i18n: fix missing translations for /clear command

feat(i18n): update Russian translation with new strings

docs/users/features/_meta.ts

@@ -9,4 +9,5 @@ export default {
   mcp: 'MCP',
   'token-caching': 'Token Caching',
   sandbox: 'Sandboxing',
+  language: 'i18n',
 };

docs/users/features/commands.md

@@ -48,7 +48,7 @@ Commands specifically for controlling interface and output language.
 | → `ui [language]`     | Set UI interface language        | `/language ui zh-CN`       |
 | → `output [language]` | Set LLM output language          | `/language output Chinese` |
 
-- Available UI languages: `zh-CN` (Simplified Chinese), `en-US` (English)
+- Available built-in UI languages: `zh-CN` (Simplified Chinese), `en-US` (English), `ru-RU` (Russian), `de-DE` (German)
 - Output language examples: `Chinese`, `English`, `Japanese`, etc.
 
 ### 1.4 Tool and Model Management
@@ -72,17 +72,16 @@ Commands for managing AI tools and models.
 
 Commands for obtaining information and performing system settings.
 
-| Command         | Description                                     | Usage Examples                                   |
-| --------------- | ----------------------------------------------- | ------------------------------------------------ |
-| `/help`         | Display help information for available commands | `/help` or `/?`                                  |
-| `/about`        | Display version information                     | `/about`                                         |
-| `/stats`        | Display detailed statistics for current session | `/stats`                                         |
-| `/settings`     | Open settings editor                            | `/settings`                                      |
-| `/auth`         | Change authentication method                    | `/auth`                                          |
-| `/bug`          | Submit issue about Qwen Code                    | `/bug Button click unresponsive`                 |
-| `/copy`         | Copy last output content to clipboard           | `/copy`                                          |
-| `/quit-confirm` | Show confirmation dialog before quitting        | `/quit-confirm` (shortcut: press `Ctrl+C` twice) |
-| `/quit`         | Exit Qwen Code immediately                      | `/quit` or `/exit`                               |
+| Command     | Description                                     | Usage Examples                   |
+| ----------- | ----------------------------------------------- | -------------------------------- |
+| `/help`     | Display help information for available commands | `/help` or `/?`                  |
+| `/about`    | Display version information                     | `/about`                         |
+| `/stats`    | Display detailed statistics for current session | `/stats`                         |
+| `/settings` | Open settings editor                            | `/settings`                      |
+| `/auth`     | Change authentication method                    | `/auth`                          |
+| `/bug`      | Submit issue about Qwen Code                    | `/bug Button click unresponsive` |
+| `/copy`     | Copy last output content to clipboard           | `/copy`                          |
+| `/quit`     | Exit Qwen Code immediately                      | `/quit` or `/exit`               |
 
 ### 1.6 Common Shortcuts
 

docs/users/features/language.md

@@ -0,0 +1,136 @@
+# Internationalization (i18n) & Language
+
+Qwen Code is built for multilingual workflows: it supports UI localization (i18n/l10n) in the CLI, lets you choose the assistant output language, and allows custom UI language packs.
+
+## Overview
+
+From a user point of view, Qwen Code’s “internationalization” spans multiple layers:
+
+| Capability / Setting     | What it controls                                                       | Where stored                 |
+| ------------------------ | ---------------------------------------------------------------------- | ---------------------------- |
+| `/language ui`           | Terminal UI text (menus, system messages, prompts)                     | `~/.qwen/settings.json`      |
+| `/language output`       | Language the AI responds in (an output preference, not UI translation) | `~/.qwen/output-language.md` |
+| Custom UI language packs | Overrides/extends built-in UI translations                             | `~/.qwen/locales/*.js`       |
+
+## UI Language
+
+This is the CLI’s UI localization layer (i18n/l10n): it controls the language of menus, prompts, and system messages.
+
+### Setting the UI Language
+
+Use the `/language ui` command:
+
+```bash
+/language ui zh-CN    # Chinese
+/language ui en-US    # English
+/language ui ru-RU    # Russian
+/language ui de-DE    # German
+```
+
+Aliases are also supported:
+
+```bash
+/language ui zh       # Chinese
+/language ui en       # English
+/language ui ru       # Russian
+/language ui de       # German
+```
+
+### Auto-detection
+
+On first startup, Qwen Code detects your system locale and sets the UI language automatically.
+
+Detection priority:
+
+1. `QWEN_CODE_LANG` environment variable
+2. `LANG` environment variable
+3. System locale via JavaScript Intl API
+4. Default: English
+
+## LLM Output Language
+
+The LLM output language controls what language the AI assistant responds in, regardless of what language you type your questions in.
+
+### How It Works
+
+The LLM output language is controlled by a rule file at `~/.qwen/output-language.md`. This file is automatically included in the LLM's context during startup, instructing it to respond in the specified language.
+
+### Auto-detection
+
+On first startup, if no `output-language.md` file exists, Qwen Code automatically creates one based on your system locale. For example:
+
+- System locale `zh` creates a rule for Chinese responses
+- System locale `en` creates a rule for English responses
+- System locale `ru` creates a rule for Russian responses
+- System locale `de` creates a rule for German responses
+
+### Manual Setting
+
+Use `/language output <language>` to change:
+
+```bash
+/language output Chinese
+/language output English
+/language output Japanese
+/language output German
+```
+
+Any language name works. The LLM will be instructed to respond in that language.
+
+> [!note]
+>
+> After changing the output language, restart Qwen Code for the change to take effect.
+
+### File Location
+
+```
+~/.qwen/output-language.md
+```
+
+## Configuration
+
+### Via Settings Dialog
+
+1. Run `/settings`
+2. Find "Language" under General
+3. Select your preferred UI language
+
+### Via Environment Variable
+
+```bash
+export QWEN_CODE_LANG=zh
+```
+
+This influences auto-detection on first startup (if you haven’t set a UI language and no `output-language.md` file exists yet).
+
+## Custom Language Packs
+
+For UI translations, you can create custom language packs in `~/.qwen/locales/`:
+
+- Example: `~/.qwen/locales/es.js` for Spanish
+- Example: `~/.qwen/locales/fr.js` for French
+
+User directory takes precedence over built-in translations.
+
+> [!tip]
+>
+> Contributions are welcome! If you’d like to improve built-in translations or add new languages.
+> For a concrete example, see [PR #1238: feat(i18n): add Russian language support](https://github.com/QwenLM/qwen-code/pull/1238).
+
+### Language Pack Format
+
+```javascript
+// ~/.qwen/locales/es.js
+export default {
+  Hello: 'Hola',
+  Settings: 'Configuracion',
+  // ... more translations
+};
+```
+
+## Related Commands
+
+- `/language` - Show current language settings
+- `/language ui [lang]` - Set UI language
+- `/language output <language>` - Set LLM output language
+- `/settings` - Open settings dialog

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@qwen-code/qwen-code",
-  "version": "0.6.0-preview.0",
+  "version": "0.6.0",
   "engines": {
     "node": ">=20.0.0"
   },
@@ -13,11 +13,14 @@
     "url": "git+https://github.com/QwenLM/qwen-code.git"
   },
   "config": {
-    "sandboxImageUri": "ghcr.io/qwenlm/qwen-code:0.6.0-preview.0"
+    "sandboxImageUri": "ghcr.io/qwenlm/qwen-code:0.6.0"
   },
   "scripts": {
     "start": "cross-env node scripts/start.js",
     "debug": "cross-env DEBUG=1 node --inspect-brk scripts/start.js",
+    "auth:npm": "npx google-artifactregistry-auth",
+    "auth:docker": "gcloud auth configure-docker us-west1-docker.pkg.dev",
+    "auth": "npm run auth:npm && npm run auth:docker",
     "generate": "node scripts/generate-git-commit-info.js",
     "build": "node scripts/build.js",
     "build-and-start": "npm run build && npm run start",
@@ -92,6 +95,7 @@
     "eslint-plugin-react-hooks": "^5.2.0",
     "glob": "^10.5.0",
     "globals": "^16.0.0",
+    "google-artifactregistry-auth": "^3.4.0",
     "husky": "^9.1.7",
     "json": "^11.0.0",
     "lint-staged": "^16.1.6",

packages/cli/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@qwen-code/qwen-code",
-  "version": "0.6.0-preview.0",
+  "version": "0.6.0",
   "description": "Qwen Code",
   "repository": {
     "type": "git",
@@ -33,13 +33,13 @@
     "dist"
   ],
   "config": {
-    "sandboxImageUri": "ghcr.io/qwenlm/qwen-code:0.6.0-preview.0"
+    "sandboxImageUri": "ghcr.io/qwenlm/qwen-code:0.6.0"
   },
   "dependencies": {
-    "@google/genai": "1.30.0",
+    "@google/genai": "1.16.0",
     "@iarna/toml": "^2.2.5",
     "@qwen-code/qwen-code-core": "file:../core",
-    "@modelcontextprotocol/sdk": "^1.25.1",
+    "@modelcontextprotocol/sdk": "^1.15.1",
     "@types/update-notifier": "^6.0.8",
     "ansi-regex": "^6.2.2",
     "command-exists": "^1.2.9",

packages/cli/src/config/auth.ts

@@ -26,23 +26,5 @@ export function validateAuthMethod(authMethod: string): string | null {
     return null;
   }
 
-  if (authMethod === AuthType.USE_GEMINI) {
-    const hasApiKey = process.env['GEMINI_API_KEY'];
-    if (!hasApiKey) {
-      return 'GEMINI_API_KEY environment variable not found. Please set it in your .env file or environment variables.';
-    }
-    return null;
-  }
-
-  if (authMethod === AuthType.USE_VERTEX_AI) {
-    const hasApiKey = process.env['GOOGLE_API_KEY'];
-    if (!hasApiKey) {
-      return 'GOOGLE_API_KEY environment variable not found. Please set it in your .env file or environment variables.';
-    }
-
-    process.env['GOOGLE_GENAI_USE_VERTEXAI'] = 'true';
-    return null;
-  }
-
   return 'Invalid auth method selected.';
 }

packages/cli/src/config/config.ts

@@ -460,12 +460,7 @@ export async function parseArguments(settings: Settings): Promise<CliArgs> {
         })
         .option('auth-type', {
           type: 'string',
-          choices: [
-            AuthType.USE_OPENAI,
-            AuthType.QWEN_OAUTH,
-            AuthType.USE_GEMINI,
-            AuthType.USE_VERTEX_AI,
-          ],
+          choices: [AuthType.USE_OPENAI, AuthType.QWEN_OAUTH],
           description: 'Authentication type',
         })
         .deprecateOption(

packages/cli/src/config/extension.test.ts

@@ -56,17 +56,6 @@ vi.mock('simple-git', () => ({
   }),
 }));
 
-vi.mock('./extensions/github.js', async (importOriginal) => {
-  const actual =
-    await importOriginal<typeof import('./extensions/github.js')>();
-  return {
-    ...actual,
-    downloadFromGitHubRelease: vi
-      .fn()
-      .mockRejectedValue(new Error('Mocked GitHub release download failure')),
-  };
-});
-
 vi.mock('os', async (importOriginal) => {
   const mockedOs = await importOriginal<typeof os>();
   return {

packages/cli/src/config/extensions/update.test.ts

@@ -41,17 +41,6 @@ vi.mock('simple-git', () => ({
   }),
 }));
 
-vi.mock('../extensions/github.js', async (importOriginal) => {
-  const actual =
-    await importOriginal<typeof import('../extensions/github.js')>();
-  return {
-    ...actual,
-    downloadFromGitHubRelease: vi
-      .fn()
-      .mockRejectedValue(new Error('Mocked GitHub release download failure')),
-  };
-});
-
 vi.mock('os', async (importOriginal) => {
   const mockedOs = await importOriginal<typeof os>();
   return {

packages/cli/src/core/initializer.ts

@@ -15,6 +15,7 @@ import { type LoadedSettings, SettingScope } from '../config/settings.js';
 import { performInitialAuth } from './auth.js';
 import { validateTheme } from './theme.js';
 import { initializeI18n } from '../i18n/index.js';
+import { initializeLlmOutputLanguage } from '../ui/commands/languageCommand.js';
 
 export interface InitializationResult {
   authError: string | null;
@@ -41,6 +42,9 @@ export async function initializeApp(
     'auto';
   await initializeI18n(languageSetting);
 
+  // Auto-detect and set LLM output language on first use
+  initializeLlmOutputLanguage();
+
   const authType = settings.merged.security?.auth?.selectedType;
   const authError = await performInitialAuth(config, authType);
 

packages/cli/src/gemini.tsx

@@ -4,8 +4,13 @@
  * SPDX-License-Identifier: Apache-2.0
  */
 
-import type { Config, AuthType } from '@qwen-code/qwen-code-core';
-import { InputFormat, logUserPrompt } from '@qwen-code/qwen-code-core';
+import type { Config } from '@qwen-code/qwen-code-core';
+import {
+  AuthType,
+  getOauthClient,
+  InputFormat,
+  logUserPrompt,
+} from '@qwen-code/qwen-code-core';
 import { render } from 'ink';
 import dns from 'node:dns';
 import os from 'node:os';
@@ -394,6 +399,15 @@ export async function main() {
       initializationResult = await initializeApp(config, settings);
     }
 
+    if (
+      settings.merged.security?.auth?.selectedType ===
+        AuthType.LOGIN_WITH_GOOGLE &&
+      config.isBrowserLaunchSuppressed()
+    ) {
+      // Do oauth before app renders to make copying the link possible.
+      await getOauthClient(settings.merged.security.auth.selectedType, config);
+    }
+
     if (config.getExperimentalZedIntegration()) {
       return runAcpAgent(config, settings, extensions, argv);
     }

packages/cli/src/i18n/index.ts

@@ -1,6 +1,6 @@
 /**
  * @license
- * Copyright 2025 Qwen
+ * Copyright 2025 Qwen team
  * SPDX-License-Identifier: Apache-2.0
  */
 
@@ -8,15 +8,21 @@ import * as fs from 'node:fs';
 import * as path from 'node:path';
 import { fileURLToPath, pathToFileURL } from 'node:url';
 import { homedir } from 'node:os';
+import {
+  type SupportedLanguage,
+  getLanguageNameFromLocale,
+} from './languages.js';
 
-export type SupportedLanguage = 'en' | 'zh' | 'ru' | string; // Allow custom language codes
+export type { SupportedLanguage };
+export { getLanguageNameFromLocale };
 
 // State
 let currentLanguage: SupportedLanguage = 'en';
-let translations: Record<string, string> = {};
+let translations: Record<string, string | string[]> = {};
 
 // Cache
-type TranslationDict = Record<string, string>;
+type TranslationValue = string | string[];
+type TranslationDict = Record<string, TranslationValue>;
 const translationCache: Record<string, TranslationDict> = {};
 const loadingPromises: Record<string, Promise<TranslationDict>> = {};
 
