vibecodingmachine-core 2025.12.6-1702 → 2025.12.22-2230

package/src/llm/direct-llm-manager.cjs

@@ -5,6 +5,7 @@
 
 const https = require('https');
 const http = require('http');
+const quotaManagement = require('../quota-management');
 
 class DirectLLMManager {
   constructor(sharedProviderManager = null) {
@@ -22,7 +23,7 @@ class DirectLLMManager {
       }
     }
   }
-  
+
   /**
    * Detect and save rate limit from error message
    * @param {string} provider - Provider name
@@ -31,17 +32,17 @@ class DirectLLMManager {
    */
   detectAndSaveRateLimit(provider, model, errorMessage) {
     if (!this.providerManager) return;
-    
+
     // Check for rate limit indicators
-    const isRateLimit = errorMessage.includes('rate limit') || 
-                       errorMessage.includes('Rate limit') ||
-                       errorMessage.includes('too many requests') ||
-                       errorMessage.includes('429') ||
-                       errorMessage.includes('quota') ||
-                       errorMessage.includes('Weekly limit reached') ||
-                       errorMessage.includes('Daily limit reached') ||
-                       errorMessage.includes('limit reached');
-    
+    const isRateLimit = errorMessage.includes('rate limit') ||
+      errorMessage.includes('Rate limit') ||
+      errorMessage.includes('too many requests') ||
+      errorMessage.includes('429') ||
+      errorMessage.includes('quota') ||
+      errorMessage.includes('Weekly limit reached') ||
+      errorMessage.includes('Daily limit reached') ||
+      errorMessage.includes('limit reached');
+
     if (isRateLimit) {
       this.providerManager.markRateLimited(provider, model, errorMessage);
     }
@@ -56,10 +57,10 @@ class DirectLLMManager {
    */
   async callOllama(model, prompt, options = {}) {
     const { onChunk, onComplete, onError, temperature = 0.2 } = options;
-    
+
     return new Promise((resolve) => {
       let fullResponse = '';
-      
+
       const postData = JSON.stringify({
         model: model,
         prompt: prompt,
@@ -80,22 +81,22 @@ class DirectLLMManager {
         }
       }, (res) => {
         let buffer = '';
-        
+
         res.on('data', (chunk) => {
           buffer += chunk.toString();
           const lines = buffer.split('\n');
           buffer = lines.pop(); // Keep incomplete line in buffer
-          
+
           for (const line of lines) {
             if (!line.trim()) continue;
-            
+
             try {
               const data = JSON.parse(line);
               if (data.response) {
                 fullResponse += data.response;
                 if (onChunk) onChunk(data.response);
               }
-              
+
               if (data.done) {
                 if (onComplete) onComplete(fullResponse);
                 resolve({
@@ -110,7 +111,7 @@ class DirectLLMManager {
             }
           }
         });
-        
+
         res.on('end', () => {
           if (!fullResponse) {
             const error = 'No response received from Ollama';
@@ -140,7 +141,7 @@ class DirectLLMManager {
    */
   async callAnthropic(model, prompt, options = {}) {
     const { apiKey, onChunk, onComplete, onError, temperature = 0.2, maxTokens = 8192 } = options;
-    
+
     if (!apiKey) {
       const error = 'Anthropic API key required';
       if (onError) onError(error);
@@ -149,7 +150,7 @@ class DirectLLMManager {
 
