s9n-devops-agent 2.0.9 → 2.0.11-dev.0

package/scripts/contract-automation/analyze-with-llm.js

@@ -0,0 +1,497 @@
+#!/usr/bin/env node
+
+/**
+ * ============================================================================
+ * LLM-POWERED CONTRACT ANALYSIS SCRIPT
+ * ============================================================================
+ * 
+ * This script uses Groq LLM to perform intelligent analysis of code and
+ * generate human-readable contract documentation.
+ * 
+ * Usage:
+ *   node scripts/contract-automation/analyze-with-llm.js --scan-results=path/to/results.json
+ *   node scripts/contract-automation/analyze-with-llm.js --analyze-file=src/features/auth/index.js
+ *   node scripts/contract-automation/analyze-with-llm.js --validate-contracts
+ * 
+ * Options:
+ *   --scan-results=<path>   Use scan results from generate-contracts.js
+ *   --analyze-file=<path>   Analyze specific file
+ *   --validate-contracts    Validate existing contracts for completeness
+ *   --model=<name>          Groq model to use (default: llama-3.1-70b-versatile)
+ *   --verbose               Detailed logging
+ * 
+ * Environment Variables:
+ *   OPENAI_API_KEY          Required - Groq API key (OpenAI-compatible endpoint)
+ * 
+ * ============================================================================
+ */
+
+import fs from 'fs';
+import path from 'path';
+import Groq from 'groq-sdk';
+import { fileURLToPath } from 'url';
+import { dirname } from 'path';
+import { credentialsManager } from '../../src/credentials-manager.js';
+
+const __filename = fileURLToPath(import.meta.url);
+const __dirname = dirname(__filename);
+
+// Attempt to load credentials manager for local key injection
+credentialsManager.injectEnv();
+
+// Configuration
+const CONFIG = {
+  rootDir: process.cwd(),
+  contractsDir: path.join(process.cwd(), 'House_Rules_Contracts'),
+  verbose: process.argv.includes('--verbose'),
+  scanResults: getArgValue('--scan-results'),
+  analyzeFile: getArgValue('--analyze-file'),
+  validateContracts: process.argv.includes('--validate-contracts'),
+  model: getArgValue('--model') || 'llama-3.1-70b-versatile'
+};
+
+// Initialize Groq client
+const groq = new Groq({
+  apiKey: process.env.GROQ_API_KEY || process.env.OPENAI_API_KEY
+});
+
+// Helper: Get command line argument value
+function getArgValue(argName) {
+  const arg = process.argv.find(a => a.startsWith(argName + '='));
+  return arg ? arg.split('=')[1] : null;
+}
+
+// Helper: Log with optional verbose mode
+function log(message, level = 'info') {
+  const prefix = {
+    info: '[INFO]',
+    warn: '[WARN]',
+    error: '[ERROR]',
+    success: '[SUCCESS]',
+    debug: '[DEBUG]'
+  }[level];
+  
+  if (level === 'debug' && !CONFIG.verbose) return;
+  
+  console.log(`${prefix} ${message}`);
+}
+
+// Helper: Read file safely
+function readFileSafe(filePath) {
+  try {
+    return fs.readFileSync(filePath, 'utf8');
+  } catch (error) {
+    log(`Failed to read ${filePath}: ${error.message}`, 'warn');
+    return '';
+  }
+}
+
+// Helper: Write JSON file
+function writeJSON(filePath, data) {
+  fs.writeFileSync(filePath, JSON.stringify(data, null, 2), 'utf8');
+}
+
+// ============================================================================
+// LLM INTERACTION
+// ============================================================================
+
+async function callLLM(prompt, systemPrompt = 'You are a helpful assistant.') {
+  try {
+    log(`Calling LLM (${CONFIG.model})...`, 'debug');
+    
+    const response = await groq.chat.completions.create({
+      model: CONFIG.model,
+      messages: [
+        { role: 'system', content: systemPrompt },
+        { role: 'user', content: prompt }
+      ],
+      temperature: 0.3, // Lower temperature for more deterministic output
+      max_tokens: 4000
+    });
+    
+    return response.choices[0].message.content;
+  } catch (error) {
+    log(`LLM call failed: ${error.message}`, 'error');
+    throw error;
+  }
+}
+
+// ============================================================================
+// FEATURE ANALYSIS
+// ============================================================================
+
+async function analyzeFeatures(features) {
+  log('Analyzing features with LLM...');
+  
+  const systemPrompt = `You are a software architect analyzing a codebase. 
