@cloudcli-ai/cloudcli 1.32.0 → 1.33.1

package/server/claude-sdk.js

@@ -17,7 +17,8 @@ import crypto from 'crypto';
 import { promises as fs } from 'fs';
 import path from 'path';
 import os from 'os';
-import { CLAUDE_MODELS } from '../shared/modelConstants.js';
+import { CLAUDE_FALLBACK_MODELS } from './modules/providers/list/claude/claude-models.provider.js';
+import { providerModelsService } from './modules/providers/services/provider-models.service.js';
 import { resolveClaudeCodeExecutablePath } from './shared/claude-cli-path.js';
 import {
   createNotificationEvent,
@@ -204,7 +205,7 @@ function mapCliOptionsToSDK(options = {}) {
 
   // Map model (default to sonnet)
   // Valid models: sonnet, opus, haiku, opusplan, sonnet[1m]
-  sdkOptions.model = options.model || CLAUDE_MODELS.DEFAULT;
+  sdkOptions.model = options.model || CLAUDE_FALLBACK_MODELS.DEFAULT;
   // Model logged at query start below
 
   // Map system prompt configuration
@@ -284,43 +285,68 @@ function transformMessage(sdkMessage) {
   return sdkMessage;
 }
 
+function readNumber(value) {
+  const parsed = Number(value);
+  return Number.isFinite(parsed) ? parsed : 0;
+}
+
 /**
- * Extracts token usage from SDK result messages
- * @param {Object} resultMessage - SDK result message
+ * Extracts token usage from SDK messages.
+ * Prefers per-step `message.usage` (Claude message payload), then falls back
+ * to result-level usage/modelUsage for compatibility across SDK versions.
+ * @param {Object} sdkMessage - SDK stream message
  * @returns {Object|null} Token budget object or null
  */
-function extractTokenBudget(resultMessage) {
-  if (resultMessage.type !== 'result' || !resultMessage.modelUsage) {
+function extractTokenBudget(sdkMessage) {
+  if (!sdkMessage || typeof sdkMessage !== 'object') {
     return null;
   }
 
-  // Get the first model's usage data
-  const modelKey = Object.keys(resultMessage.modelUsage)[0];
-  const modelData = resultMessage.modelUsage[modelKey];
+  const messageUsage = sdkMessage.message?.usage || sdkMessage.usage;
+  if (messageUsage && typeof messageUsage === 'object') {
+    const inputTokens = readNumber(messageUsage.input_tokens ?? messageUsage.inputTokens);
+    const outputTokens = readNumber(messageUsage.output_tokens ?? messageUsage.outputTokens);
+    const totalUsed = inputTokens + outputTokens;
+    const contextWindow = parseInt(process.env.CONTEXT_WINDOW, 10) || 160000;
 
-  if (!modelData) {
-    return null;
+    return {
+      used: totalUsed,
+      total: contextWindow,
+      inputTokens,
+      outputTokens,
+      breakdown: {
+        input: inputTokens,
+        output: outputTokens,
+      },
+    };
   }
 
-  // Use cumulative tokens if available (tracks total for the session)
-  // Otherwise fall back to per-request tokens
-  const inputTokens = modelData.cumulativeInputTokens || modelData.inputTokens || 0;
-  const outputTokens = modelData.cumulativeOutputTokens || modelData.outputTokens || 0;
-  const cacheReadTokens = modelData.cumulativeCacheReadInputTokens || modelData.cacheReadInputTokens || 0;
-  const cacheCreationTokens = modelData.cumulativeCacheCreationInputTokens || modelData.cacheCreationInputTokens || 0;
+  if (!sdkMessage.modelUsage || typeof sdkMessage.modelUsage !== 'object') {
+    return null;
+  }
 
-  // Total used = input + output + cache tokens
-  const totalUsed = inputTokens + outputTokens + cacheReadTokens + cacheCreationTokens;
+  // Fallback for older SDK messages with only modelUsage
+  const modelKey = Object.keys(sdkMessage.modelUsage)[0];
+  const modelData = sdkMessage.modelUsage[modelKey];
 