@@ -52,11 +58,13 @@ export function detectSystemLanguage(): SupportedLanguage {
   if (envLang?.startsWith('zh')) return 'zh';
   if (envLang?.startsWith('en')) return 'en';
   if (envLang?.startsWith('ru')) return 'ru';
+  if (envLang?.startsWith('de')) return 'de';
 
   try {
     const locale = Intl.DateTimeFormat().resolvedOptions().locale;
     if (locale.startsWith('zh')) return 'zh';
     if (locale.startsWith('ru')) return 'ru';
+    if (locale.startsWith('de')) return 'de';
   } catch {
     // Fallback to default
   }
@@ -224,9 +232,25 @@ export function getCurrentLanguage(): SupportedLanguage {
 
 export function t(key: string, params?: Record<string, string>): string {
   const translation = translations[key] ?? key;
+  if (Array.isArray(translation)) {
+    return key;
+  }
   return interpolate(translation, params);
 }
 
+/**
+ * Get a translation that is an array of strings.
+ * @param key The translation key
+ * @returns The array of strings, or an empty array if not found or not an array
+ */
+export function ta(key: string): string[] {
+  const translation = translations[key];
+  if (Array.isArray(translation)) {
+    return translation;
+  }
+  return [];
+}
+
 export async function initializeI18n(
   lang?: SupportedLanguage | 'auto',
 ): Promise<void> {

packages/cli/src/i18n/languages.ts

@@ -0,0 +1,48 @@
+/**
+ * @license
+ * Copyright 2025 Qwen team
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+export type SupportedLanguage = 'en' | 'zh' | 'ru' | 'de' | string;
+
+export interface LanguageDefinition {
+  /** The internal locale code used by the i18n system (e.g., 'en', 'zh'). */
+  code: SupportedLanguage;
+  /** The standard name used in UI settings (e.g., 'en-US', 'zh-CN'). */
+  id: string;
+  /** The full English name of the language (e.g., 'English', 'Chinese'). */
+  fullName: string;
+}
+
+export const SUPPORTED_LANGUAGES: readonly LanguageDefinition[] = [
+  {
+    code: 'en',
+    id: 'en-US',
+    fullName: 'English',
+  },
+  {
+    code: 'zh',
+    id: 'zh-CN',
+    fullName: 'Chinese',
+  },
+  {
+    code: 'ru',
+    id: 'ru-RU',
+    fullName: 'Russian',
+  },
+  {
+    code: 'de',
+    id: 'de-DE',
+    fullName: 'German',
+  },
+];
+
+/**
+ * Maps a locale code to its English language name.
+ * Used for LLM output language instructions.
+ */
+export function getLanguageNameFromLocale(locale: SupportedLanguage): string {
+  const lang = SUPPORTED_LANGUAGES.find((l) => l.code === locale);
+  return lang?.fullName || 'English';
+}

packages/cli/src/i18n/locales/en.js

@@ -102,8 +102,8 @@ export default {
   'Theme "{{themeName}}" not found.': 'Theme "{{themeName}}" not found.',
   'Theme "{{themeName}}" not found in selected scope.':
     'Theme "{{themeName}}" not found in selected scope.',
-  'clear the screen and conversation history':
-    'clear the screen and conversation history',
+  'Clear conversation history and free up context':
+    'Clear conversation history and free up context',
   'Compresses the context by replacing it with a summary.':
     'Compresses the context by replacing it with a summary.',
   'open full Qwen Code documentation in your browser':
@@ -604,9 +604,10 @@ export default {
   // ============================================================================
   // Commands - Clear
   // ============================================================================
-  'Clearing terminal and resetting chat.':
-    'Clearing terminal and resetting chat.',
-  'Clearing terminal.': 'Clearing terminal.',
+  'Starting a new session, resetting chat, and clearing terminal.':
+    'Starting a new session, resetting chat, and clearing terminal.',
+  'Starting a new session and clearing.':
+    'Starting a new session and clearing.',
 
   // ============================================================================
   // Commands - Compress
@@ -927,192 +928,138 @@ export default {
   // ============================================================================
   'Waiting for user confirmation...': 'Waiting for user confirmation...',
   '(esc to cancel, {{time}})': '(esc to cancel, {{time}})',
-  "I'm Feeling Lucky": "I'm Feeling Lucky",
-  'Shipping awesomeness... ': 'Shipping awesomeness... ',
-  'Painting the serifs back on...': 'Painting the serifs back on...',
-  'Navigating the slime mold...': 'Navigating the slime mold...',
-  'Consulting the digital spirits...': 'Consulting the digital spirits...',
-  'Reticulating splines...': 'Reticulating splines...',
-  'Warming up the AI hamsters...': 'Warming up the AI hamsters...',
-  'Asking the magic conch shell...': 'Asking the magic conch shell...',
-  'Generating witty retort...': 'Generating witty retort...',
-  'Polishing the algorithms...': 'Polishing the algorithms...',
-  "Don't rush perfection (or my code)...":
+
+  // ============================================================================
+  // Loading Phrases
+  // ============================================================================
+  WITTY_LOADING_PHRASES: [
+    "I'm Feeling Lucky",
+    'Shipping awesomeness... ',
+    'Painting the serifs back on...',
+    'Navigating the slime mold...',
+    'Consulting the digital spirits...',
+    'Reticulating splines...',
+    'Warming up the AI hamsters...',
+    'Asking the magic conch shell...',
+    'Generating witty retort...',
+    'Polishing the algorithms...',
     "Don't rush perfection (or my code)...",
-  'Brewing fresh bytes...': 'Brewing fresh bytes...',
-  'Counting electrons...': 'Counting electrons...',
-  'Engaging cognitive processors...': 'Engaging cognitive processors...',
-  'Checking for syntax errors in the universe...':
+    'Brewing fresh bytes...',
+    'Counting electrons...',
+    'Engaging cognitive processors...',
     'Checking for syntax errors in the universe...',
-  'One moment, optimizing humor...': 'One moment, optimizing humor...',
-  'Shuffling punchlines...': 'Shuffling punchlines...',
-  'Untangling neural nets...': 'Untangling neural nets...',
-  'Compiling brilliance...': 'Compiling brilliance...',
-  'Loading wit.exe...': 'Loading wit.exe...',
-  'Summoning the cloud of wisdom...': 'Summoning the cloud of wisdom...',
-  'Preparing a witty response...': 'Preparing a witty response...',
-  "Just a sec, I'm debugging reality...":
+    'One moment, optimizing humor...',
+    'Shuffling punchlines...',
+    'Untangling neural nets...',
+    'Compiling brilliance...',
+    'Loading wit.exe...',
+    'Summoning the cloud of wisdom...',
+    'Preparing a witty response...',
     "Just a sec, I'm debugging reality...",
-  'Confuzzling the options...': 'Confuzzling the options...',
-  'Tuning the cosmic frequencies...': 'Tuning the cosmic frequencies...',
-  'Crafting a response worthy of your patience...':
+    'Confuzzling the options...',
+    'Tuning the cosmic frequencies...',
     'Crafting a response worthy of your patience...',
-  'Compiling the 1s and 0s...': 'Compiling the 1s and 0s...',
-  'Resolving dependencies... and existential crises...':
+    'Compiling the 1s and 0s...',
     'Resolving dependencies... and existential crises...',
-  'Defragmenting memories... both RAM and personal...':
     'Defragmenting memories... both RAM and personal...',
-  'Rebooting the humor module...': 'Rebooting the humor module...',
-  'Caching the essentials (mostly cat memes)...':
+    'Rebooting the humor module...',
     'Caching the essentials (mostly cat memes)...',
-  'Optimizing for ludicrous speed': 'Optimizing for ludicrous speed',
-  "Swapping bits... don't tell the bytes...":
+    'Optimizing for ludicrous speed',
     "Swapping bits... don't tell the bytes...",
-  'Garbage collecting... be right back...':
     'Garbage collecting... be right back...',
-  'Assembling the interwebs...': 'Assembling the interwebs...',
-  'Converting coffee into code...': 'Converting coffee into code...',
-  'Updating the syntax for reality...': 'Updating the syntax for reality...',
-  'Rewiring the synapses...': 'Rewiring the synapses...',
-  'Looking for a misplaced semicolon...':
+    'Assembling the interwebs...',
+    'Converting coffee into code...',
+    'Updating the syntax for reality...',
+    'Rewiring the synapses...',
     'Looking for a misplaced semicolon...',
-  "Greasin' the cogs of the machine...": "Greasin' the cogs of the machine...",
-  'Pre-heating the servers...': 'Pre-heating the servers...',
-  'Calibrating the flux capacitor...': 'Calibrating the flux capacitor...',
-  'Engaging the improbability drive...': 'Engaging the improbability drive...',
-  'Channeling the Force...': 'Channeling the Force...',
-  'Aligning the stars for optimal response...':
+    "Greasin' the cogs of the machine...",
+    'Pre-heating the servers...',
+    'Calibrating the flux capacitor...',
+    'Engaging the improbability drive...',
+    'Channeling the Force...',
     'Aligning the stars for optimal response...',
-  'So say we all...': 'So say we all...',
-  'Loading the next great idea...': 'Loading the next great idea...',
-  "Just a moment, I'm in the zone...": "Just a moment, I'm in the zone...",
-  'Preparing to dazzle you with brilliance...':
+    'So say we all...',
+    'Loading the next great idea...',
+    "Just a moment, I'm in the zone...",
     'Preparing to dazzle you with brilliance...',
-  "Just a tick, I'm polishing my wit...":
     "Just a tick, I'm polishing my wit...",
-  "Hold tight, I'm crafting a masterpiece...":
     "Hold tight, I'm crafting a masterpiece...",
-  "Just a jiffy, I'm debugging the universe...":
     "Just a jiffy, I'm debugging the universe...",
-  "Just a moment, I'm aligning the pixels...":
     "Just a moment, I'm aligning the pixels...",
-  "Just a sec, I'm optimizing the humor...":
     "Just a sec, I'm optimizing the humor...",
-  "Just a moment, I'm tuning the algorithms...":
     "Just a moment, I'm tuning the algorithms...",
-  'Warp speed engaged...': 'Warp speed engaged...',
-  'Mining for more Dilithium crystals...':
+    'Warp speed engaged...',
     'Mining for more Dilithium crystals...',
-  "Don't panic...": "Don't panic...",
-  'Following the white rabbit...': 'Following the white rabbit...',
-  'The truth is in here... somewhere...':
+    "Don't panic...",
+    'Following the white rabbit...',
     'The truth is in here... somewhere...',
-  'Blowing on the cartridge...': 'Blowing on the cartridge...',
-  'Loading... Do a barrel roll!': 'Loading... Do a barrel roll!',
-  'Waiting for the respawn...': 'Waiting for the respawn...',
-  'Finishing the Kessel Run in less than 12 parsecs...':
+    'Blowing on the cartridge...',
+    'Loading... Do a barrel roll!',
+    'Waiting for the respawn...',
     'Finishing the Kessel Run in less than 12 parsecs...',
-  "The cake is not a lie, it's just still loading...":
     "The cake is not a lie, it's just still loading...",
-  'Fiddling with the character creation screen...':
     'Fiddling with the character creation screen...',
-  "Just a moment, I'm finding the right meme...":
     "Just a moment, I'm finding the right meme...",
-  "Pressing 'A' to continue...": "Pressing 'A' to continue...",
-  'Herding digital cats...': 'Herding digital cats...',
-  'Polishing the pixels...': 'Polishing the pixels...',
-  'Finding a suitable loading screen pun...':
+    "Pressing 'A' to continue...",
+    'Herding digital cats...',
+    'Polishing the pixels...',
     'Finding a suitable loading screen pun...',
-  'Distracting you with this witty phrase...':
     'Distracting you with this witty phrase...',
-  'Almost there... probably...': 'Almost there... probably...',
-  'Our hamsters are working as fast as they can...':
+    'Almost there... probably...',
     'Our hamsters are working as fast as they can...',
-  'Giving Cloudy a pat on the head...': 'Giving Cloudy a pat on the head...',
-  'Petting the cat...': 'Petting the cat...',
-  'Rickrolling my boss...': 'Rickrolling my boss...',
-  'Never gonna give you up, never gonna let you down...':
+    'Giving Cloudy a pat on the head...',
+    'Petting the cat...',
+    'Rickrolling my boss...',
     'Never gonna give you up, never gonna let you down...',
-  'Slapping the bass...': 'Slapping the bass...',
-  'Tasting the snozberries...': 'Tasting the snozberries...',
-  "I'm going the distance, I'm going for speed...":
+    'Slapping the bass...',
+    'Tasting the snozberries...',
     "I'm going the distance, I'm going for speed...",
-  'Is this the real life? Is this just fantasy?...':
     'Is this the real life? Is this just fantasy?...',
-  "I've got a good feeling about this...":
     "I've got a good feeling about this...",
-  'Poking the bear...': 'Poking the bear...',
-  'Doing research on the latest memes...':
+    'Poking the bear...',
     'Doing research on the latest memes...',
-  'Figuring out how to make this more witty...':
     'Figuring out how to make this more witty...',
-  'Hmmm... let me think...': 'Hmmm... let me think...',
-  'What do you call a fish with no eyes? A fsh...':
+    'Hmmm... let me think...',
     'What do you call a fish with no eyes? A fsh...',
-  'Why did the computer go to therapy? It had too many bytes...':
     'Why did the computer go to therapy? It had too many bytes...',
-  "Why don't programmers like nature? It has too many bugs...":
     "Why don't programmers like nature? It has too many bugs...",
-  'Why do programmers prefer dark mode? Because light attracts bugs...':
     'Why do programmers prefer dark mode? Because light attracts bugs...',
-  'Why did the developer go broke? Because they used up all their cache...':
     'Why did the developer go broke? Because they used up all their cache...',
-  "What can you do with a broken pencil? Nothing, it's pointless...":
     "What can you do with a broken pencil? Nothing, it's pointless...",
-  'Applying percussive maintenance...': 'Applying percussive maintenance...',
-  'Searching for the correct USB orientation...':
+    'Applying percussive maintenance...',
     'Searching for the correct USB orientation...',
-  'Ensuring the magic smoke stays inside the wires...':
     'Ensuring the magic smoke stays inside the wires...',
-  'Rewriting in Rust for no particular reason...':
     'Rewriting in Rust for no particular reason...',
-  'Trying to exit Vim...': 'Trying to exit Vim...',
-  'Spinning up the hamster wheel...': 'Spinning up the hamster wheel...',
-  "That's not a bug, it's an undocumented feature...":
+    'Trying to exit Vim...',
+    'Spinning up the hamster wheel...',
     "That's not a bug, it's an undocumented feature...",
-  'Engage.': 'Engage.',
-  "I'll be back... with an answer.": "I'll be back... with an answer.",
-  'My other process is a TARDIS...': 'My other process is a TARDIS...',
-  'Communing with the machine spirit...':
+    'Engage.',
+    "I'll be back... with an answer.",
+    'My other process is a TARDIS...',
     'Communing with the machine spirit...',
-  'Letting the thoughts marinate...': 'Letting the thoughts marinate...',
-  'Just remembered where I put my keys...':
+    'Letting the thoughts marinate...',
     'Just remembered where I put my keys...',
-  'Pondering the orb...': 'Pondering the orb...',
-  "I've seen things you people wouldn't believe... like a user who reads loading messages.":
+    'Pondering the orb...',
     "I've seen things you people wouldn't believe... like a user who reads loading messages.",
-  'Initiating thoughtful gaze...': 'Initiating thoughtful gaze...',
-  "What's a computer's favorite snack? Microchips.":
+    'Initiating thoughtful gaze...',
     "What's a computer's favorite snack? Microchips.",
-  "Why do Java developers wear glasses? Because they don't C#.":
     "Why do Java developers wear glasses? Because they don't C#.",
-  'Charging the laser... pew pew!': 'Charging the laser... pew pew!',
-  'Dividing by zero... just kidding!': 'Dividing by zero... just kidding!',
-  'Looking for an adult superviso... I mean, processing.':
+    'Charging the laser... pew pew!',
+    'Dividing by zero... just kidding!',
     'Looking for an adult superviso... I mean, processing.',
-  'Making it go beep boop.': 'Making it go beep boop.',
-  'Buffering... because even AIs need a moment.':
+    'Making it go beep boop.',
     'Buffering... because even AIs need a moment.',
-  'Entangling quantum particles for a faster response...':
     'Entangling quantum particles for a faster response...',
-  'Polishing the chrome... on the algorithms.':
     'Polishing the chrome... on the algorithms.',
-  'Are you not entertained? (Working on it!)':
     'Are you not entertained? (Working on it!)',
-  'Summoning the code gremlins... to help, of course.':
     'Summoning the code gremlins... to help, of course.',
-  'Just waiting for the dial-up tone to finish...':
     'Just waiting for the dial-up tone to finish...',
-  'Recalibrating the humor-o-meter.': 'Recalibrating the humor-o-meter.',
-  'My other loading screen is even funnier.':
+    'Recalibrating the humor-o-meter.',
     'My other loading screen is even funnier.',
-  "Pretty sure there's a cat walking on the keyboard somewhere...":
     "Pretty sure there's a cat walking on the keyboard somewhere...",
-  'Enhancing... Enhancing... Still loading.':
     'Enhancing... Enhancing... Still loading.',
-  "It's not a bug, it's a feature... of this loading screen.":
     "It's not a bug, it's a feature... of this loading screen.",
-  'Have you tried turning it off and on again? (The loading screen, not me.)':
     'Have you tried turning it off and on again? (The loading screen, not me.)',
-  'Constructing additional pylons...': 'Constructing additional pylons...',
+    'Constructing additional pylons...',
+  ],
 };