+Your task is to analyze feature information and provide detailed documentation including:
+- Clear feature descriptions
+- User stories
+- Acceptance criteria
+- Dependencies
+- Priority assessment
+
+Output should be in JSON format.`;
+  
+  const prompt = `Analyze these features and provide detailed documentation for each:
+
+${JSON.stringify(features, null, 2)}
+
+For each feature, provide:
+1. A clear description (2-3 sentences)
+2. A user story in format: "As a [user type], I want to [action] so that [benefit]"
+3. 3-5 acceptance criteria
+4. Likely dependencies on other features
+5. Priority assessment (critical/high/medium/low) with reasoning
+
+Return as JSON array with same structure plus new fields: description, userStory, acceptanceCriteria, dependencies, priorityReasoning.`;
+  
+  const result = await callLLM(prompt, systemPrompt);
+  
+  try {
+    return JSON.parse(result);
+  } catch (error) {
+    log('Failed to parse LLM response as JSON, returning raw text', 'warn');
+    return { raw: result };
+  }
+}
+
+// ============================================================================
+// API ENDPOINT ANALYSIS
+// ============================================================================
+
+async function analyzeAPIEndpoints(endpoints) {
+  log('Analyzing API endpoints with LLM...');
+  
+  const systemPrompt = `You are an API documentation expert.
+Analyze API endpoints and provide comprehensive documentation including:
+- Endpoint purpose and description
+- Expected request/response formats
+- Authentication requirements
+- Error scenarios
+- Usage examples
+
+Output should be in JSON format.`;
+  
+  const prompt = `Analyze these API endpoints and provide detailed documentation:
+
+${JSON.stringify(endpoints.slice(0, 20), null, 2)}
+
+For each endpoint, infer and provide:
+1. Purpose and description
+2. Likely authentication requirement (yes/no)
+3. Expected request parameters (path, query, body)
+4. Expected response format
+5. Common error scenarios
+6. Usage example (curl command)
+
+Return as JSON array with enhanced documentation.`;
+  
+  const result = await callLLM(prompt, systemPrompt);
+  
+  try {
+    return JSON.parse(result);
+  } catch (error) {
+    log('Failed to parse LLM response as JSON, returning raw text', 'warn');
+    return { raw: result };
+  }
+}
+
+// ============================================================================
+// SQL QUERY ANALYSIS
+// ============================================================================
+
+async function analyzeSQLQueries(queries) {
+  log('Analyzing SQL queries with LLM...');
+  
+  const systemPrompt = `You are a database expert analyzing SQL queries.
+Provide insights on:
+- Query purpose and description
+- Performance considerations
+- Security concerns
+- Suggested optimizations
+- Parameter types and validation
+
+Output should be in JSON format.`;
+  
+  const queryList = Object.values(queries).slice(0, 10);
+  
+  const prompt = `Analyze these SQL queries and provide documentation:
+
+${JSON.stringify(queryList, null, 2)}
+
+For each query, provide:
+1. Clear description of what it does
+2. Performance notes (indexes needed, complexity)
+3. Security notes (SQL injection risks, data sensitivity)
+4. Parameter types and validation rules
+5. Suggested optimizations if any
+
+Return as JSON array with enhanced documentation.`;
+  
+  const result = await callLLM(prompt, systemPrompt);
+  
+  try {
+    return JSON.parse(result);
+  } catch (error) {
+    log('Failed to parse LLM response as JSON, returning raw text', 'warn');
+    return { raw: result };
+  }
+}
+
+// ============================================================================
+// THIRD-PARTY INTEGRATION ANALYSIS
+// ============================================================================
+
+async function analyzeIntegrations(integrations) {
+  log('Analyzing third-party integrations with LLM...');
+  
+  const systemPrompt = `You are a software integration expert.