     return new Promise((resolve) => {
       let fullResponse = '';
-      
+
       const postData = JSON.stringify({
         model: model,
         max_tokens: maxTokens,
@@ -172,21 +173,21 @@ class DirectLLMManager {
         }
       }, (res) => {
         let buffer = '';
-        
+
         res.on('data', (chunk) => {
           buffer += chunk.toString();
           const lines = buffer.split('\n');
           buffer = lines.pop();
-          
+
           for (const line of lines) {
             if (!line.trim() || !line.startsWith('data: ')) continue;
-            
+
             try {
               const jsonStr = line.slice(6); // Remove "data: " prefix
               if (jsonStr === '[DONE]') continue;
-              
+
               const data = JSON.parse(jsonStr);
-              
+
               if (data.type === 'content_block_delta' && data.delta?.text) {
                 fullResponse += data.delta.text;
                 if (onChunk) onChunk(data.delta.text);
@@ -203,7 +204,7 @@ class DirectLLMManager {
             }
           }
         });
-        
+
         res.on('end', () => {
           if (!fullResponse) {
             const error = 'No response received from Anthropic';
@@ -233,7 +234,7 @@ class DirectLLMManager {
    */
   async callGroq(model, prompt, options = {}) {
     const { apiKey, onChunk, onComplete, onError, temperature = 0.2, maxTokens = 8192 } = options;
-    
+
     if (!apiKey) {
       const error = 'Groq API key required';
       if (onError) onError(error);
@@ -242,7 +243,7 @@ class DirectLLMManager {
 
     return new Promise((resolve) => {
       let fullResponse = '';
-      
+
       const postData = JSON.stringify({
         model: model,
         messages: [
@@ -265,7 +266,7 @@ class DirectLLMManager {
       }, (res) => {
         let buffer = '';
         let statusCode = res.statusCode;
-        
+
         // Check for rate limit or error status codes
         if (statusCode === 429 || statusCode >= 400) {
           let errorBody = '';
@@ -280,15 +281,15 @@ class DirectLLMManager {
           });
           return;
         }
-        
+
         res.on('data', (chunk) => {
           buffer += chunk.toString();
           const lines = buffer.split('\n');
           buffer = lines.pop();
-          
+
           for (const line of lines) {
             if (!line.trim() || !line.startsWith('data: ')) continue;
-            
+
             try {
               const jsonStr = line.slice(6);
               if (jsonStr === '[DONE]') {
@@ -300,10 +301,10 @@ class DirectLLMManager {
                 });
                 return;
               }
-              
+
               const data = JSON.parse(jsonStr);
               const content = data.choices?.[0]?.delta?.content;
-              
+
               if (content) {
                 fullResponse += content;
                 if (onChunk) onChunk(content);
@@ -313,7 +314,7 @@ class DirectLLMManager {
             }
           }
         });
-        
+
         res.on('end', () => {
           if (fullResponse) {
             if (onComplete) onComplete(fullResponse);
@@ -348,7 +349,7 @@ class DirectLLMManager {
    */
   async callBedrock(model, prompt, options = {}) {
     const { region, accessKeyId, secretAccessKey, onChunk, onComplete, onError, temperature = 0.2, maxTokens = 8192 } = options;
-    
+
     if (!region || !accessKeyId || !secretAccessKey) {
       const error = 'AWS credentials required (region, accessKeyId, secretAccessKey)';
       if (onError) onError(error);
@@ -358,7 +359,7 @@ class DirectLLMManager {
     try {
       // Use AWS SDK v3 for Bedrock
       const { BedrockRuntimeClient, InvokeModelWithResponseStreamCommand } = require('@aws-sdk/client-bedrock-runtime');
-      
+
       const client = new BedrockRuntimeClient({
         region: region,
         credentials: {
@@ -401,14 +402,14 @@ class DirectLLMManager {
       for await (const event of response.body) {
         if (event.chunk) {
           const chunk = JSON.parse(new TextDecoder().decode(event.chunk.bytes));
-          
+
           let text = '';
           if (chunk.delta?.text) {
             text = chunk.delta.text; // Anthropic format
           } else if (chunk.generation) {
             text = chunk.generation; // Meta Llama format
           }
-          
+
           if (text) {
             fullResponse += text;
             if (onChunk) onChunk(text);
@@ -418,7 +419,7 @@ class DirectLLMManager {
 