packages/cli/src/i18n/locales/ru.js

@@ -103,8 +103,8 @@ export default {
   'Theme "{{themeName}}" not found.': 'Тема "{{themeName}}" не найдена.',
   'Theme "{{themeName}}" not found in selected scope.':
     'Тема "{{themeName}}" не найдена в выбранной области.',
-  'clear the screen and conversation history':
-    'Очистка экрана и истории диалога',
+  'Clear conversation history and free up context':
+    'Очистить историю диалога и освободить контекст',
   'Compresses the context by replacing it with a summary.':
     'Сжатие контекста заменой на краткую сводку',
   'open full Qwen Code documentation in your browser':
@@ -313,6 +313,7 @@ export default {
   'Tool Output Truncation Lines': 'Лимит строк вывода инструментов',
   'Folder Trust': 'Доверие к папке',
   'Vision Model Preview': 'Визуальная модель (предпросмотр)',
+  'Tool Schema Compliance': 'Соответствие схеме инструмента',
   // Варианты перечислений настроек
   'Auto (detect from system)': 'Авто (определить из системы)',
   Text: 'Текст',
@@ -341,8 +342,8 @@ export default {
     'Установка предпочитаемого внешнего редактора',
   'Manage extensions': 'Управление расширениями',
   'List active extensions': 'Показать активные расширения',
-  'Update extensions. Usage: update |--all':
-    'Обновить расширения. Использование: update |--all',
+  'Update extensions. Usage: update <extension-names>|--all':
+    'Обновить расширения. Использование: update <extension-names>|--all',
   'manage IDE integration': 'Управление интеграцией с IDE',
   'check status of IDE integration': 'Проверить статус интеграции с IDE',
   'install required IDE companion for {{ideName}}':
@@ -400,7 +401,8 @@ export default {
   'Set LLM output language': 'Установка языка вывода LLM',
   'Usage: /language ui [zh-CN|en-US]':
     'Использование: /language ui [zh-CN|en-US|ru-RU]',
-  'Usage: /language output ': 'Использование: /language output ',
+  'Usage: /language output <language>':
+    'Использование: /language output <language>',
   'Example: /language output 中文': 'Пример: /language output 中文',
   'Example: /language output English': 'Пример: /language output English',
   'Example: /language output 日本語': 'Пример: /language output 日本語',
@@ -417,9 +419,8 @@ export default {
   'To request additional UI language packs, please open an issue on GitHub.':
     'Для запроса дополнительных языковых пакетов интерфейса, пожалуйста, создайте обращение на GitHub.',
   'Available options:': 'Доступные варианты:',
-  ' - zh-CN: Simplified Chinese': ' - zh-CN: Упрощенный китайский',
-  ' - en-US: English': ' - en-US: Английский',
-  ' - ru-RU: Russian': ' - ru-RU: Русский',
+  '  - zh-CN: Simplified Chinese': ' - zh-CN: Упрощенный китайский',
+  '  - en-US: English': ' - en-US: Английский',
   'Set UI language to Simplified Chinese (zh-CN)':
     'Установить язык интерфейса на упрощенный китайский (zh-CN)',
   'Set UI language to English (en-US)':
@@ -435,8 +436,8 @@ export default {
     'Режим подтверждения изменен на: {{mode}}',
   'Approval mode changed to: {{mode}} (saved to {{scope}} settings{{location}})':
     'Режим подтверждения изменен на: {{mode}} (сохранено в настройках {{scope}}{{location}})',
-  'Usage: /approval-mode [--session|--user|--project]':
-    'Использование: /approval-mode [--session|--user|--project]',
+  'Usage: /approval-mode <mode> [--session|--user|--project]':
+    'Использование: /approval-mode <mode> [--session|--user|--project]',
   'Scope subcommands do not accept additional arguments.':
     'Подкоманды области не принимают дополнительных аргументов.',
   'Plan mode - Analyze only, do not modify files or execute commands':
@@ -588,8 +589,8 @@ export default {
     'Ошибка при экспорте диалога: {{error}}',
   'Conversation shared to {{filePath}}': 'Диалог экспортирован в {{filePath}}',
   'No conversation found to share.': 'Нет диалога для экспорта.',
-  'Share the current conversation to a markdown or json file. Usage: /chat share <путь-к-файлу>':
-    'Экспортировать текущий диалог в markdown или json файл. Использование: /chat share <путь-к-файлу>',
+  'Share the current conversation to a markdown or json file. Usage: /chat share <file>':
+    'Экспортировать текущий диалог в markdown или json файл. Использование: /chat share <файл>',
 
   // ============================================================================
   // Команды - Резюме
@@ -618,8 +619,9 @@ export default {
   // ============================================================================
   // Команды - Очистка
   // ============================================================================
-  'Clearing terminal and resetting chat.': 'Очистка терминала и сброс чата.',
-  'Clearing terminal.': 'Очистка терминала.',
+  'Starting a new session, resetting chat, and clearing terminal.':
+    'Начало новой сессии, сброс чата и очистка терминала.',
+  'Starting a new session and clearing.': 'Начало новой сессии и очистка.',
 