-  // Use configured context window budget from environment (default 160000)
-  // This is the user's budget limit, not the model's context window
-  const contextWindow = parseInt(process.env.CONTEXT_WINDOW) || 160000;
+  if (!modelData || typeof modelData !== 'object') {
+    return null;
+  }
 
-  // Token calc logged via token-budget WS event
+  const inputTokens = readNumber(modelData.cumulativeInputTokens ?? modelData.inputTokens);
+  const outputTokens = readNumber(modelData.cumulativeOutputTokens ?? modelData.outputTokens);
+  const totalUsed = inputTokens + outputTokens;
+  const contextWindow = parseInt(process.env.CONTEXT_WINDOW, 10) || 160000;
 
   return {
     used: totalUsed,
-    total: contextWindow
+    total: contextWindow,
+    inputTokens,
+    outputTokens,
+    breakdown: {
+      input: inputTokens,
+      output: outputTokens,
+    },
   };
 }
 
@@ -491,8 +517,17 @@ async function queryClaudeSDK(command, options = {}, ws) {
   };
 
   try {
+    const resolvedModel = await providerModelsService.resolveResumeModel(
+      'claude',
+      sessionId,
+      options.model,
+    );
+
     // Map CLI options to SDK format
-    const sdkOptions = mapCliOptionsToSDK(options);
+    const sdkOptions = mapCliOptionsToSDK({
+      ...options,
+      model: resolvedModel || options.model,
+    });
 
     // Load MCP configuration
     const mcpServers = await loadMcpConfig(options.cwd);
@@ -674,16 +709,10 @@ async function queryClaudeSDK(command, options = {}, ws) {
         ws.send(msg);
       }
 
-      // Extract and send token budget updates from result messages
-      if (message.type === 'result') {
-        const models = Object.keys(message.modelUsage || {});
-        if (models.length > 0) {
-          // Model info available in result message
-        }
-        const tokenBudgetData = extractTokenBudget(message);
-        if (tokenBudgetData) {
-          ws.send(createNormalizedMessage({ kind: 'status', text: 'token_budget', tokenBudget: tokenBudgetData, sessionId: capturedSessionId || sessionId || null, provider: 'claude' }));
-        }
+      // Extract and send token budget updates from assistant/result usage payloads
+      const tokenBudgetData = extractTokenBudget(message);
+      if (tokenBudgetData) {
+        ws.send(createNormalizedMessage({ kind: 'status', text: 'token_budget', tokenBudget: tokenBudgetData, sessionId: capturedSessionId || sessionId || null, provider: 'claude' }));
       }
     }
 

package/server/cursor-cli.js

@@ -3,6 +3,7 @@ import crossSpawn from 'cross-spawn';
 import { notifyRunFailed, notifyRunStopped } from './services/notification-orchestrator.js';
 import { sessionsService } from './modules/providers/services/sessions.service.js';
 import { providerAuthService } from './modules/providers/services/provider-auth.service.js';
+import { providerModelsService } from './modules/providers/services/provider-models.service.js';
 import { createNormalizedMessage } from './shared/utils.js';
 
 // Use cross-spawn on Windows for better command execution
@@ -28,6 +29,7 @@ function isWorkspaceTrustPrompt(text = '') {
 async function spawnCursor(command, options = {}, ws) {
   return new Promise(async (resolve, reject) => {
     const { sessionId, projectPath, cwd, resume, toolsSettings, skipPermissions, model, sessionSummary } = options;
+    const resolvedModel = await providerModelsService.resolveResumeModel('cursor', sessionId, model);
     let capturedSessionId = sessionId; // Track session ID throughout the process
     let sessionCreatedSent = false; // Track if we've already sent session-created event
     let hasRetriedWithTrust = false;
@@ -52,9 +54,10 @@ async function spawnCursor(command, options = {}, ws) {
       // Provide a prompt (works for both new and resumed sessions)
       baseArgs.push('-p', command);
 
