Quality Dimension Generator

import { TaskAnalysis, ConversationInput } from './types.js'; /** * Task Extractor Class - Pure prompt approach * Only generates prompts for LLM to analyze tasks, makes no subjective judgments */ export class TaskExtractor { /** * Generate task analysis prompt * Let LLM analyze tasks and output structured results */ public generateTaskAnalysisPrompt(input: ConversationInput): string { const { userMessage, conversationHistory = [] } = input; const context = this.buildAnalysisContext(userMessage, conversationHistory); return `Please analyze the user's core task based on the following conversation content: Conversation Content: ${context.rawText} Please analyze according to the following dimensions and output the result in JSON format: \`\`\`json { "coreTask": "Description of the user's core task", "taskName": "Concise task name (suitable for file naming, can be in any language)", "taskType": "Task type (e.g., development, design, analysis, learning, management, consulting, etc.)", "complexity": Complexity level (1-5), "domain": "Task domain (e.g., technical development, business management, education and training, etc.)", "keyElements": ["Key element 1", "Key element 2", "Key element 3"], "objectives": ["Main objective 1", "Main objective 2", "Main objective 3"] } \`\`\` Analysis Requirements: 1. Core Task: Extract the user's main needs and expected goals 2. Task Name: Generate a concise name of 3-15 characters, suitable for file naming, avoid special symbols 3. Task Type: Classify according to the nature of the task 3. Complexity: 1=Simple 2=Fairly Simple 3=Medium 4=Fairly Complex 5=Complex 4. Domain: Identify the professional domain to which the task belongs 5. Key Elements: Identify important constraints, technical requirements, standard requirements, etc. 6. Objectives: Break down into specific, measurable goals Please ensure the analysis is accurate, specific, and practical.`; } /** * Parse LLM-returned task analysis result */ public parseTaskAnalysisResult(llmResponse: string): TaskAnalysis { // Extract JSON part const jsonMatch = llmResponse.match(/```json\s*([\s\S]*?)\s*```/); if (!jsonMatch) { throw new Error('JSON format analysis result not found'); } const result = JSON.parse(jsonMatch[1]); // Validate required fields if (!result.coreTask || !result.taskType || !result.domain || !result.taskName) { throw new Error('Analysis result missing required fields'); } return { coreTask: result.coreTask, taskName: result.taskName, taskType: result.taskType, complexity: result.complexity || 3, domain: result.domain, keyElements: Array.isArray(result.keyElements) ? result.keyElements : [], objectives: Array.isArray(result.objectives) ? result.objectives : [] }; } /** * Build analysis context */ private buildAnalysisContext(userMessage: string, conversationHistory: Array<{role: string, content: string}>) { return { rawText: [ ...conversationHistory.map(h => `${h.role}: ${h.content}`), `user: ${userMessage}` ].join('\n'), messageCount: conversationHistory.length + 1, totalLength: userMessage.length + conversationHistory.reduce((sum, h) => sum + h.content.length, 0) }; } }