   // ============================================================================
   // Команды - Сжатие
@@ -650,8 +652,8 @@ export default {
     'Команда /directory add не поддерживается в ограничительных профилях песочницы. Пожалуйста, используйте --include-directories при запуске сессии.',
   "Error adding '{{path}}': {{error}}":
     "Ошибка при добавлении '{{path}}': {{error}}",
-  'Successfully added GEMINI.md files from the following directories if there are:\n- {{directories}}':
-    'Успешно добавлены файлы GEMINI.md из следующих директорий (если они есть):\n- {{directories}}',
+  'Successfully added QWEN.md files from the following directories if there are:\n- {{directories}}':
+    'Успешно добавлены файлы QWEN.md из следующих директорий (если они есть):\n- {{directories}}',
   'Error refreshing memory: {{error}}':
     'Ошибка при обновлении памяти: {{error}}',
   'Successfully added directories:\n- {{directories}}':
@@ -884,6 +886,7 @@ export default {
   // Экран выхода / Статистика
   // ============================================================================
   'Agent powering down. Goodbye!': 'Агент завершает работу. До свидания!',
+  'To continue this session, run': 'Для продолжения этой сессии, выполните',
   'Interaction Summary': 'Сводка взаимодействия',
   'Session ID:': 'ID сессии:',
   'Tool Calls:': 'Вызовы инструментов:',
@@ -943,179 +946,140 @@ export default {
   'Waiting for user confirmation...':
     'Ожидание подтверждения от пользователя...',
   '(esc to cancel, {{time}})': '(esc для отмены, {{time}})',
-  "I'm Feeling Lucky": 'Мне повезёт!',
-  'Shipping awesomeness... ': 'Доставляем крутизну... ',
-  'Painting the serifs back on...': 'Рисуем засечки на буквах...',
-  'Navigating the slime mold...': 'Пробираемся через слизевиков..',
-  'Consulting the digital spirits...': 'Советуемся с цифровыми духами...',
-  'Reticulating splines...': 'Сглаживание сплайнов...',
-  'Warming up the AI hamsters...': 'Разогреваем ИИ-хомячков...',
-  'Asking the magic conch shell...': 'Спрашиваем волшебную ракушку...',
-  'Generating witty retort...': 'Генерируем остроумный ответ...',
-  'Polishing the algorithms...': 'Полируем алгоритмы...',
-  "Don't rush perfection (or my code)...":
+
+  // ============================================================================
+
+  // ============================================================================
+  // Loading Phrases
+  // ============================================================================
+  WITTY_LOADING_PHRASES: [
+    'Мне повезёт!',
+    'Доставляем крутизну... ',
+    'Рисуем засечки на буквах...',
+    'Пробираемся через слизевиков..',
+    'Советуемся с цифровыми духами...',
+    'Сглаживание сплайнов...',
+    'Разогреваем ИИ-хомячков...',
+    'Спрашиваем волшебную ракушку...',
+    'Генерируем остроумный ответ...',
+    'Полируем алгоритмы...',
     'Не торопите совершенство (или мой код)...',
-  'Brewing fresh bytes...': 'Завариваем свежие байты...',
-  'Counting electrons...': 'Пересчитываем электроны...',
-  'Engaging cognitive processors...': 'Задействуем когнитивные процессоры...',
-  'Checking for syntax errors in the universe...':
+    'Завариваем свежие байты...',
+    'Пересчитываем электроны...',
+    'Задействуем когнитивные процессоры...',
     'Ищем синтаксические ошибки во вселенной...',
-  'One moment, optimizing humor...': 'Секундочку, оптимизируем юмор...',
-  'Shuffling punchlines...': 'Перетасовываем панчлайны...',
-  'Untangling neural nets...': 'Распутаваем нейросети...',
-  'Compiling brilliance...': 'Компилируем гениальность...',
-  'Loading wit.exe...': 'Загружаем yumor.exe...',
-  'Summoning the cloud of wisdom...': 'Призываем облако мудрости...',
-  'Preparing a witty response...': 'Готовим остроумный ответ...',
-  "Just a sec, I'm debugging reality...": 'Секунду, идёт отладка реальности...',
-  'Confuzzling the options...': 'Запутываем варианты...',
-  'Tuning the cosmic frequencies...': 'Настраиваем космические частоты...',
-  'Crafting a response worthy of your patience...':
+    'Секундочку, оптимизируем юмор...',
+    'Перетасовываем панчлайны...',
+    'Распутаваем нейросети...',
+    'Компилируем гениальность...',
+    'Загружаем yumor.exe...',
+    'Призываем облако мудрости...',
+    'Готовим остроумный ответ...',
+    'Секунду, идёт отладка реальности...',
+    'Запутываем варианты...',
+    'Настраиваем космические частоты...',
     'Создаем ответ, достойный вашего терпения...',
-  'Compiling the 1s and 0s...': 'Компилируем единички и нолики...',
-  'Resolving dependencies... and existential crises...':
+    'Компилируем единички и нолики...',
     'Разрешаем зависимости... и экзистенциальные кризисы...',
-  'Defragmenting memories... both RAM and personal...':
     'Дефрагментация памяти... и оперативной, и личной...',
-  'Rebooting the humor module...': 'Перезагрузка модуля юмора...',
-  'Caching the essentials (mostly cat memes)...':
+    'Перезагрузка модуля юмора...',
     'Кэшируем самое важное (в основном мемы с котиками)...',
-  'Optimizing for ludicrous speed': 'Оптимизация для безумной скорости',
-  "Swapping bits... don't tell the bytes...":
+    'Оптимизация для безумной скорости',
     'Меняем биты... только байтам не говорите...',
-  'Garbage collecting... be right back...': 'Сборка мусора... скоро вернусь...',
-  'Assembling the interwebs...': 'Сборка интернетов...',
-  'Converting coffee into code...': 'Превращаем кофе в код...',
-  'Updating the syntax for reality...': 'Обновляем синтаксис реальности...',
-  'Rewiring the synapses...': 'Переподключаем синапсы...',
-  'Looking for a misplaced semicolon...': 'Ищем лишнюю точку с запятой...',
-  "Greasin' the cogs of the machine...": 'Смазываем шестерёнки машины...',
-  'Pre-heating the servers...': 'Разогреваем серверы...',
-  'Calibrating the flux capacitor...': 'Калибруем потоковый накопитель...',
-  'Engaging the improbability drive...': 'Включаем двигатель невероятности...',
-  'Channeling the Force...': 'Направляем Силу...',
-  'Aligning the stars for optimal response...':
+    'Сборка мусора... скоро вернусь...',
+    'Сборка интернетов...',
+    'Превращаем кофе в код...',
+    'Обновляем синтаксис реальности...',
+    'Переподключаем синапсы...',
+    'Ищем лишнюю точку с запятой...',
+    'Смазываем шестерёнки машины...',
+    'Разогреваем серверы...',
+    'Калибруем потоковый накопитель...',
+    'Включаем двигатель невероятности...',
+    'Направляем Силу...',
     'Выравниваем звёзды для оптимального ответа...',
-  'So say we all...': 'Так скажем мы все...',
-  'Loading the next great idea...': 'Загрузка следующей великой идеи...',
-  "Just a moment, I'm in the zone...": 'Минутку, я в потоке...',
-  'Preparing to dazzle you with brilliance...':
+    'Так скажем мы все...',
+    'Загрузка следующей великой идеи...',
+    'Минутку, я в потоке...',
     'Готовлюсь ослепить вас гениальностью...',
-  "Just a tick, I'm polishing my wit...": 'Секунду, полирую остроумие...',
-  "Hold tight, I'm crafting a masterpiece...": 'Держитесь, создаю шедевр...',
-  "Just a jiffy, I'm debugging the universe...":
+    'Секунду, полирую остроумие...',
+    'Держитесь, создаю шедевр...',
     'Мигом, отлаживаю вселенную...',
-  "Just a moment, I'm aligning the pixels...": 'Момент, выравниваю пиксели...',
-  "Just a sec, I'm optimizing the humor...": 'Секунду, оптимизирую юмор...',
-  "Just a moment, I'm tuning the algorithms...":
+    'Момент, выравниваю пиксели...',
+    'Секунду, оптимизирую юмор...',
     'Момент, настраиваю алгоритмы...',
-  'Warp speed engaged...': 'Варп-скорость включена...',
-  'Mining for more Dilithium crystals...': 'Добываем кристаллы дилития...',
-  "Don't panic...": 'Без паники...',
-  'Following the white rabbit...': 'Следуем за белым кроликом...',
-  'The truth is in here... somewhere...': 'Истина где-то здесь... внутри...',
-  'Blowing on the cartridge...': 'Продуваем картридж...',
-  'Loading... Do a barrel roll!': 'Загрузка... Сделай бочку!',
-  'Waiting for the respawn...': 'Ждем респауна...',
-  'Finishing the Kessel Run in less than 12 parsecs...':
+    'Варп-прыжок активирован...',
+    'Добываем кристаллы дилития...',
+    'Без паники...',
+    'Следуем за белым кроликом...',
+    'Истина где-то здесь... внутри...',
+    'Продуваем картридж...',
+    'Загрузка... Сделай бочку!',
+    'Ждем респауна...',
     'Делаем Дугу Кесселя менее чем за 12 парсеков...',
-  "The cake is not a lie, it's just still loading...":
     'Тортик — не ложь, он просто ещё грузится...',
-  'Fiddling with the character creation screen...':
     'Возимся с экраном создания персонажа...',
-  "Just a moment, I'm finding the right meme...":
     'Минутку, ищу подходящий мем...',
-  "Pressing 'A' to continue...": "Нажимаем 'A' для продолжения...",
-  'Herding digital cats...': 'Пасём цифровых котов...',
-  'Polishing the pixels...': 'Полируем пиксели...',
-  'Finding a suitable loading screen pun...':
+    "Нажимаем 'A' для продолжения...",
+    'Пасём цифровых котов...',
+    'Полируем пиксели...',
     'Ищем подходящий каламбур для экрана загрузки...',
-  'Distracting you with this witty phrase...':
     'Отвлекаем вас этой остроумной фразой...',
-  'Almost there... probably...': 'Почти готово... вроде...',
-  'Our hamsters are working as fast as they can...':
+    'Почти готово... вроде...',
     'Наши хомячки работают изо всех сил...',
-  'Giving Cloudy a pat on the head...': 'Гладим Облачко по голове...',
-  'Petting the cat...': 'Гладим кота...',
-  'Rickrolling my boss...': 'Рикроллим начальника...',
-  'Never gonna give you up, never gonna let you down...':
+    'Гладим Облачко по голове...',
+    'Гладим кота...',
+    'Рикроллим начальника...',
     'Never gonna give you up, never gonna let you down...',
-  'Slapping the bass...': 'Лабаем бас-гитару...',
-  'Tasting the snozberries...': 'Пробуем снузберри на вкус...',
-  "I'm going the distance, I'm going for speed...":
+    'Лабаем бас-гитару...',
+    'Пробуем снузберри на вкус...',
     'Иду до конца, иду на скорость...',
-  'Is this the real life? Is this just fantasy?...':
     'Is this the real life? Is this just fantasy?...',
-  "I've got a good feeling about this...": 'У меня хорошее предчувствие...',
-  'Poking the bear...': 'Дразним медведя... (Не лезь...)',
-  'Doing research on the latest memes...': 'Изучаем свежие мемы...',
-  'Figuring out how to make this more witty...':
+    'У меня хорошее предчувствие...',
+    'Дразним медведя... (Не лезь...)',
+    'Изучаем свежие мемы...',
     'Думаем, как сделать это остроумнее...',
-  'Hmmm... let me think...': 'Хмм... дайте подумать...',
-  'What do you call a fish with no eyes? A fsh...':
+    'Хмм... дайте подумать...',
     'Как называется бумеранг, который не возвращается? Палка...',
-  'Why did the computer go to therapy? It had too many bytes...':
     'Почему компьютер простудился? Потому что оставил окна открытыми...',
-  "Why don't programmers like nature? It has too many bugs...":
     'Почему программисты не любят гулять на улице? Там среда не настроена...',
-  'Why do programmers prefer dark mode? Because light attracts bugs...':
     'Почему программисты предпочитают тёмную тему? Потому что в темноте не видно багов...',
-  'Why did the developer go broke? Because they used up all their cache...':
     'Почему разработчик разорился? Потому что потратил весь свой кэш...',
-  "What can you do with a broken pencil? Nothing, it's pointless...":
     'Что можно делать со сломанным карандашом? Ничего — он тупой...',
-  'Applying percussive maintenance...': 'Провожу настройку методом тыка...',
-  'Searching for the correct USB orientation...':
+    'Провожу настройку методом тыка...',
     'Ищем, какой стороной вставлять флешку...',
-  'Ensuring the magic smoke stays inside the wires...':
     'Следим, чтобы волшебный дым не вышел из проводов...',
-  'Rewriting in Rust for no particular reason...':
     'Переписываем всё на Rust без особой причины...',
-  'Trying to exit Vim...': 'Пытаемся выйти из Vim...',
-  'Spinning up the hamster wheel...': 'Раскручиваем колесо для хомяка...',
-  "That's not a bug, it's an undocumented feature...": 'Это не баг, а фича...',
-  'Engage.': 'Поехали!',
-  "I'll be back... with an answer.": 'Я вернусь... с ответом.',
-  'My other process is a TARDIS...': 'Мой другой процесс — это ТАРДИС...',
-  'Communing with the machine spirit...': 'Общаемся с духом машины...',
-  'Letting the thoughts marinate...': 'Даем мыслям замариноваться...',
-  'Just remembered where I put my keys...':
+    'Пытаемся выйти из Vim...',
+    'Раскручиваем колесо для хомяка...',
+    'Это не баг, а фича...',
+    'Поехали!',
+    'Я вернусь... с ответом.',
+    'Мой другой процесс — это ТАРДИС...',
+    'Общаемся с духом машины...',
+    'Даем мыслям замариноваться...',
     'Только что вспомнил, куда положил ключи...',
-  'Pondering the orb...': 'Размышляю над сферой...',
-  "I've seen things you people wouldn't believe... like a user who reads loading messages.":
-    'Я видел такое, во что вы, люди, просто не поверите... например, пользователя, читающего сообщения загрузки.',
-  'Initiating thoughtful gaze...': 'Инициируем задумчивый взгляд...',
-  "What's a computer's favorite snack? Microchips.":
+    'Размышляю над сферой...',
+    'Я видел такое, что вам, людям, и не снилось... пользователя, читающего эти сообщения.',
+    'Инициируем задумчивый взгляд...',
     'Что сервер заказывает в баре? Пинг-коладу.',
-  "Why do Java developers wear glasses? Because they don't C#.":
     'Почему Java-разработчики не убираются дома? Они ждут сборщик мусора...',
-  'Charging the laser... pew pew!': 'Заряжаем лазер... пиу-пиу!',
-  'Dividing by zero... just kidding!': 'Делим на ноль... шучу!',
-  'Looking for an adult superviso... I mean, processing.':
+    'Заряжаем лазер... пиу-пиу!',
+    'Делим на ноль... шучу!',
     'Ищу взрослых для присмот... в смысле, обрабатываю.',
-  'Making it go beep boop.': 'Делаем бип-буп.',
-  'Buffering... because even AIs need a moment.':
-    'Буферизация... даже ИИ нужно мгновение.',
-  'Entangling quantum particles for a faster response...':
+    'Делаем бип-буп.',
+    'Буферизация... даже ИИ нужно время подумать.',
     'Запутываем квантовые частицы для быстрого ответа...',
-  'Polishing the chrome... on the algorithms.':
     'Полируем хром... на алгоритмах.',
-  'Are you not entertained? (Working on it!)':
     'Вы ещё не развлеклись?! Разве вы не за этим сюда пришли?!',
-  'Summoning the code gremlins... to help, of course.':
     'Призываем гремлинов кода... для помощи, конечно же.',
-  'Just waiting for the dial-up tone to finish...':
     'Ждем, пока закончится звук dial-up модема...',
-  'Recalibrating the humor-o-meter.': 'Перекалибровка юморометра.',
-  'My other loading screen is even funnier.':
+    'Перекалибровка юморометра.',
     'Мой другой экран загрузки ещё смешнее.',
-  "Pretty sure there's a cat walking on the keyboard somewhere...":
     'Кажется, где-то по клавиатуре гуляет кот...',
-  'Enhancing... Enhancing... Still loading.':
     'Улучшаем... Ещё улучшаем... Всё ещё грузится.',
-  "It's not a bug, it's a feature... of this loading screen.":
     'Это не баг, это фича... экрана загрузки.',
-  'Have you tried turning it off and on again? (The loading screen, not me.)':
     'Пробовали выключить и включить снова? (Экран загрузки, не меня!)',
-  'Constructing additional pylons...': 'Нужно построить больше пилонов...',
+    'Нужно построить больше пилонов...',
+  ],
 };

packages/cli/src/i18n/locales/zh.js

@@ -101,7 +101,7 @@ export default {
   'Theme "{{themeName}}" not found.': '未找到主题 "{{themeName}}"。',
   'Theme "{{themeName}}" not found in selected scope.':
     '在所选作用域中未找到主题 "{{themeName}}"。',
-  'clear the screen and conversation history': '清屏并清除对话历史',
+  'Clear conversation history and free up context': '清除对话历史并释放上下文',
   'Compresses the context by replacing it with a summary.':
     '通过用摘要替换来压缩上下文',
   'open full Qwen Code documentation in your browser':
@@ -573,8 +573,9 @@ export default {
   // ============================================================================
   // Commands - Clear
   // ============================================================================
-  'Clearing terminal and resetting chat.': '正在清屏并重置聊天',
-  'Clearing terminal.': '正在清屏',
+  'Starting a new session, resetting chat, and clearing terminal.':
+    '正在开始新会话，重置聊天并清屏。',
+  'Starting a new session and clearing.': '正在开始新会话并清屏。',
 