       if (onComplete) onComplete(fullResponse);
       return { success: true, response: fullResponse, model };
-      
+
     } catch (error) {
       const errorMsg = `AWS Bedrock error: ${error.message}`;
       if (onError) onError(errorMsg);
@@ -436,32 +437,32 @@ class DirectLLMManager {
   async callClaudeCode(model, prompt, options = {}) {
     const { onChunk, onComplete, onError } = options;
     const { spawn } = require('child_process');
-    
+
     return new Promise((resolve) => {
       let fullResponse = '';
       let errorOutput = '';
-      
+
       // Call claude CLI with the prompt
       const claude = spawn('claude', ['--dangerously-skip-permissions'], {
         stdio: ['pipe', 'pipe', 'pipe']
       });
-      
+
       // Send prompt to stdin
       claude.stdin.write(prompt);
       claude.stdin.end();
-      
+
       // Capture stdout
       claude.stdout.on('data', (data) => {
         const chunk = data.toString();
         fullResponse += chunk;
         if (onChunk) onChunk(chunk);
       });
-      
+
       // Capture stderr
       claude.stderr.on('data', (data) => {
         errorOutput += data.toString();
       });
-      
+
       // Handle completion
       claude.on('close', (code) => {
         if (code === 0) {
@@ -475,7 +476,7 @@ class DirectLLMManager {
           resolve({ success: false, error });
         }
       });
-      
+
       // Handle spawn errors
       claude.on('error', (err) => {
         const error = `Failed to start Claude CLI: ${err.message}`;
@@ -493,27 +494,63 @@ class DirectLLMManager {
    * @returns {Promise<{success: boolean, response?: string, error?: string}>}
    */
   async call(config, prompt, options = {}) {
-    const { provider, model, apiKey, region, accessKeyId, secretAccessKey } = config;
-
-    switch (provider) {
-      case 'ollama':
-        return this.callOllama(model, prompt, options);
-      
-      case 'anthropic':
-        return this.callAnthropic(model, prompt, { ...options, apiKey });
-      
-      case 'groq':
-        return this.callGroq(model, prompt, { ...options, apiKey });
-      
-      case 'bedrock':
-        return this.callBedrock(model, prompt, { ...options, region, accessKeyId, secretAccessKey });
-      
-      case 'claude-code':
-        return this.callClaudeCode(model, prompt, options);
-      
-      default:
-        return { success: false, error: `Unknown provider: ${provider}` };
+    const { provider, model, apiKey, region, accessKeyId, secretAccessKey, fallbackModels = [] } = config;
+    const modelsToTry = [model, ...fallbackModels];
+    let lastError = null;
+
+    for (const currentModel of modelsToTry) {
+      if (currentModel !== model) {
+        this.logger.log(`⚠️  Quota/Limit reached for previous model, failing over to ${currentModel}...`);
+      }
+
+      const agentId = `${provider}:${currentModel}`;
+      try {
+        const quota = await quotaManagement.fetchQuotaForAgent(agentId);
+        if (quota.isExceeded()) {
+          const errorMessage = `Quota limit reached for ${currentModel}. Resets at ${quota.resetsAt ? quota.resetsAt.toLocaleString() : 'a later time'}.`;
+          lastError = { success: false, error: errorMessage };
+          continue; // Try next model
+        }
+      } catch (error) {
+        this.logger.error(`Failed to check quota for ${agentId}: ${error.message}`);
+      }
+
+      const currentConfig = { ...config, model: currentModel };
+      let result;
+
+      switch (provider) {
+        case 'ollama':
+          result = await this.callOllama(currentModel, prompt, options);
+          break;
+        case 'anthropic':
+          result = await this.callAnthropic(currentModel, prompt, { ...options, apiKey });
+          break;
+        case 'groq':
+          result = await this.callGroq(currentModel, prompt, { ...options, apiKey });
+          break;
+        case 'bedrock':
+          result = await this.callBedrock(currentModel, prompt, { ...options, region, accessKeyId, secretAccessKey });
+          break;
+        case 'claude-code':
+          result = await this.callClaudeCode(currentModel, prompt, options);
+          break;
+        default:
+          return { success: false, error: `Unknown provider: ${provider}` };
+      }
+
+      if (result.success) {
+        return result;
+      }
+
+      // If failed, check for rate limit to save it
+      this.detectAndSaveRateLimit(provider, currentModel, result.error || '');
+      lastError = result;
+
+      // If it's a "fatal" error that isn't a rate limit, we might want to stop? 
+      // But usually we want to try the next model if possible.
     }
+
+    return lastError || { success: false, error: `All models for ${provider} failed.` };
   }
 
