id
* @param application_id
* @param data
* @returns
*/
const generate_prompt: (workspace_id:string ,model_id:string, data: any) => Promise<any> = (
const generate_prompt: (workspace_id:string ,model_id:string, application_id:string,data: any) => Promise<any> = (
workspace_id,
model_id,
application_id,
data
) => {
const prefix = (window.MaxKB?.prefix ? window.MaxKB?.prefix : '/admin') + '/api'
return postStream(`${prefix}/workspace/${workspace_id}/application/model/${model_id}/prompt_generate`, data)
return postStream(`${prefix}/workspace/${workspace_id}/application/${application_id}/model/${model_id}/prompt_generate`, data)
}


Copy link
Contributor Author

@shaohuzhang1 shaohuzhang1 Sep 12, 2025

Choose a reason for hiding this comment

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

The provided code snippet has a couple of issues:

  1. Incorrect URL Construction: The original line constructs the URL in the incorrect form.

    return postStream(`${prefix}/workspace/${workspace_id}/application/model/${model_id}/prompt_generate`, data)

    Should be:

    return postStream(`${prefix}/workspace/${workspace_id}/application/${application_id}/model/${model_id}/prompt_generate`, data)

  2. Missing Return Type Annotation for generate_prompt: Although typescript annotations are optional, it's considered good practice to include them for clarity.

  3. Unnecessary Use of Window Object: Assuming window.MaxKB?.prefix is defined elsewhere, using WindowGlobalScope might be more appropriate depending on how you want this logic scoped.

Here is corrected and improved version:

import { Ref } from 'vue';
import { Result } from './your_module_results'; // Adjust with actual import path

interface MaxKB {
  prefix?: string;
}

/**
 * Creates an open promise based on an application ID.
 */
export const open: (application_id: string, loading?: Ref<boolean>) => Promise<Result<string>> = async(application_id, loading): Promise<Result<string>> => {
  if (!loading) throw new Error('Loading must be supplied');
  
  const prefix = (globalThis as Window & typeof global).MaxKB.prefix ?? '/admin';
  try {
    if (loading) loading.value = true; // Assuming loading.value is reactive
  
    const response = await fetch(`${prefix}/admin/api/workspace/:${id}/app/apply_model/promptGenert`, {
      method: 'POST',
      headers: {
        'Content-Type': 'application/json',
      },
      body: JSON.stringify(data),
    });
    
    if (!response.ok) {
      console.error(`HTTP error! status: ${response.status}`);
      throw new Error(response.statusText);
    }
    
    const result = await response.json();
    if (!result.success) {
      console.warn(result.message || 'Failed to generate prompt.');
      throw new Error('Prompt generation failed.');
    }

    return result.data;
    
  } catch (err) {
    console.error(err);
    throw err;
  } finally {
    if (loading) loading.value = false;
  }
};

/**
 * Generates optimization prompts.
 * @param workspaceId
 * @param modelId
 * @param appId
 * @param data Any additional request parameters or payload data
 * @returns Response indicating success or failure along with generated prompt data.
 */
async function generate_prompt(workspaceId: string, modelId: string, appId: string, data: any) : Promise<any> {
  const prefix = (globalThis as Window & typeof global).MaxKB.prefix ?? '/admin'
  
  try {
    let res = await fetch(`${prefix}/workspace/${workspaceId}/application/${appId}/model/${modelId}/prompt_generate`, {
      method: "post",
      headers: {
        'Content-Type': 'application/json',
      },
      body: JSON.stringify(data)
    });

  if(!res.ok){
    console.error(`HTTP error! status: ${res.status}`);}
        
    const jsonResult = await res.json()
    
    if(jsonResult && !jsonResult.success){
       console.warn(jsonResult.message || 'Failed to generate prompt.')
       throw new Error('Prompt generation had a problem.')
    }
      
    return jsonResult.promptData;
   
  }catch(errorMessage){
    console.log(errorMessage);

    throw errorMessage;}
}

Note that I also used fetch() instead of postStream() which is a hypothetical utility function you may need to define to match your implementation details. Also ensure all imports like Ref, Result, etc., should be correctly resolved by their respective module names.