   // ============================================================================
   // Commands - Compress
@@ -880,165 +881,39 @@ export default {
   // ============================================================================
   'Waiting for user confirmation...': '等待用户确认...',
   '(esc to cancel, {{time}})': '（按 esc 取消，{{time}}）',
-  "I'm Feeling Lucky": '我感觉很幸运',
-  'Shipping awesomeness... ': '正在运送精彩内容... ',
-  'Painting the serifs back on...': '正在重新绘制衬线...',
-  'Navigating the slime mold...': '正在导航粘液霉菌...',
-  'Consulting the digital spirits...': '正在咨询数字精灵...',
-  'Reticulating splines...': '正在网格化样条曲线...',
-  'Warming up the AI hamsters...': '正在预热 AI 仓鼠...',
-  'Asking the magic conch shell...': '正在询问魔法海螺壳...',
-  'Generating witty retort...': '正在生成机智的反驳...',
-  'Polishing the algorithms...': '正在打磨算法...',
-  "Don't rush perfection (or my code)...": '不要急于追求完美（或我的代码）...',
-  'Brewing fresh bytes...': '正在酿造新鲜字节...',
-  'Counting electrons...': '正在计算电子...',
-  'Engaging cognitive processors...': '正在启动认知处理器...',
-  'Checking for syntax errors in the universe...':
-    '正在检查宇宙中的语法错误...',
-  'One moment, optimizing humor...': '稍等片刻，正在优化幽默感...',
-  'Shuffling punchlines...': '正在洗牌笑点...',
-  'Untangling neural nets...': '正在解开神经网络...',
-  'Compiling brilliance...': '正在编译智慧...',
-  'Loading wit.exe...': '正在加载 wit.exe...',
-  'Summoning the cloud of wisdom...': '正在召唤智慧云...',
-  'Preparing a witty response...': '正在准备机智的回复...',
-  "Just a sec, I'm debugging reality...": '稍等片刻，我正在调试现实...',
-  'Confuzzling the options...': '正在混淆选项...',
-  'Tuning the cosmic frequencies...': '正在调谐宇宙频率...',
-  'Crafting a response worthy of your patience...':
-    '正在制作值得您耐心等待的回复...',
-  'Compiling the 1s and 0s...': '正在编译 1 和 0...',
-  'Resolving dependencies... and existential crises...':
-    '正在解决依赖关系...和存在主义危机...',
-  'Defragmenting memories... both RAM and personal...':
-    '正在整理记忆碎片...包括 RAM 和个人记忆...',
-  'Rebooting the humor module...': '正在重启幽默模块...',
-  'Caching the essentials (mostly cat memes)...':
-    '正在缓存必需品（主要是猫咪表情包）...',
-  'Optimizing for ludicrous speed': '正在优化到荒谬的速度',
-  "Swapping bits... don't tell the bytes...": '正在交换位...不要告诉字节...',
-  'Garbage collecting... be right back...': '正在垃圾回收...马上回来...',
-  'Assembling the interwebs...': '正在组装互联网...',
-  'Converting coffee into code...': '正在将咖啡转换为代码...',
-  'Updating the syntax for reality...': '正在更新现实的语法...',
-  'Rewiring the synapses...': '正在重新连接突触...',
-  'Looking for a misplaced semicolon...': '正在寻找放错位置的分号...',
-  "Greasin' the cogs of the machine...": '正在给机器的齿轮上油...',
-  'Pre-heating the servers...': '正在预热服务器...',
-  'Calibrating the flux capacitor...': '正在校准通量电容器...',
-  'Engaging the improbability drive...': '正在启动不可能性驱动器...',
-  'Channeling the Force...': '正在引导原力...',
-  'Aligning the stars for optimal response...': '正在对齐星星以获得最佳回复...',
-  'So say we all...': '我们都说...',
-  'Loading the next great idea...': '正在加载下一个伟大的想法...',
-  "Just a moment, I'm in the zone...": '稍等片刻，我正进入状态...',
-  'Preparing to dazzle you with brilliance...': '正在准备用智慧让您眼花缭乱...',
-  "Just a tick, I'm polishing my wit...": '稍等片刻，我正在打磨我的智慧...',
-  "Hold tight, I'm crafting a masterpiece...": '请稍等，我正在制作杰作...',
-  "Just a jiffy, I'm debugging the universe...": '稍等片刻，我正在调试宇宙...',
-  "Just a moment, I'm aligning the pixels...": '稍等片刻，我正在对齐像素...',
-  "Just a sec, I'm optimizing the humor...": '稍等片刻，我正在优化幽默感...',
-  "Just a moment, I'm tuning the algorithms...": '稍等片刻，我正在调整算法...',
-  'Warp speed engaged...': '曲速已启动...',
-  'Mining for more Dilithium crystals...': '正在挖掘更多二锂晶体...',
-  "Don't panic...": '不要惊慌...',
-  'Following the white rabbit...': '正在跟随白兔...',
-  'The truth is in here... somewhere...': '真相在这里...某个地方...',
-  'Blowing on the cartridge...': '正在吹卡带...',
-  'Loading... Do a barrel roll!': '正在加载...做个桶滚！',
-  'Waiting for the respawn...': '等待重生...',
-  'Finishing the Kessel Run in less than 12 parsecs...':
-    '正在以不到 12 秒差距完成凯塞尔航线...',
-  "The cake is not a lie, it's just still loading...":
-    '蛋糕不是谎言，只是还在加载...',
-  'Fiddling with the character creation screen...': '正在摆弄角色创建界面...',
-  "Just a moment, I'm finding the right meme...":
-    '稍等片刻，我正在寻找合适的表情包...',
-  "Pressing 'A' to continue...": "按 'A' 继续...",
-  'Herding digital cats...': '正在放牧数字猫...',
-  'Polishing the pixels...': '正在打磨像素...',
-  'Finding a suitable loading screen pun...': '正在寻找合适的加载屏幕双关语...',
-  'Distracting you with this witty phrase...':
-    '正在用这个机智的短语分散您的注意力...',
-  'Almost there... probably...': '快到了...可能...',
-  'Our hamsters are working as fast as they can...':
-    '我们的仓鼠正在尽可能快地工作...',
-  'Giving Cloudy a pat on the head...': '正在拍拍 Cloudy 的头...',
-  'Petting the cat...': '正在抚摸猫咪...',
-  'Rickrolling my boss...': '正在 Rickroll 我的老板...',
-  'Never gonna give you up, never gonna let you down...':
-    '永远不会放弃你，永远不会让你失望...',
-  'Slapping the bass...': '正在拍打低音...',
-  'Tasting the snozberries...': '正在品尝 snozberries...',
-  "I'm going the distance, I'm going for speed...":
-    '我要走得更远，我要追求速度...',
-  'Is this the real life? Is this just fantasy?...':
-    '这是真实的生活吗？还是只是幻想？...',
-  "I've got a good feeling about this...": '我对这个感觉很好...',
-  'Poking the bear...': '正在戳熊...',
-  'Doing research on the latest memes...': '正在研究最新的表情包...',
-  'Figuring out how to make this more witty...': '正在想办法让这更有趣...',
-  'Hmmm... let me think...': '嗯...让我想想...',
-  'What do you call a fish with no eyes? A fsh...':
-    '没有眼睛的鱼叫什么？一条鱼...',
-  'Why did the computer go to therapy? It had too many bytes...':
-    '为什么电脑去看心理医生？因为它有太多字节...',
-  "Why don't programmers like nature? It has too many bugs...":
-    '为什么程序员不喜欢大自然？因为虫子太多了...',
-  'Why do programmers prefer dark mode? Because light attracts bugs...':
-    '为什么程序员喜欢暗色模式？因为光会吸引虫子...',
-  'Why did the developer go broke? Because they used up all their cache...':
-    '为什么开发者破产了？因为他们用完了所有缓存...',
-  "What can you do with a broken pencil? Nothing, it's pointless...":
-    '你能用断了的铅笔做什么？什么都不能，因为它没有笔尖...',
-  'Applying percussive maintenance...': '正在应用敲击维护...',
-  'Searching for the correct USB orientation...': '正在寻找正确的 USB 方向...',
-  'Ensuring the magic smoke stays inside the wires...':
-    '确保魔法烟雾留在电线内...',
-  'Rewriting in Rust for no particular reason...':
-    '正在用 Rust 重写，没有特别的原因...',
-  'Trying to exit Vim...': '正在尝试退出 Vim...',
-  'Spinning up the hamster wheel...': '正在启动仓鼠轮...',
-  "That's not a bug, it's an undocumented feature...":
-    '这不是一个错误，这是一个未记录的功能...',
-  'Engage.': '启动。',
-  "I'll be back... with an answer.": '我会回来的...带着答案。',
-  'My other process is a TARDIS...': '我的另一个进程是 TARDIS...',
-  'Communing with the machine spirit...': '正在与机器精神交流...',
-  'Letting the thoughts marinate...': '让想法慢慢酝酿...',
-  'Just remembered where I put my keys...': '刚刚想起我把钥匙放在哪里了...',
-  'Pondering the orb...': '正在思考球体...',
-  "I've seen things you people wouldn't believe... like a user who reads loading messages.":
-    '我见过你们不会相信的事情...比如一个阅读加载消息的用户。',
-  'Initiating thoughtful gaze...': '正在启动深思凝视...',
-  "What's a computer's favorite snack? Microchips.":
-    '电脑最喜欢的零食是什么？微芯片。',
-  "Why do Java developers wear glasses? Because they don't C#.":
-    '为什么 Java 开发者戴眼镜？因为他们不会 C#。',
-  'Charging the laser... pew pew!': '正在给激光充电...砰砰！',
-  'Dividing by zero... just kidding!': '除以零...只是开玩笑！',
-  'Looking for an adult superviso... I mean, processing.':
-    '正在寻找成人监督...我是说，处理中。',
-  'Making it go beep boop.': '让它发出哔哔声。',
-  'Buffering... because even AIs need a moment.':
-    '正在缓冲...因为即使是 AI 也需要片刻。',
-  'Entangling quantum particles for a faster response...':
-    '正在纠缠量子粒子以获得更快的回复...',
-  'Polishing the chrome... on the algorithms.': '正在打磨铬...在算法上。',
-  'Are you not entertained? (Working on it!)': '你不觉得有趣吗？（正在努力！）',
-  'Summoning the code gremlins... to help, of course.':
-    '正在召唤代码小精灵...当然是来帮忙的。',
-  'Just waiting for the dial-up tone to finish...': '只是等待拨号音结束...',
-  'Recalibrating the humor-o-meter.': '正在重新校准幽默计。',
-  'My other loading screen is even funnier.': '我的另一个加载屏幕更有趣。',
-  "Pretty sure there's a cat walking on the keyboard somewhere...":
-    '很确定有只猫在某个地方键盘上走...',
-  'Enhancing... Enhancing... Still loading.':
-    '正在增强...正在增强...仍在加载。',
-  "It's not a bug, it's a feature... of this loading screen.":
-    '这不是一个错误，这是一个功能...这个加载屏幕的功能。',
-  'Have you tried turning it off and on again? (The loading screen, not me.)':
-    '你试过把它关掉再打开吗？（加载屏幕，不是我。）',
-  'Constructing additional pylons...': '正在建造额外的能量塔...',
+  WITTY_LOADING_PHRASES: [
+    // --- 职场搬砖系列 ---
+    '正在努力搬砖，请稍候...',
+    '老板在身后，快加载啊！',
+    '头发掉光前，一定能加载完...',
+    '服务器正在深呼吸，准备放大招...',
+    '正在向服务器投喂咖啡...',
+
+    // --- 大厂黑话系列 ---
+    '正在赋能全链路，寻找关键抓手...',
+    '正在降本增效，优化加载路径...',
+    '正在打破部门壁垒，沉淀方法论...',
+    '正在拥抱变化，迭代核心价值...',
+    '正在对齐颗粒度，打磨底层逻辑...',
+    '大力出奇迹，正在强行加载...',
+
+    // --- 程序员自嘲系列 ---
+    '只要我不写代码，代码就没有 Bug...',
+    '正在把 Bug 转化为 Feature...',
+    '只要我不尴尬，Bug 就追不上我...',
+    '正在试图理解去年的自己写了什么...',
+    '正在猿力觉醒中，请耐心等待...',
+
+    // --- 合作愉快系列 ---
+    '正在询问产品经理：这需求是真的吗？',
+    '正在给产品经理画饼，请稍等...',
+
+    // --- 温暖治愈系列 ---
+    '每一行代码，都在努力让世界变得更好一点点...',
+    '每一个伟大的想法，都值得这份耐心的等待...',
+    '别急，美好的事物总是需要一点时间去酝酿...',
+    '愿你的代码永无 Bug，愿你的梦想终将成真...',
+    '哪怕只有 0.1% 的进度，也是在向目标靠近...',
+    '加载的是字节，承载的是对技术的热爱...',
+  ],
 };

packages/cli/src/nonInteractive/io/BaseJsonOutputAdapter.ts

@@ -610,6 +610,8 @@ export abstract class BaseJsonOutputAdapter {
         const errorText = parseAndFormatApiError(
           event.value.error,
           this.config.getContentGeneratorConfig()?.authType,
+          undefined,
+          this.config.getModel(),
         );
         this.appendText(state, errorText, null);
         break;

packages/cli/src/nonInteractiveCli.ts

@@ -221,6 +221,8 @@ export async function runNonInteractive(
               const errorText = parseAndFormatApiError(
                 event.value.error,
                 config.getContentGeneratorConfig()?.authType,
+                undefined,
+                config.getModel(),
               );
               process.stderr.write(`${errorText}\n`);
             }

packages/cli/src/services/McpPromptLoader.test.ts

@@ -28,7 +28,7 @@ const mockPrompt = {
     { name: 'trail', required: false, description: "The animal's trail." },
   ],
   invoke: vi.fn().mockResolvedValue({
-    messages: [{ content: { type: 'text', text: 'Hello, world!' } }],
+    messages: [{ content: { text: 'Hello, world!' } }],
   }),
 };
 

packages/cli/src/services/McpPromptLoader.ts

@@ -123,10 +123,7 @@ export class McpPromptLoader implements ICommandLoader {
                 };
               }
 
-              const firstMessage = result.messages?.[0];
-              const content = firstMessage?.content;
-
-              if (content?.type !== 'text') {
+              if (!result.messages?.[0]?.content?.['text']) {
                 return {
                   type: 'message',
                   messageType: 'error',
@@ -137,7 +134,7 @@ export class McpPromptLoader implements ICommandLoader {
 
               return {
                 type: 'submit_prompt',
-                content: JSON.stringify(content.text),
+                content: JSON.stringify(result.messages[0].content.text),
               };
             } catch (error) {
               return {

packages/cli/src/ui/AppContainer.test.tsx

@@ -23,6 +23,7 @@ import {
 } from '@qwen-code/qwen-code-core';
 import type { LoadedSettings } from '../config/settings.js';
 import type { InitializationResult } from '../core/initializer.js';
+import { useQuotaAndFallback } from './hooks/useQuotaAndFallback.js';
 import { UIStateContext, type UIState } from './contexts/UIStateContext.js';
 import {
   UIActionsContext,
@@ -55,6 +56,7 @@ vi.mock('./App.js', () => ({
   App: TestContextConsumer,
 }));
 
+vi.mock('./hooks/useQuotaAndFallback.js');
 vi.mock('./hooks/useHistoryManager.js');
 vi.mock('./hooks/useThemeCommand.js');
 vi.mock('./auth/useAuth.js');
@@ -120,6 +122,7 @@ describe('AppContainer State Management', () => {
   let mockInitResult: InitializationResult;
 
   // Create typed mocks for all hooks
+  const mockedUseQuotaAndFallback = useQuotaAndFallback as Mock;
   const mockedUseHistory = useHistory as Mock;
   const mockedUseThemeCommand = useThemeCommand as Mock;
   const mockedUseAuthCommand = useAuthCommand as Mock;
@@ -161,6 +164,10 @@ describe('AppContainer State Management', () => {
     capturedUIActions = null!;
 
     // **Provide a default return value for EVERY mocked hook.**
+    mockedUseQuotaAndFallback.mockReturnValue({
+      proQuotaRequest: null,
+      handleProQuotaChoice: vi.fn(),
+    });
     mockedUseHistory.mockReturnValue({
       history: [],
       addItem: vi.fn(),
@@ -560,6 +567,75 @@ describe('AppContainer State Management', () => {
     });
   });
 
+  describe('Quota and Fallback Integration', () => {
+    it('passes a null proQuotaRequest to UIStateContext by default', () => {
+      // The default mock from beforeEach already sets proQuotaRequest to null
+      render(
+        <AppContainer
+          config={mockConfig}
+          settings={mockSettings}
+          version="1.0.0"
+          initializationResult={mockInitResult}
+        />,
+      );
+
+      // Assert that the context value is as expected
+      expect(capturedUIState.proQuotaRequest).toBeNull();
+    });
+
+    it('passes a valid proQuotaRequest to UIStateContext when provided by the hook', () => {
+      // Arrange: Create a mock request object that a UI dialog would receive
+      const mockRequest = {
+        failedModel: 'gemini-pro',
+        fallbackModel: 'gemini-flash',
+        resolve: vi.fn(),
+      };
+      mockedUseQuotaAndFallback.mockReturnValue({
+        proQuotaRequest: mockRequest,
+        handleProQuotaChoice: vi.fn(),
+      });
+
+      // Act: Render the container
+      render(
+        <AppContainer
+          config={mockConfig}
+          settings={mockSettings}
+          version="1.0.0"
+          initializationResult={mockInitResult}
+        />,
+      );
+
+      // Assert: The mock request is correctly passed through the context
+      expect(capturedUIState.proQuotaRequest).toEqual(mockRequest);
+    });
+
+    it('passes the handleProQuotaChoice function to UIActionsContext', () => {
+      // Arrange: Create a mock handler function
+      const mockHandler = vi.fn();
+      mockedUseQuotaAndFallback.mockReturnValue({
+        proQuotaRequest: null,
+        handleProQuotaChoice: mockHandler,
+      });
+
+      // Act: Render the container
+      render(
+        <AppContainer
+          config={mockConfig}
+          settings={mockSettings}
+          version="1.0.0"
+          initializationResult={mockInitResult}
+        />,
+      );
+
+      // Assert: The action in the context is the mock handler we provided
+      expect(capturedUIActions.handleProQuotaChoice).toBe(mockHandler);
+
+      // You can even verify that the plumbed function is callable
+      capturedUIActions.handleProQuotaChoice('auth');
+      expect(mockHandler).toHaveBeenCalledWith('auth');
+    });
+  });
+
   describe('Terminal Title Update Feature', () => {
     beforeEach(() => {
       // Reset mock stdout for each test

packages/cli/src/ui/AppContainer.tsx

@@ -32,6 +32,7 @@ import {
   type Config,
   type IdeInfo,
   type IdeContext,
+  type UserTierId,
   DEFAULT_GEMINI_FLASH_MODEL,
   IdeClient,
   ideContextStore,
@@ -47,6 +48,7 @@ import { useHistory } from './hooks/useHistoryManager.js';
 import { useMemoryMonitor } from './hooks/useMemoryMonitor.js';
 import { useThemeCommand } from './hooks/useThemeCommand.js';
 import { useAuthCommand } from './auth/useAuth.js';
+import { useQuotaAndFallback } from './hooks/useQuotaAndFallback.js';
 import { useEditorSettings } from './hooks/useEditorSettings.js';
 import { useSettingsCommand } from './hooks/useSettingsCommand.js';
 import { useModelCommand } from './hooks/useModelCommand.js';
@@ -190,6 +192,8 @@ export const AppContainer = (props: AppContainerProps) => {
 
   const [currentModel, setCurrentModel] = useState(getEffectiveModel());
 