   /**
@@ -548,20 +585,20 @@ class DirectLLMManager {
    */
   async isClaudeCodeAvailable() {
     const { spawn } = require('child_process');
-    
+
     return new Promise((resolve) => {
       const claude = spawn('claude', ['--version'], {
         stdio: ['ignore', 'pipe', 'pipe']
       });
-      
+
       claude.on('close', (code) => {
         resolve(code === 0);
       });
-      
+
       claude.on('error', () => {
         resolve(false);
       });
-      
+
       // Timeout after 2 seconds
       setTimeout(() => {
         claude.kill();
@@ -583,11 +620,11 @@ class DirectLLMManager {
         method: 'GET'
       }, (res) => {
         let data = '';
-        
+
         res.on('data', (chunk) => {
           data += chunk.toString();
         });
-        
+
         res.on('end', () => {
           try {
             const json = JSON.parse(data);

package/src/quota-management/index.js

@@ -0,0 +1,108 @@
+const sharedAuth = require('../auth/shared-auth-storage');
+const ProviderManager = require('../ide-integration/provider-manager.cjs');
+
+// In-memory cache for quota data.
+const quotaCache = new Map();
+
+/**
+ * Represents the quota details for a specific agent.
+ */
+class Quota {
+  /**
+   * @param {string} agentId - The unique identifier for the agent (e.g., 'openai:gpt-4').
+   * @param {number} limit - The total quota limit.
+   * @param {number} remaining - The remaining quota.
+   * @param {Date} resetsAt - The timestamp when the quota resets.
+   * @param {string} type - The type of quota ('global', 'rate-limit', 'infinite').
+   */
+  constructor(agentId, limit, remaining, resetsAt, type = 'rate-limit') {
+    this.agentId = agentId;
+    this.limit = limit;
+    this.remaining = remaining;
+    this.resetsAt = resetsAt;
+    this.lastUpdated = new Date();
+    this.type = type;
+  }
+
+  /**
+   * Checks if the quota has been exceeded.
+   * @returns {boolean} True if the remaining quota is zero or less.
+   */
+  isExceeded() {
+    return this.remaining <= 0;
+  }
+}
+
+/**
+ * Fetches quota information for a given agent.
+ * 
+ * @param {string} agentId - The ID of the agent to fetch quota for (e.g., 'anthropic:claude-3').
+ * @returns {Promise<Quota>} A promise that resolves to a Quota object.
+ */
+async function fetchQuotaForAgent(agentId) {
+  // 1. Handle Global Daily Iteration Quota
+  if (agentId === 'global:iterations') {
+    const quotaInfo = await sharedAuth.canRunAutoMode();
+    const today = new Date();
+    const tonight = new Date(today);
+    tonight.setHours(24, 0, 0, 0);
+
+    const quota = new Quota(
+      'global:iterations',
+      quotaInfo.maxIterations || 10,
+      quotaInfo.todayUsage !== undefined ? Math.max(0, (quotaInfo.maxIterations || 10) - quotaInfo.todayUsage) : 10,
+      tonight,
+      'global'
+    );
+    quotaCache.set(agentId, quota);
+    return quota;
+  }
+
+  // 2. Handle Local Agent (Infinite Quota)
+  if (agentId.startsWith('local-') || agentId.includes('ollama')) {
+    const quota = new Quota(agentId, Infinity, Infinity, null, 'infinite');
+    quotaCache.set(agentId, quota);
+    return quota;
+  }
+
+  // 3. Handle Provider Rate Limits via ProviderManager
+  try {
+    const providerManager = new ProviderManager();
+    const parts = agentId.split(':');
+    const provider = parts[0];
+    const model = parts[1] || provider;
+
+    const isLimited = providerManager.isRateLimited(provider, model);
+    const timeUntilReset = providerManager.getTimeUntilReset(provider, model);
+
+    const quota = new Quota(
+      agentId,
+      1, // Binary limit for rate limits (1 if available, 0 if limited)
+      isLimited ? 0 : 1,
+      timeUntilReset ? new Date(Date.now() + timeUntilReset) : null,
+      'rate-limit'
+    );
+
+    quotaCache.set(agentId, quota);
+    return quota;
+  } catch (error) {
+    console.error(`Error fetching provider quota for ${agentId}:`, error);
+    return new Quota(agentId, 1, 1, null, 'rate-limit');
+  }
+}
+
+/**
+ * Retrieves the cached quota for a given agent.
+ * 
+ * @param {string} agentId - The ID of the agent.
+ * @returns {Quota | undefined} The cached Quota object or undefined if not found.
+ */
+function getCachedQuota(agentId) {
+  return quotaCache.get(agentId);
+}
+
+module.exports = {
+  Quota,
+  fetchQuotaForAgent,
+  getCachedQuota,
+};