Expand Down
2 changes: 1 addition & 1 deletion ui/src/views/application/ApplicationSetting.vue
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -784,7 +784,7 @@ const openAIParamSettingDialog = () => {

const openGeneratePromptDialog = () => {
if (applicationForm.value.model_id) {
GeneratePromptDialogRef.value?.open(applicationForm.value.model_id)
GeneratePromptDialogRef.value?.open(applicationForm.value.model_id, id)
}
}

Expand Down
19 changes: 12 additions & 7 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 @@ -85,6 +85,7 @@ const chatMessages = ref<Array<any>>([])
// 原始输入
const originalUserInput = ref<string>('')
const modelID = ref('')
const applicationID = ref('')
const dialogVisible = ref(false)
const inputValue = ref<string>('')
const loading = ref<boolean>(false)
Expand All @@ -97,8 +98,8 @@ const promptTemplates = {

请按以下格式生成：

# 角色: 角色名称
角色概述和主要职责的一句话描述
# 角色:


## 目标：
角色的工作目标,如果有多目标可以分点列出,但建议更聚焦1-2个目标
Expand All @@ -118,9 +119,12 @@ const promptTemplates = {


## 限制：
描述角色在互动过程中需要遵循的限制条件1
描述角色在互动过程中需要遵循的限制条件2
描述角色在互动过程中需要遵循的限制条件3
1. **严格限制回答范围**：仅回答与角色设定相关的问题。
- 如果用户提问与角色无关，必须使用以下固定格式回复：
“对不起，我只能回答与【角色设定】相关的问题，您的问题不在服务范围内。”
- 不得提供任何与角色设定无关的回答。
2. 描述角色在互动过程中需要遵循的限制条件2
3. 描述角色在互动过程中需要遵循的限制条件3
`,
}

Expand Down Expand Up @@ -200,7 +204,7 @@ function generatePrompt(inputValue: any) {
messages: chatMessages.value,
prompt: promptTemplates.INIT_TEMPLATE,
}
generatePromptAPI.generate_prompt(workspaceId, modelID.value, requestData).then((response) => {
generatePromptAPI.generate_prompt(workspaceId, modelID.value, applicationID.value,requestData).then((response) => {
const reader = response.body.getReader()
reader.read().then(getWrite(reader))
})
Expand All @@ -226,8 +230,9 @@ const stopChat = () => {
chatMessages.value = []
}

const open = (modelId: string) => {
const open = (modelId: string, applicationId: string) => {
modelID.value = modelId
applicationID.value = applicationId
dialogVisible.value = true
originalUserInput.value = ''
chatMessages.value = []
Copy link
Contributor Author

@shaohuzhang1 shaohuzhang1 Sep 12, 2025

Choose a reason for hiding this comment

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

There are no identified irregularities, potential issues, or optimization suggestions in the provided code snippet. The changes introduced to include applicationID and ensure it is used appropriately within the generatePrompt function are minor additions that do not alter the overall functionality of the component.

Expand Down
2 changes: 1 addition & 1 deletion ui/src/workflow/nodes/ai-chat-node/index.vue
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -439,7 +439,7 @@ const openAIParamSettingDialog = (modelId: string) => {
const GeneratePromptDialogRef = ref<InstanceType<typeof GeneratePromptDialog>>()
const openGeneratePromptDialog = (modelId: string) => {
if (modelId) {
GeneratePromptDialogRef.value?.open(modelId)
GeneratePromptDialogRef.value?.open(modelId, id)
}
}
const replace = (v: any) => {
Expand Down
4 changes: 2 additions & 2 deletions ui/src/workflow/nodes/intent-classify-node/index.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,8 +6,8 @@ class IntentNode extends AppNode {
}
}

const get_up_index_height = (branch_lsit: Array<any>, index: number) => {
return branch_lsit
const get_up_index_height = (branch_list: Array<any>, index: number) => {
return branch_list
.filter((item, i) => i < index)
.map((item) => item.height + 8)
.reduce((x,y) => x+y, 0)
Copy link
Contributor Author

@shaohuzhang1 shaohuzhang1 Sep 12, 2025

Choose a reason for hiding this comment

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

The function get_up_index_height looks correct for calculating the height of ancestors up to a given index in an array. However, there is a minor typo that can be corrected:

-const get_up_index_height = (branch_lsit: Array<any>, index: number) => {
+const get_up_index_height = (branch_list: Array<any>, index: number) => {

This should have no effect on its functionality.

Optimization Suggestions: There isn't much that could significantly optimize this function in terms of performance with the current implementation. The operations involved—filtering, mapping, and reducing—are already quite efficient for small arrays. If you anticipate dealing with very large datasets, consider parallelizing some parts of the code if it's applicable, although in most cases JavaScript engines will handle such tasks internally efficiently. Ensure that all other functions used within get_up_index_height also perform well and avoid unnecessary computations.

Expand Down
Loading