-      // Add model flag if specified (only meaningful for new sessions; harmless on resume)
-      if (!sessionId && model) {
-        baseArgs.push('--model', model);
+      // Model overrides are applied to both new and resumed sessions so a
+      // session-scoped change request can take effect on the next turn.
+      if (resolvedModel) {
+        baseArgs.push('--model', resolvedModel);
       }
 
       // Request streaming JSON when we are providing a prompt

package/server/gemini-cli.js

@@ -9,6 +9,7 @@ import sessionManager from './sessionManager.js';
 import GeminiResponseHandler from './gemini-response-handler.js';
 import { notifyRunFailed, notifyRunStopped } from './services/notification-orchestrator.js';
 import { providerAuthService } from './modules/providers/services/provider-auth.service.js';
+import { providerModelsService } from './modules/providers/services/provider-models.service.js';
 import { createNormalizedMessage } from './shared/utils.js';
 
 // Use cross-spawn on Windows for correct .cmd resolution (same pattern as cursor-cli.js)
@@ -120,6 +121,11 @@ async function buildGeminiProcessEnv() {
 
 async function spawnGemini(command, options = {}, ws) {
     const { sessionId, projectPath, cwd, toolsSettings, permissionMode, images, sessionSummary } = options;
+    const resolvedModel = await providerModelsService.resolveResumeModel(
+        'gemini',
+        sessionId,
+        options.model
+    );
     let capturedSessionId = sessionId; // Track session ID throughout the process
     let sessionCreatedSent = false; // Track if we've already sent session-created event
     let assistantBlocks = []; // Accumulate the full response blocks including tools
@@ -244,7 +250,7 @@ async function spawnGemini(command, options = {}, ws) {
     }
 
     // Add model for all sessions (both new and resumed)
-    let modelToUse = options.model || 'gemini-2.5-flash';
+    let modelToUse = resolvedModel || 'gemini-2.5-flash';
     args.push('--model', modelToUse);
     args.push('--output-format', 'stream-json');
 

package/server/gemini-response-handler.js

@@ -1,5 +1,32 @@
 // Gemini Response Handler - JSON Stream processing
 import { sessionsService } from './modules/providers/services/sessions.service.js';
+import { createNormalizedMessage } from './shared/utils.js';
+
+function buildGeminiTokenBudget(tokens) {
+  if (!tokens || typeof tokens !== 'object') {
+    return null;
+  }
+
+  const parsedInputTokens = Number(tokens.input);
+  const parsedOutputTokens = Number(tokens.output);
+  const inputTokens = Number.isFinite(parsedInputTokens) ? parsedInputTokens : 0;
+  const outputTokens = Number.isFinite(parsedOutputTokens) ? parsedOutputTokens : 0;
+  const parsedUsed = Number(tokens.total);
+  const used = Number.isFinite(parsedUsed) ? parsedUsed : inputTokens + outputTokens;
+  if (!Number.isFinite(used) || used <= 0) {
+    return null;
+  }
+
+  return {
+    used,
+    inputTokens,
+    outputTokens,
+    breakdown: {
+      input: inputTokens,
+      output: outputTokens,
+    },
+  };
+}
 
 class GeminiResponseHandler {
   constructor(ws, options = {}) {
@@ -60,6 +87,17 @@ class GeminiResponseHandler {
     for (const msg of normalized) {
       this.ws.send(msg);
     }
+
+    const tokenBudget = buildGeminiTokenBudget(event.tokens);
+    if (tokenBudget) {
+      this.ws.send(createNormalizedMessage({
+        kind: 'status',
+        text: 'token_budget',
+        tokenBudget,
+        sessionId: sid,
+        provider: 'gemini',
+      }));
+    }
   }
 
   forceFlush() {