package/src/sync/sync-engine.js

@@ -54,17 +54,27 @@ class SyncEngine extends EventEmitter {
       // Initialize DynamoDB client
       const { DynamoDBClient } = require('@aws-sdk/client-dynamodb');
       const { DynamoDBDocumentClient } = require('@aws-sdk/lib-dynamodb');
-      
+
       const region = process.env.AWS_REGION || 'us-east-1';
       const client = new DynamoDBClient({ region });
       this.dynamoClient = DynamoDBDocumentClient.from(client);
-      
+
       // Initialize WebSocket client for real-time updates
       await this._initializeWebSocket();
-      
+
       this.emit('initialized');
       return true;
     } catch (error) {
+      // If AWS SDK is not installed in this environment (common in local dev),
+      // fall back to offline mode instead of throwing so the CLI remains usable.
+      if (error && error.code === 'MODULE_NOT_FOUND' && /@aws-sdk\//.test(error.message)) {
+        this.options.offlineMode = true;
+        this.isOnline = false;
+        this.emit('warning', { type: 'offline-fallback', message: 'AWS SDK not available, running in offline mode', error });
+        this.emit('initialized');
+        return false;
+      }
+
       this.emit('error', { type: 'initialization', error });
       throw error;
     }
@@ -177,11 +187,16 @@ class SyncEngine extends EventEmitter {
    * Fetch remote changes from DynamoDB
    */
   async _fetchRemoteChanges() {
+    if (!this.dynamoClient) {
+      this.emit('warning', { type: 'no-dynamo', message: 'DynamoDB client not initialized, skipping remote fetch' });
+      return [];
+    }
+
     const { ScanCommand } = require('@aws-sdk/lib-dynamodb');
-    
+
     const tableName = process.env.DYNAMODB_TABLE_NAME || 'vibecodingmachine-requirements';
     const lastSync = this.lastSyncTime || 0;
-    
+
     try {
       // Use Scan with filter instead of Query since we need to check all items
       // In production, consider using DynamoDB Streams for real-time updates
@@ -195,7 +210,7 @@ class SyncEngine extends EventEmitter {
           ':lastSync': lastSync
         }
       });
-      
+
       const response = await this.dynamoClient.send(command);
       return response.Items || [];
     } catch (error) {
@@ -303,10 +318,17 @@ class SyncEngine extends EventEmitter {
    * Push local changes to remote
    */
   async _pushLocalChanges(changes) {
+    if (!this.dynamoClient) {
+      this.emit('warning', { type: 'no-dynamo', message: 'DynamoDB client not initialized, queuing local changes' });
+      // Queue all changes for later push
+      this.offlineQueue.push(...changes);
+      return;
+    }
+
     const { PutCommand } = require('@aws-sdk/lib-dynamodb');
-    
+
     const tableName = process.env.DYNAMODB_TABLE_NAME || 'vibecodingmachine-requirements';
-    
+
     for (const change of changes) {
       try {
         const command = new PutCommand({
@@ -317,12 +339,12 @@ class SyncEngine extends EventEmitter {
             ...change
           }
         });
-        
+
         await this.dynamoClient.send(command);
         this.emit('local-change-pushed', change);
       } catch (error) {
         this.emit('error', { type: 'push-local', change, error });
-        
+
         // Add to offline queue if push fails
         if (!this.isOnline) {
           this.offlineQueue.push(change);