+  const [userTier] = useState<UserTierId | undefined>(undefined);
+
   const [isConfigInitialized, setConfigInitialized] = useState(false);
 
   const [userMessages, setUserMessages] = useState<string[]>([]);
@@ -363,6 +367,14 @@ export const AppContainer = (props: AppContainerProps) => {
     cancelAuthentication,
   } = useAuthCommand(settings, config, historyManager.addItem);
 
+  const { proQuotaRequest, handleProQuotaChoice } = useQuotaAndFallback({
+    config,
+    historyManager,
+    userTier,
+    setAuthState,
+    setModelSwitchedFromQuotaError,
+  });
+
   useInitializationAuthError(initializationResult.authError, onAuthError);
 
   // Sync user tier from config when authentication changes
@@ -740,7 +752,8 @@ export const AppContainer = (props: AppContainerProps) => {
     !initError &&
     !isProcessing &&
     (streamingState === StreamingState.Idle ||
-      streamingState === StreamingState.Responding);
+      streamingState === StreamingState.Responding) &&
+    !proQuotaRequest;
 
   const [controlsHeight, setControlsHeight] = useState(0);
 
@@ -1193,6 +1206,7 @@ export const AppContainer = (props: AppContainerProps) => {
     isAuthenticating ||
     isEditorDialogOpen ||
     showIdeRestartPrompt ||
+    !!proQuotaRequest ||
     isSubagentCreateDialogOpen ||
     isAgentsManagerDialogOpen ||
     isApprovalModeDialogOpen ||
@@ -1263,6 +1277,8 @@ export const AppContainer = (props: AppContainerProps) => {
       showWorkspaceMigrationDialog,
       workspaceExtensions,
       currentModel,
+      userTier,
+      proQuotaRequest,
       contextFileNames,
       errorCount,
       availableTerminalHeight,
@@ -1351,6 +1367,8 @@ export const AppContainer = (props: AppContainerProps) => {
       showAutoAcceptIndicator,
       showWorkspaceMigrationDialog,
       workspaceExtensions,
+      userTier,
+      proQuotaRequest,
       contextFileNames,
       errorCount,
       availableTerminalHeight,
@@ -1412,6 +1430,7 @@ export const AppContainer = (props: AppContainerProps) => {
       handleClearScreen,
       onWorkspaceMigrationDialogOpen,
       onWorkspaceMigrationDialogClose,
+      handleProQuotaChoice,
       // Vision switch dialog
       handleVisionSwitchSelect,
       // Welcome back dialog
@@ -1449,6 +1468,7 @@ export const AppContainer = (props: AppContainerProps) => {
       handleClearScreen,
       onWorkspaceMigrationDialogOpen,
       onWorkspaceMigrationDialogClose,
+      handleProQuotaChoice,
       handleVisionSwitchSelect,
       handleWelcomeBackSelection,
       handleWelcomeBackClose,

packages/cli/src/ui/auth/AuthDialog.test.tsx

@@ -168,7 +168,7 @@ describe('AuthDialog', () => {
 
     it('should not show the GEMINI_API_KEY message if QWEN_DEFAULT_AUTH_TYPE is set to something else', () => {
       process.env['GEMINI_API_KEY'] = 'foobar';
-      process.env['QWEN_DEFAULT_AUTH_TYPE'] = AuthType.USE_OPENAI;
+      process.env['QWEN_DEFAULT_AUTH_TYPE'] = AuthType.LOGIN_WITH_GOOGLE;
 
       const settings: LoadedSettings = new LoadedSettings(
         {
@@ -212,7 +212,7 @@ describe('AuthDialog', () => {
 
     it('should show the GEMINI_API_KEY message if QWEN_DEFAULT_AUTH_TYPE is set to use api key', () => {
       process.env['GEMINI_API_KEY'] = 'foobar';
-      process.env['QWEN_DEFAULT_AUTH_TYPE'] = AuthType.USE_OPENAI;
+      process.env['QWEN_DEFAULT_AUTH_TYPE'] = AuthType.USE_GEMINI;
 
       const settings: LoadedSettings = new LoadedSettings(
         {
@@ -504,12 +504,12 @@ describe('AuthDialog', () => {
       },
       {
         settings: {
-          security: { auth: { selectedType: AuthType.USE_OPENAI } },
+          security: { auth: { selectedType: AuthType.LOGIN_WITH_GOOGLE } },
           ui: { customThemes: {} },
           mcpServers: {},
         },
         originalSettings: {
-          security: { auth: { selectedType: AuthType.USE_OPENAI } },
+          security: { auth: { selectedType: AuthType.LOGIN_WITH_GOOGLE } },
           ui: { customThemes: {} },
           mcpServers: {},
         },

packages/cli/src/ui/auth/useAuth.ts

@@ -225,24 +225,16 @@ export const useAuthCommand = (
     const defaultAuthType = process.env['QWEN_DEFAULT_AUTH_TYPE'];
     if (
       defaultAuthType &&
-      ![
-        AuthType.QWEN_OAUTH,
-        AuthType.USE_OPENAI,
-        AuthType.USE_GEMINI,
-        AuthType.USE_VERTEX_AI,
-      ].includes(defaultAuthType as AuthType)
+      ![AuthType.QWEN_OAUTH, AuthType.USE_OPENAI].includes(
+        defaultAuthType as AuthType,
+      )
     ) {
       onAuthError(
         t(
           'Invalid QWEN_DEFAULT_AUTH_TYPE value: "{{value}}". Valid values are: {{validValues}}',
           {
             value: defaultAuthType,
-            validValues: [
-              AuthType.QWEN_OAUTH,
-              AuthType.USE_OPENAI,
-              AuthType.USE_GEMINI,
-              AuthType.USE_VERTEX_AI,
-            ].join(', '),
+            validValues: [AuthType.QWEN_OAUTH, AuthType.USE_OPENAI].join(', '),
           },
         ),
       );

packages/cli/src/ui/commands/ideCommand.test.ts

@@ -15,6 +15,7 @@ vi.mock('@qwen-code/qwen-code-core', async (importOriginal) => {
   const original = await importOriginal<typeof core>();
   return {
     ...original,
+    getOauthClient: vi.fn(original.getOauthClient),
     getIdeInstaller: vi.fn(original.getIdeInstaller),
     IdeClient: {
       getInstance: vi.fn(),

packages/cli/src/ui/commands/languageCommand.test.ts

@@ -13,6 +13,16 @@ import { createMockCommandContext } from '../../test-utils/mockCommandContext.js
 vi.mock('../../i18n/index.js', () => ({
   setLanguageAsync: vi.fn().mockResolvedValue(undefined),
   getCurrentLanguage: vi.fn().mockReturnValue('en'),
+  detectSystemLanguage: vi.fn().mockReturnValue('en'),
+  getLanguageNameFromLocale: vi.fn((locale: string) => {
+    const map: Record<string, string> = {
+      zh: 'Chinese',
+      en: 'English',
+      ru: 'Russian',
+      de: 'German',
+    };
+    return map[locale] || 'English';
+  }),
   t: vi.fn((key: string) => key),
 }));
 
@@ -61,7 +71,10 @@ vi.mock('@qwen-code/qwen-code-core', async (importOriginal) => {
 
 // Import modules after mocking
 import * as i18n from '../../i18n/index.js';
-import { languageCommand } from './languageCommand.js';
+import {
+  languageCommand,
+  initializeLlmOutputLanguage,
+} from './languageCommand.js';
 
 describe('languageCommand', () => {
   let mockContext: CommandContext;
@@ -186,6 +199,39 @@ describe('languageCommand', () => {
         content: expect.stringContaining('Chinese'),
       });
     });
+
+    it('should parse Unicode LLM output language from marker', async () => {
+      vi.mocked(fs.existsSync).mockReturnValue(true);
+      vi.mocked(fs.readFileSync).mockReturnValue(
+        [
+          '# ⚠️ CRITICAL: 中文 Output Language Rule - HIGHEST PRIORITY ⚠️',
+          '<!-- qwen-code:llm-output-language: 中文 -->',
+          '',
+          'Some other content...',
+        ].join('\n'),
+      );
+
+      vi.mocked(i18n.t).mockImplementation(
+        (key: string, params?: Record<string, string>) => {
+          if (params && key.includes('{{lang}}')) {
+            return key.replace('{{lang}}', params['lang'] || '');
+          }
+          return key;
+        },
+      );
+
+      if (!languageCommand.action) {
+        throw new Error('The language command must have an action.');
+      }
+
+      const result = await languageCommand.action(mockContext, '');
+
+      expect(result).toEqual({
+        type: 'message',
+        messageType: 'info',
+        content: expect.stringContaining('中文'),
+      });
+    });
   });
 
   describe('main command action - config not available', () => {
@@ -400,6 +446,34 @@ describe('languageCommand', () => {
       });
     });
 
+    it('should normalize locale code "ru" to "Russian"', async () => {
+      if (!languageCommand.action) {
+        throw new Error('The language command must have an action.');
+      }
+
+      await languageCommand.action(mockContext, 'output ru');
+
+      expect(fs.writeFileSync).toHaveBeenCalledWith(
+        expect.stringContaining('output-language.md'),
+        expect.stringContaining('Russian'),
+        'utf-8',
+      );
+    });
+
+    it('should normalize locale code "de" to "German"', async () => {
+      if (!languageCommand.action) {
+        throw new Error('The language command must have an action.');
+      }
+
+      await languageCommand.action(mockContext, 'output de');
+
+      expect(fs.writeFileSync).toHaveBeenCalledWith(
+        expect.stringContaining('output-language.md'),
+        expect.stringContaining('German'),
+        'utf-8',
+      );
+    });
+
     it('should handle file write errors gracefully', async () => {
       vi.mocked(fs.writeFileSync).mockImplementation(() => {
         throw new Error('Permission denied');
@@ -481,6 +555,8 @@ describe('languageCommand', () => {
       const nestedNames = uiSubcommand?.subCommands?.map((c) => c.name);
       expect(nestedNames).toContain('zh-CN');
       expect(nestedNames).toContain('en-US');
+      expect(nestedNames).toContain('ru-RU');
+      expect(nestedNames).toContain('de-DE');
     });
 
     it('should have action that sets language', async () => {
@@ -542,6 +618,9 @@ describe('languageCommand', () => {
     const enUSSubcommand = uiSubcommand?.subCommands?.find(
       (c) => c.name === 'en-US',
     );
+    const deDESubcommand = uiSubcommand?.subCommands?.find(
+      (c) => c.name === 'de-DE',
+    );
 
     it('zh-CN should have aliases', () => {
       expect(zhCNSubcommand?.altNames).toContain('zh');
@@ -553,6 +632,12 @@ describe('languageCommand', () => {
       expect(enUSSubcommand?.altNames).toContain('english');
     });
 
+    it('de-DE should have aliases', () => {
+      expect(deDESubcommand?.altNames).toContain('de');
+      expect(deDESubcommand?.altNames).toContain('german');
+      expect(deDESubcommand?.altNames).toContain('deutsch');
+    });
+
     it('zh-CN action should set Chinese', async () => {
       if (!zhCNSubcommand?.action) {
         throw new Error('zh-CN subcommand must have an action.');
@@ -583,6 +668,21 @@ describe('languageCommand', () => {
       });
     });
 
+    it('de-DE action should set German', async () => {
+      if (!deDESubcommand?.action) {
+        throw new Error('de-DE subcommand must have an action.');
+      }
+
+      const result = await deDESubcommand.action(mockContext, '');
+
+      expect(i18n.setLanguageAsync).toHaveBeenCalledWith('de');
+      expect(result).toEqual({
+        type: 'message',
+        messageType: 'info',
+        content: expect.stringContaining('UI language changed'),
+      });
+    });
+
     it('should reject extra arguments', async () => {
       if (!zhCNSubcommand?.action) {
         throw new Error('zh-CN subcommand must have an action.');
@@ -597,4 +697,74 @@ describe('languageCommand', () => {
       });
     });
   });
+
+  describe('initializeLlmOutputLanguage', () => {
+    beforeEach(() => {
+      vi.clearAllMocks();
+      vi.mocked(fs.existsSync).mockReturnValue(false);
+      vi.mocked(fs.mkdirSync).mockImplementation(() => undefined);
+      vi.mocked(fs.writeFileSync).mockImplementation(() => undefined);
+    });
+
+    it('should create file when it does not exist', () => {
+      vi.mocked(fs.existsSync).mockReturnValue(false);
+      vi.mocked(i18n.detectSystemLanguage).mockReturnValue('en');
+
+      initializeLlmOutputLanguage();
+
+      expect(fs.mkdirSync).toHaveBeenCalled();
+      expect(fs.writeFileSync).toHaveBeenCalledWith(
+        expect.stringContaining('output-language.md'),
+        expect.stringContaining('English'),
+        'utf-8',
+      );
+    });
+
+    it('should NOT overwrite existing file', () => {
+      vi.mocked(fs.existsSync).mockReturnValue(true);
+
+      initializeLlmOutputLanguage();
+
+      expect(fs.writeFileSync).not.toHaveBeenCalled();
+    });
+
+    it('should detect Chinese locale and create Chinese rule file', () => {
+      vi.mocked(fs.existsSync).mockReturnValue(false);
+      vi.mocked(i18n.detectSystemLanguage).mockReturnValue('zh');
+
+      initializeLlmOutputLanguage();
+
+      expect(fs.writeFileSync).toHaveBeenCalledWith(
+        expect.stringContaining('output-language.md'),
+        expect.stringContaining('Chinese'),
+        'utf-8',
+      );
+    });
+
+    it('should detect Russian locale and create Russian rule file', () => {
+      vi.mocked(fs.existsSync).mockReturnValue(false);
+      vi.mocked(i18n.detectSystemLanguage).mockReturnValue('ru');
+
+      initializeLlmOutputLanguage();
+
+      expect(fs.writeFileSync).toHaveBeenCalledWith(
+        expect.stringContaining('output-language.md'),
+        expect.stringContaining('Russian'),
+        'utf-8',
+      );
+    });
+
+    it('should detect German locale and create German rule file', () => {
+      vi.mocked(fs.existsSync).mockReturnValue(false);
+      vi.mocked(i18n.detectSystemLanguage).mockReturnValue('de');
+
+      initializeLlmOutputLanguage();
+
+      expect(fs.writeFileSync).toHaveBeenCalledWith(
+        expect.stringContaining('output-language.md'),
+        expect.stringContaining('German'),
+        'utf-8',
+      );
+    });
+  });
 });

packages/cli/src/ui/commands/languageCommand.ts

@@ -1,6 +1,6 @@
 /**
  * @license
- * Copyright 2025 Google LLC
+ * Copyright 2025 Qwen team
  * SPDX-License-Identifier: Apache-2.0
  */
 
@@ -15,51 +15,72 @@ import { SettingScope } from '../../config/settings.js';
 import {
   setLanguageAsync,
   getCurrentLanguage,
+  detectSystemLanguage,
+  getLanguageNameFromLocale,
   type SupportedLanguage,
   t,
 } from '../../i18n/index.js';
+import {
+  SUPPORTED_LANGUAGES,
+  type LanguageDefinition,
+} from '../../i18n/languages.js';
 import * as fs from 'node:fs';
 import * as path from 'node:path';
 import { Storage } from '@qwen-code/qwen-code-core';
 
 const LLM_OUTPUT_LANGUAGE_RULE_FILENAME = 'output-language.md';
+const LLM_OUTPUT_LANGUAGE_MARKER_PREFIX = 'qwen-code:llm-output-language:';
+
+function parseUiLanguageArg(input: string): SupportedLanguage | null {
+  const lowered = input.trim().toLowerCase();
+  if (!lowered) return null;
+  for (const lang of SUPPORTED_LANGUAGES) {
+    if (
+      lowered === lang.code ||
+      lowered === lang.id.toLowerCase() ||
+      lowered === lang.fullName.toLowerCase()
+    ) {
+      return lang.code;
+    }
+  }
+  return null;
+}
+
+function formatUiLanguageDisplay(lang: SupportedLanguage): string {
+  const option = SUPPORTED_LANGUAGES.find((o) => o.code === lang);
+  return option ? `${option.fullName}（${option.id}）` : lang;
+}
+
+function sanitizeLanguageForMarker(language: string): string {
+  // HTML comments cannot contain "--" or end marker "-->" safely.
+  // Also avoid newlines to keep the marker single-line and robust to parsing.
+  return language
+    .replace(/[\r\n]/g, ' ')
+    .replace(/-->/g, '')
+    .replace(/--/g, '');
+}
 
 /**
  * Generates the LLM output language rule template based on the language name.
  */
 function generateLlmOutputLanguageRule(language: string): string {
-  return `# ⚠️ CRITICAL: ${language} Output Language Rule - HIGHEST PRIORITY ⚠️
+  const markerLanguage = sanitizeLanguageForMarker(language);
+  return `# Output language preference: ${language}
+<!-- ${LLM_OUTPUT_LANGUAGE_MARKER_PREFIX} ${markerLanguage} -->
 
-## 🚨 MANDATORY RULE - NO EXCEPTIONS 🚨
+## Goal
+Prefer responding in **${language}** for normal assistant messages and explanations.
 