+Analyze third-party service integrations and provide:
+- Service purpose and use cases
+- Best practices for integration
+- Error handling strategies
+- Security considerations
+- Cost optimization tips
+
+Output should be in JSON format.`;
+  
+  const prompt = `Analyze these third-party integrations and provide documentation:
+
+${JSON.stringify(integrations, null, 2)}
+
+For each integration, provide:
+1. Detailed purpose and use cases
+2. Recommended error handling strategy
+3. Security best practices
+4. Rate limiting considerations
+5. Cost optimization suggestions
+6. Alternative services (if applicable)
+
+Return as JSON array with enhanced documentation.`;
+  
+  const result = await callLLM(prompt, systemPrompt);
+  
+  try {
+    return JSON.parse(result);
+  } catch (error) {
+    log('Failed to parse LLM response as JSON, returning raw text', 'warn');
+    return { raw: result };
+  }
+}
+
+// ============================================================================
+// FILE ANALYSIS
+// ============================================================================
+
+async function analyzeFile(filePath) {
+  log(`Analyzing file: ${filePath}`);
+  
+  const content = readFileSafe(filePath);
+  if (!content) {
+    throw new Error(`Could not read file: ${filePath}`);
+  }
+  
+  const systemPrompt = `You are a code analysis expert.
+Analyze the provided code file and extract:
+- Purpose and functionality
+- API endpoints (if any)
+- Database queries (if any)
+- Third-party integrations (if any)
+- Environment variables used
+- Dependencies on other modules
+- Security considerations
+- Suggested improvements
+
+Provide structured analysis in JSON format.`;
+  
+  const prompt = `Analyze this code file and provide comprehensive documentation:
+
+File: ${filePath}
+
+\`\`\`
+${content.slice(0, 10000)} ${content.length > 10000 ? '... (truncated)' : ''}
+\`\`\`
+
+Provide:
+1. Module purpose and description
+2. Exported functions/classes with descriptions
+3. API endpoints defined (method, path, purpose)
+4. Database queries used (with purpose)
+5. Third-party services integrated
+6. Environment variables required
+7. Dependencies on other modules
+8. Security considerations
+9. Suggested improvements or concerns
+
+Return as structured JSON.`;
+  
+  const result = await callLLM(prompt, systemPrompt);
+  
+  try {
+    return JSON.parse(result);
+  } catch (error) {
+    log('Failed to parse LLM response as JSON, returning raw text', 'warn');
+    return { raw: result };
+  }
+}
+
+// ============================================================================
+// CONTRACT VALIDATION
+// ============================================================================
+
+async function validateContracts() {
+  log('Validating existing contracts with LLM...');
+  
+  const contracts = {
+    features: path.join(CONFIG.contractsDir, 'FEATURES_CONTRACT.md'),
+    api: path.join(CONFIG.contractsDir, 'API_CONTRACT.md'),
+    database: path.join(CONFIG.contractsDir, 'DATABASE_SCHEMA_CONTRACT.md'),
+    sql: path.join(CONFIG.contractsDir, 'SQL_CONTRACT.json'),
+    integrations: path.join(CONFIG.contractsDir, 'THIRD_PARTY_INTEGRATIONS.md'),
+    infra: path.join(CONFIG.contractsDir, 'INFRA_CONTRACT.md')
+  };
+  
+  const validation = {};
+  
+  for (const [name, filePath] of Object.entries(contracts)) {
+    if (!fs.existsSync(filePath)) {
+      validation[name] = {
+        exists: false,
+        complete: false,
+        issues: ['Contract file does not exist']
+      };
+      continue;
+    }
+    
+    const content = readFileSafe(filePath);
+    
+    const systemPrompt = `You are a documentation quality expert.
+Analyze contract documentation and identify:
+- Completeness (are all sections filled?)
+- Missing information
+- Inconsistencies
+- Suggestions for improvement
+
+Output should be in JSON format.`;
+    
+    const prompt = `Analyze this contract file for completeness and quality:
+
+File: ${name}
+
+\`\`\`
+${content.slice(0, 5000)} ${content.length > 5000 ? '... (truncated)' : ''}
+\`\`\`
+
+Provide:
+1. Is the contract complete? (true/false)
+2. Completion percentage (0-100)
+3. List of missing sections or information
+4. List of inconsistencies or errors
+5. Suggestions for improvement
+
+Return as JSON with fields: complete, completionPercentage, missingSections, inconsistencies, suggestions.`;
+    
+    const result = await callLLM(prompt, systemPrompt);
+    
+    try {
+      validation[name] = JSON.parse(result);
+      validation[name].exists = true;
+    } catch (error) {
+      validation[name] = {
+        exists: true,
+        complete: false,
+        error: 'Failed to parse validation result',
+        raw: result
+      };
+    }
+  }
+  
+  return validation;
+}
+
+// ============================================================================
+// MAIN EXECUTION
+// ============================================================================
+
+async function main() {
+  log('='.repeat(80));
+  log('LLM-POWERED CONTRACT ANALYSIS');
+  log('='.repeat(80));
+  
+  // Check API key
+  if (!process.env.GROQ_API_KEY && !process.env.OPENAI_API_KEY) {
+    throw new Error('GROQ_API_KEY environment variable is required. Please set it or run "s9n-devops-agent setup" to configure your Groq API key.');
+  }
+  
+  let results = {};
+  
+  // Analyze scan results
+  if (CONFIG.scanResults) {
+    log(`Loading scan results from: ${CONFIG.scanResults}`);
+    const scanData = JSON.parse(readFileSafe(CONFIG.scanResults));
+    
+    if (scanData.results.features) {
+      results.features = await analyzeFeatures(scanData.results.features);
+    }
+    
+    if (scanData.results.api) {
+      results.api = await analyzeAPIEndpoints(scanData.results.api);
+    }
+    
+    if (scanData.results.sql) {
+      results.sql = await analyzeSQLQueries(scanData.results.sql);
+    }
+    
+    if (scanData.results.integrations) {
+      results.integrations = await analyzeIntegrations(scanData.results.integrations);
+    }
+  }
+  
+  // Analyze specific file
+  if (CONFIG.analyzeFile) {
+    results.fileAnalysis = await analyzeFile(CONFIG.analyzeFile);
+  }
+  
+  // Validate contracts
+  if (CONFIG.validateContracts) {
+    results.validation = await validateContracts();
+  }
+  
+  // Save results
+  const outputPath = path.join(CONFIG.contractsDir, 'llm-analysis-results.json');
+  writeJSON(outputPath, {
+    generated: new Date().toISOString(),
+    model: CONFIG.model,
+    results
+  });
+  
+  log(`Results saved to: ${outputPath}`, 'success');
+  
+  // Summary
+  log('='.repeat(80));
+  log('ANALYSIS COMPLETE', 'success');
+  log('='.repeat(80));
+  
+  if (results.validation) {
+    log('Contract Validation Results:');
+    for (const [name, result] of Object.entries(results.validation)) {
+      if (result.exists) {
+        const status = result.complete ? '✅' : '⚠️';
+        const pct = result.completionPercentage || 0;
+        log(`  ${status} ${name}: ${pct}% complete`);
+      } else {
+        log(`  ❌ ${name}: Does not exist`);
+      }
+    }
+  }
+  
+  log('='.repeat(80));
+}
+
+// Run
+main().catch(error => {
+  log(`Fatal error: ${error.message}`, 'error');
+  console.error(error);
+  process.exit(1);
+});