-**YOU MUST RESPOND IN ${language.toUpperCase()} FOR EVERY SINGLE OUTPUT, REGARDLESS OF THE USER'S INPUT LANGUAGE.**
+## Keep technical artifacts unchanged
+Do **not** translate or rewrite:
+- Code blocks, CLI commands, file paths, stack traces, logs, JSON keys, identifiers
+- Exact quoted text from the user (keep quotes verbatim)
 
-This is a **NON-NEGOTIABLE** requirement. Even if the user writes in English, says "hi", asks a simple question, or explicitly requests another language, **YOU MUST ALWAYS RESPOND IN ${language.toUpperCase()}.**
+## When a conflict exists
+If higher-priority instructions (system/developer) require a different behavior, follow them.
 
-## What Must Be in ${language}
-
-**EVERYTHING** you output: conversation replies, tool call descriptions, success/error messages, generated file content (comments, documentation), and all explanatory text.
-
-**Tool outputs**: All descriptive text from \`read_file\`, \`write_file\`, \`codebase_search\`, \`run_terminal_cmd\`, \`todo_write\`, \`web_search\`, etc. MUST be in ${language}.
-
-## Examples
-
-### ✅ CORRECT:
-- User says "hi" → Respond in ${language} (e.g., "Bonjour" if ${language} is French)
-- Tool result → "已成功读取文件 config.json" (if ${language} is Chinese)
-- Error → "无法找到指定的文件" (if ${language} is Chinese)
-
-### ❌ WRONG:
-- User says "hi" → "Hello" in English
-- Tool result → "Successfully read file" in English
-- Error → "File not found" in English
-
-## Notes
-
-- Code elements (variable/function names, syntax) can remain in English
-- Comments, documentation, and all other text MUST be in ${language}
-
-**THIS RULE IS ACTIVE NOW. ALL OUTPUTS MUST BE IN ${language.toUpperCase()}. NO EXCEPTIONS.**
+## Tool / system outputs
+Raw tool/system outputs may contain fixed-format English. Preserve them verbatim, and if needed, add a short **${language}** explanation below.
 `;
 }
 
@@ -73,6 +94,80 @@ function getLlmOutputLanguageRulePath(): string {
   );
 }
 
+/**
+ * Normalizes a language input to its full English name.
+ * If the input is a known locale code (e.g., "ru", "zh"), converts it to the full name.
+ * Otherwise, returns the input as-is (e.g., "Japanese" stays "Japanese").
+ */
+function normalizeLanguageName(language: string): string {
+  const lowered = language.toLowerCase();
+  // Check if it's a known locale code and convert to full name
+  const fullName = getLanguageNameFromLocale(lowered);
+  // If getLanguageNameFromLocale returned a different value, use it
+  // Otherwise, use the original input (preserves case for unknown languages)
+  if (fullName !== 'English' || lowered === 'en') {
+    return fullName;
+  }
+  return language;
+}
+
+function extractLlmOutputLanguageFromRuleFileContent(
+  content: string,
+): string | null {
+  // Preferred: machine-readable marker that supports Unicode and spaces.
+  // Example: <!-- qwen-code:llm-output-language: 中文 -->
+  const markerMatch = content.match(
+    new RegExp(
+      String.raw`<!--\s*${LLM_OUTPUT_LANGUAGE_MARKER_PREFIX}\s*(.*?)\s*-->`,
+      'i',
+    ),
+  );
+  if (markerMatch?.[1]) {
+    const lang = markerMatch[1].trim();
+    if (lang) return lang;
+  }
+
+  // Backward compatibility: parse the heading line.
+  // Example: "# CRITICAL: Chinese Output Language Rule - HIGHEST PRIORITY"
+  // Example: "# ⚠️ CRITICAL: 日本語 Output Language Rule - HIGHEST PRIORITY ⚠️"
+  const headingMatch = content.match(
+    /^#.*?CRITICAL:\s*(.*?)\s+Output Language Rule\b/im,
+  );
+  if (headingMatch?.[1]) {
+    const lang = headingMatch[1].trim();
+    if (lang) return lang;
+  }
+
+  return null;
+}
+
+/**
+ * Initializes the LLM output language rule file on first startup.
+ * If the file already exists, it is not overwritten (respects user preference).
+ */
+export function initializeLlmOutputLanguage(): void {
+  const filePath = getLlmOutputLanguageRulePath();
+
+  // Skip if file already exists (user preference)
+  if (fs.existsSync(filePath)) {
+    return;
+  }
+
+  // Detect system language and map to language name
+  const detectedLocale = detectSystemLanguage();
+  const languageName = getLanguageNameFromLocale(detectedLocale);
+
+  // Generate the rule file
+  const content = generateLlmOutputLanguageRule(languageName);
+
+  // Ensure directory exists
+  const dir = path.dirname(filePath);
+  fs.mkdirSync(dir, { recursive: true });
+
+  // Write file
+  fs.writeFileSync(filePath, content, 'utf-8');
+}
+
 /**
  * Gets the current LLM output language from the rule file if it exists.
  */
@@ -81,12 +176,7 @@ function getCurrentLlmOutputLanguage(): string | null {
   if (fs.existsSync(filePath)) {
     try {
       const content = fs.readFileSync(filePath, 'utf-8');
-      // Extract language name from the first line
-      // Template format: "# CRITICAL: Chinese Output Language Rule - HIGHEST PRIORITY"
-      const match = content.match(/^#.*?(\w+)\s+Output Language Rule/i);
-      if (match) {
-        return match[1];
-      }
+      return extractLlmOutputLanguageFromRuleFileContent(content);
     } catch {
       // Ignore errors
     }
@@ -127,18 +217,11 @@ async function setUiLanguage(
   // Reload commands to update their descriptions with the new language
   context.ui.reloadCommands();
 
-  // Map language codes to friendly display names
-  const langDisplayNames: Partial<Record<SupportedLanguage, string>> = {
-    zh: '中文（zh-CN）',
-    en: 'English（en-US）',
-    ru: 'Русский (ru-RU)',
-  };
-
   return {
     type: 'message',
     messageType: 'info',
     content: t('UI language changed to {{lang}}', {
-      lang: langDisplayNames[lang] || lang,
+      lang: formatUiLanguageDisplay(lang),
     }),
   };
 }
@@ -151,7 +234,9 @@ function generateLlmOutputLanguageRuleFile(
 ): Promise<MessageActionReturn> {
   try {
     const filePath = getLlmOutputLanguageRulePath();
-    const content = generateLlmOutputLanguageRule(language);
+    // Normalize locale codes (e.g., "ru" -> "Russian") to full language names
+    const normalizedLanguage = normalizeLanguageName(language);
+    const content = generateLlmOutputLanguageRule(normalizedLanguage);
 
     // Ensure directory exists
     const dir = path.dirname(filePath);
@@ -195,16 +280,6 @@ export const languageCommand: SlashCommand = {
     context: CommandContext,
     args: string,
   ): Promise<SlashCommandActionReturn> => {
-    const { services } = context;
-
-    if (!services.config) {
-      return {
-        type: 'message',
-        messageType: 'error',
-        content: t('Configuration not available.'),
-      };
-    }
-
     const trimmedArgs = args.trim();
 
     // If no arguments, show current language settings and usage
@@ -212,13 +287,15 @@ export const languageCommand: SlashCommand = {
       const currentUiLang = getCurrentLanguage();
       const currentLlmLang = getCurrentLlmOutputLanguage();
       const message = [
-        t('Current UI language: {{lang}}', { lang: currentUiLang }),
+        t('Current UI language: {{lang}}', {
+          lang: formatUiLanguageDisplay(currentUiLang as SupportedLanguage),
+        }),
         currentLlmLang
           ? t('Current LLM output language: {{lang}}', { lang: currentLlmLang })
           : t('LLM output language not set'),
         '',
         t('Available subcommands:'),
-        `  /language ui [zh-CN|en-US|ru-RU] - ${t('Set UI language')}`,
+        `  /language ui [${SUPPORTED_LANGUAGES.map((o) => o.id).join('|')}] - ${t('Set UI language')}`,
         `  /language output <language> - ${t('Set LLM output language')}`,
       ].join('\n');
 
@@ -229,115 +306,21 @@ export const languageCommand: SlashCommand = {
       };
     }
 
-    // Parse subcommand
-    const parts = trimmedArgs.split(/\s+/);
-    const subcommand = parts[0].toLowerCase();
-
-    if (subcommand === 'ui') {
-      // Handle /language ui [zh-CN|en-US|ru-RU]
-      if (parts.length === 1) {
-        // Show UI language subcommand help
-        return {
-          type: 'message',
-          messageType: 'info',
-          content: [
-            t('Set UI language'),
-            '',
-            t('Usage: /language ui [zh-CN|en-US|ru-RU]'),
-            '',
-            t('Available options:'),
-            t('  - zh-CN: Simplified Chinese'),
-            t('  - en-US: English'),
-            t('  - ru-RU: Russian'),
-            '',
-            t(
-              'To request additional UI language packs, please open an issue on GitHub.',
-            ),
-          ].join('\n'),
-        };
-      }
-
-      const langArg = parts[1].toLowerCase();
-      let targetLang: SupportedLanguage | null = null;
-
-      if (langArg === 'en' || langArg === 'english' || langArg === 'en-us') {
-        targetLang = 'en';
-      } else if (
-        langArg === 'zh' ||
-        langArg === 'chinese' ||
-        langArg === '中文' ||
-        langArg === 'zh-cn'
-      ) {
-        targetLang = 'zh';
-      } else if (
-        langArg === 'ru' ||
-        langArg === 'ru-RU' ||
-        langArg === 'russian' ||
-        langArg === 'русский'
-      ) {
-        targetLang = 'ru';
-      } else {
-        return {
-          type: 'message',
-          messageType: 'error',
-          content: t('Invalid language. Available: en-US, zh-CN, ru-RU'),
-        };
-      }
-
-      return setUiLanguage(context, targetLang);
-    } else if (subcommand === 'output') {
-      // Handle /language output <language>
-      if (parts.length === 1) {
-        return {
-          type: 'message',
-          messageType: 'info',
-          content: [
-            t('Set LLM output language'),
-            '',
-            t('Usage: /language output <language>'),
-            `  ${t('Example: /language output 中文')}`,
-          ].join('\n'),
-        };
-      }
-
-      // Join all parts after "output" as the language name
-      const language = parts.slice(1).join(' ');
-      return generateLlmOutputLanguageRuleFile(language);
-    } else {
-      // Backward compatibility: treat as UI language
-      const langArg = trimmedArgs.toLowerCase();
-      let targetLang: SupportedLanguage | null = null;
-
-      if (langArg === 'en' || langArg === 'english' || langArg === 'en-us') {
-        targetLang = 'en';
-      } else if (
-        langArg === 'zh' ||
-        langArg === 'chinese' ||
-        langArg === '中文' ||
-        langArg === 'zh-cn'
-      ) {
-        targetLang = 'zh';
-      } else if (
-        langArg === 'ru' ||
-        langArg === 'ru-RU' ||
-        langArg === 'russian' ||
-        langArg === 'русский'
-      ) {
-        targetLang = 'ru';
-      } else {
-        return {
-          type: 'message',
-          messageType: 'error',
-          content: [
-            t('Invalid command. Available subcommands:'),
-            '  - /language ui [zh-CN|en-US|ru-RU] - ' + t('Set UI language'),
-            '  - /language output <language> - ' + t('Set LLM output language'),
-          ].join('\n'),
-        };
-      }
-
+    // Handle backward compatibility for /language [lang]
+    const targetLang = parseUiLanguageArg(trimmedArgs);
+    if (targetLang) {
       return setUiLanguage(context, targetLang);
     }
+
+    return {
+      type: 'message',
+      messageType: 'error',
+      content: [
+        t('Invalid command. Available subcommands:'),
+        `  - /language ui [${SUPPORTED_LANGUAGES.map((o) => o.id).join('|')}] - ${t('Set UI language')}`,
+        '  - /language output <language> - ' + t('Set LLM output language'),
+      ].join('\n'),
+    };
   },
   subCommands: [
     {
@@ -358,11 +341,14 @@ export const languageCommand: SlashCommand = {
             content: [
               t('Set UI language'),
               '',
-              t('Usage: /language ui [zh-CN|en-US]'),
+              t('Usage: /language ui [{{options}}]', {
+                options: SUPPORTED_LANGUAGES.map((o) => o.id).join('|'),
+              }),
               '',
               t('Available options:'),
-              t('  - zh-CN: Simplified Chinese'),
-              t('  - en-US: English'),
+              ...SUPPORTED_LANGUAGES.map(
+                (o) => `  - ${o.id}: ${t(o.fullName)}`,
+              ),
               '',
               t(
                 'To request additional UI language packs, please open an issue on GitHub.',
@@ -371,99 +357,20 @@ export const languageCommand: SlashCommand = {
           };
         }
 
-        const langArg = trimmedArgs.toLowerCase();
-        let targetLang: SupportedLanguage | null = null;
-
-        if (langArg === 'en' || langArg === 'english' || langArg === 'en-us') {
-          targetLang = 'en';
-        } else if (
-          langArg === 'zh' ||
-          langArg === 'chinese' ||
-          langArg === '中文' ||
-          langArg === 'zh-cn'
-        ) {
-          targetLang = 'zh';
-        } else {
+        const targetLang = parseUiLanguageArg(trimmedArgs);
+        if (!targetLang) {
           return {
             type: 'message',
             messageType: 'error',
-            content: t('Invalid language. Available: en-US, zh-CN'),
+            content: t('Invalid language. Available: {{options}}', {
+              options: SUPPORTED_LANGUAGES.map((o) => o.id).join(','),
+            }),
           };
         }
 
         return setUiLanguage(context, targetLang);
       },
-      subCommands: [
-        {
-          name: 'zh-CN',
-          altNames: ['zh', 'chinese', '中文'],
-          get description() {
-            return t('Set UI language to Simplified Chinese (zh-CN)');
-          },
-          kind: CommandKind.BUILT_IN,
-          action: async (
-            context: CommandContext,
-            args: string,
-          ): Promise<MessageActionReturn> => {
-            if (args.trim().length > 0) {
-              return {
-                type: 'message',
-                messageType: 'error',
-                content: t(
-                  'Language subcommands do not accept additional arguments.',
-                ),
-              };
-            }
-            return setUiLanguage(context, 'zh');
-          },
-        },
-        {
-          name: 'en-US',
-          altNames: ['en', 'english'],
-          get description() {
-            return t('Set UI language to English (en-US)');
-          },
-          kind: CommandKind.BUILT_IN,
-          action: async (
-            context: CommandContext,
-            args: string,
-          ): Promise<MessageActionReturn> => {
-            if (args.trim().length > 0) {
-              return {
-                type: 'message',
-                messageType: 'error',
-                content: t(
-                  'Language subcommands do not accept additional arguments.',
-                ),
-              };
-            }
-            return setUiLanguage(context, 'en');
-          },
-        },
-        {
-          name: 'ru-RU',
-          altNames: ['ru', 'russian', 'русский'],
-          get description() {
-            return t('Set UI language to Russian (ru-RU)');
-          },
-          kind: CommandKind.BUILT_IN,
-          action: async (
-            context: CommandContext,
-            args: string,
-          ): Promise<MessageActionReturn> => {
-            if (args.trim().length > 0) {
-              return {
-                type: 'message',
-                messageType: 'error',
-                content: t(
-                  'Language subcommands do not accept additional arguments.',
-                ),
-              };
-            }
-            return setUiLanguage(context, 'ru');
-          },
-        },
-      ],
+      subCommands: SUPPORTED_LANGUAGES.map(createUiLanguageSubCommand),
     },
     {
       name: 'output',
@@ -496,3 +403,28 @@ export const languageCommand: SlashCommand = {
     },
   ],
 };
+
+/**
+ * Helper to create a UI language subcommand.
+ */
+function createUiLanguageSubCommand(option: LanguageDefinition): SlashCommand {
+  return {
+    name: option.id,
+    get description() {
+      return t('Set UI language to {{name}}', { name: option.fullName });
+    },
+    kind: CommandKind.BUILT_IN,
+    action: async (context, args) => {
+      if (args.trim().length > 0) {
+        return {
+          type: 'message',
+          messageType: 'error',
+          content: t(
+            'Language subcommands do not accept additional arguments.',
+          ),
+        };
+      }
+      return setUiLanguage(context, option.code);
+    },
+  };
+}

packages/cli/src/ui/components/DialogManager.tsx

@@ -17,6 +17,7 @@ import { AuthDialog } from '../auth/AuthDialog.js';
 import { OpenAIKeyPrompt } from './OpenAIKeyPrompt.js';
 import { EditorSettingsDialog } from './EditorSettingsDialog.js';
 import { WorkspaceMigrationDialog } from './WorkspaceMigrationDialog.js';
+import { ProQuotaDialog } from './ProQuotaDialog.js';
 import { PermissionsModifyTrustDialog } from './PermissionsModifyTrustDialog.js';
 import { ModelDialog } from './ModelDialog.js';
 import { ApprovalModeDialog } from './ApprovalModeDialog.js';
@@ -86,6 +87,15 @@ export const DialogManager = ({
       />
     );
   }
+  if (uiState.proQuotaRequest) {
+    return (
+      <ProQuotaDialog
+        failedModel={uiState.proQuotaRequest.failedModel}
+        fallbackModel={uiState.proQuotaRequest.fallbackModel}
+        onChoice={uiActions.handleProQuotaChoice}
+      />
+    );
+  }
   if (uiState.shouldShowIdePrompt) {
     return (
       <IdeIntegrationNudge

packages/cli/src/ui/components/ProQuotaDialog.test.tsx

@@ -0,0 +1,91 @@
+/**
+ * @license
+ * Copyright 2025 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import { render } from 'ink-testing-library';
+import { describe, it, expect, vi, beforeEach, type Mock } from 'vitest';
+import { ProQuotaDialog } from './ProQuotaDialog.js';
+import { RadioButtonSelect } from './shared/RadioButtonSelect.js';
+
+// Mock the child component to make it easier to test the parent
+vi.mock('./shared/RadioButtonSelect.js', () => ({
+  RadioButtonSelect: vi.fn(),
+}));
+
+describe('ProQuotaDialog', () => {
+  beforeEach(() => {
+    vi.clearAllMocks();
+  });
+
+  it('should render with correct title and options', () => {
+    const { lastFrame } = render(
+      <ProQuotaDialog
+        failedModel="gemini-2.5-pro"
+        fallbackModel="gemini-2.5-flash"
+        onChoice={() => {}}
+      />,
+    );
+
+    const output = lastFrame();
+    expect(output).toContain('Pro quota limit reached for gemini-2.5-pro.');
+
+    // Check that RadioButtonSelect was called with the correct items
+    expect(RadioButtonSelect).toHaveBeenCalledWith(
+      expect.objectContaining({
+        items: [
+          {
+            label: 'Change auth (executes the /auth command)',
+            value: 'auth',
+            key: 'auth',
+          },
+          {
+            label: `Continue with gemini-2.5-flash`,
+            value: 'continue',
+            key: 'continue',
+          },
+        ],
+      }),
+      undefined,
+    );
+  });
+
+  it('should call onChoice with "auth" when "Change auth" is selected', () => {
+    const mockOnChoice = vi.fn();
+    render(
+      <ProQuotaDialog
+        failedModel="gemini-2.5-pro"
+        fallbackModel="gemini-2.5-flash"
+        onChoice={mockOnChoice}
+      />,
+    );
+
+    // Get the onSelect function passed to RadioButtonSelect
+    const onSelect = (RadioButtonSelect as Mock).mock.calls[0][0].onSelect;
+
+    // Simulate the selection
+    onSelect('auth');
+
+    expect(mockOnChoice).toHaveBeenCalledWith('auth');
+  });
+
+  it('should call onChoice with "continue" when "Continue with flash" is selected', () => {
+    const mockOnChoice = vi.fn();
+    render(
+      <ProQuotaDialog
+        failedModel="gemini-2.5-pro"
+        fallbackModel="gemini-2.5-flash"
+        onChoice={mockOnChoice}
+      />,
+    );
+
+    // Get the onSelect function passed to RadioButtonSelect
+    const onSelect = (RadioButtonSelect as Mock).mock.calls[0][0].onSelect;
+
+    // Simulate the selection
+    onSelect('continue');
+
+    expect(mockOnChoice).toHaveBeenCalledWith('continue');
+  });
+});

packages/cli/src/ui/components/ProQuotaDialog.tsx

@@ -0,0 +1,55 @@
+/**
+ * @license
+ * Copyright 2025 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import type React from 'react';
+import { Box, Text } from 'ink';
+import { RadioButtonSelect } from './shared/RadioButtonSelect.js';
+import { theme } from '../semantic-colors.js';
+import { t } from '../../i18n/index.js';
+
+interface ProQuotaDialogProps {
+  failedModel: string;
+  fallbackModel: string;
+  onChoice: (choice: 'auth' | 'continue') => void;
+}
+
+export function ProQuotaDialog({
+  failedModel,
+  fallbackModel,
+  onChoice,
+}: ProQuotaDialogProps): React.JSX.Element {
+  const items = [
+    {
+      label: t('Change auth (executes the /auth command)'),
+      value: 'auth' as const,
+      key: 'auth',
+    },
+    {
+      label: t('Continue with {{model}}', { model: fallbackModel }),
+      value: 'continue' as const,
+      key: 'continue',
+    },
+  ];
+
+  const handleSelect = (choice: 'auth' | 'continue') => {
+    onChoice(choice);
+  };
+
+  return (
+    <Box borderStyle="round" flexDirection="column" paddingX={1}>
+      <Text bold color={theme.status.warning}>
+        {t('Pro quota limit reached for {{model}}.', { model: failedModel })}
+      </Text>
+      <Box marginTop={1}>
+        <RadioButtonSelect
+          items={items}
+          initialIndex={1}
+          onSelect={handleSelect}
+        />
+      </Box>
+    </Box>
+  );
+}

packages/cli/src/ui/contexts/UIActionsContext.tsx

@@ -55,6 +55,7 @@ export interface UIActions {
   handleClearScreen: () => void;
   onWorkspaceMigrationDialogOpen: () => void;
   onWorkspaceMigrationDialogClose: () => void;
+  handleProQuotaChoice: (choice: 'auth' | 'continue') => void;
   // Vision switch dialog
   handleVisionSwitchSelect: (outcome: VisionSwitchOutcome) => void;
   // Welcome back dialog

packages/cli/src/ui/contexts/UIStateContext.tsx

@@ -22,13 +22,21 @@ import type {
   AuthType,
   IdeContext,
   ApprovalMode,
+  UserTierId,
   IdeInfo,
+  FallbackIntent,
 } from '@qwen-code/qwen-code-core';
 import type { DOMElement } from 'ink';
 import type { SessionStatsState } from '../contexts/SessionContext.js';
 import type { ExtensionUpdateState } from '../state/extensions.js';
 import type { UpdateObject } from '../utils/updateCheck.js';
 
+export interface ProQuotaDialogRequest {
+  failedModel: string;
+  fallbackModel: string;
+  resolve: (intent: FallbackIntent) => void;
+}
+
 import { type UseHistoryManagerReturn } from '../hooks/useHistoryManager.js';
 import { type RestartReason } from '../hooks/useIdeTrustListener.js';
 
@@ -91,6 +99,8 @@ export interface UIState {
   // eslint-disable-next-line @typescript-eslint/no-explicit-any
   workspaceExtensions: any[]; // Extension[]
   // Quota-related state
+  userTier: UserTierId | undefined;
+  proQuotaRequest: ProQuotaDialogRequest | null;
   currentModel: string;
   contextFileNames: string[];
   errorCount: number;

packages/cli/src/ui/hooks/useGeminiStream.test.tsx

@@ -1323,7 +1323,7 @@ describe('useGeminiStream', () => {
     it('should call parseAndFormatApiError with the correct authType on stream initialization failure', async () => {
       // 1. Setup
       const mockError = new Error('Rate limit exceeded');
-      const mockAuthType = AuthType.USE_VERTEX_AI;
+      const mockAuthType = AuthType.LOGIN_WITH_GOOGLE;
       mockParseAndFormatApiError.mockClear();
       mockSendMessageStream.mockReturnValue(
         (async function* () {
@@ -1374,6 +1374,9 @@ describe('useGeminiStream', () => {
         expect(mockParseAndFormatApiError).toHaveBeenCalledWith(
           'Rate limit exceeded',
           mockAuthType,
+          undefined,
+          'gemini-2.5-pro',
+          'gemini-2.5-flash',
         );
       });
     });
@@ -2490,6 +2493,9 @@ describe('useGeminiStream', () => {
       expect(mockParseAndFormatApiError).toHaveBeenCalledWith(
         { message: 'Test error' },
         expect.any(String),
+        undefined,
+        'gemini-2.5-pro',
+        'gemini-2.5-flash',
       );
     });
   });

packages/cli/src/ui/hooks/useGeminiStream.ts

@@ -26,6 +26,7 @@ import {
   GitService,
   UnauthorizedError,
   UserPromptEvent,
+  DEFAULT_GEMINI_FLASH_MODEL,
   logConversationFinishedEvent,
   ConversationFinishedEvent,
   ApprovalMode,
@@ -599,6 +600,9 @@ export const useGeminiStream = (
           text: parseAndFormatApiError(
             eventValue.error,
             config.getContentGeneratorConfig()?.authType,
+            undefined,
+            config.getModel(),
+            DEFAULT_GEMINI_FLASH_MODEL,
           ),
         },
         userMessageTimestamp,
@@ -650,9 +654,6 @@ export const useGeminiStream = (
           'Response stopped due to image safety violations.',
         [FinishReason.UNEXPECTED_TOOL_CALL]:
           'Response stopped due to unexpected tool call.',
-        [FinishReason.IMAGE_PROHIBITED_CONTENT]:
-          'Response stopped due to image prohibited content.',
-        [FinishReason.NO_IMAGE]: 'Response stopped due to no image.',
       };
 
       const message = finishReasonMessages[finishReason];
@@ -769,17 +770,11 @@ export const useGeminiStream = (
       for await (const event of stream) {
         switch (event.type) {
           case ServerGeminiEventType.Thought:
-            // If the thought has a subject, it's a discrete status update rather than
-            // a streamed textual thought, so we update the thought state directly.
-            if (event.value.subject) {
-              setThought(event.value);
-            } else {
-              thoughtBuffer = handleThoughtEvent(
-                event.value,
-                thoughtBuffer,
-                userMessageTimestamp,
-              );
-            }
+            thoughtBuffer = handleThoughtEvent(
+              event.value,
+              thoughtBuffer,
+              userMessageTimestamp,
+            );
             break;
           case ServerGeminiEventType.Content:
             geminiMessageBuffer = handleContentEvent(
@@ -850,7 +845,6 @@ export const useGeminiStream = (
       handleMaxSessionTurnsEvent,
       handleSessionTokenLimitExceededEvent,
       handleCitationEvent,
-      setThought,
     ],
   );
 
@@ -993,6 +987,9 @@ export const useGeminiStream = (
                 text: parseAndFormatApiError(
                   getErrorMessage(error) || 'Unknown error',
                   config.getContentGeneratorConfig()?.authType,
+                  undefined,
+                  config.getModel(),
+                  DEFAULT_GEMINI_FLASH_MODEL,
                 ),
               },
               userMessageTimestamp,

packages/cli/src/ui/hooks/useLoadingIndicator.test.ts

@@ -8,19 +8,22 @@ import { describe, it, expect, vi, beforeEach, afterEach } from 'vitest';
 import { renderHook, act } from '@testing-library/react';
 import { useLoadingIndicator } from './useLoadingIndicator.js';
 import { StreamingState } from '../types.js';
-import {
-  WITTY_LOADING_PHRASES,
-  PHRASE_CHANGE_INTERVAL_MS,
-} from './usePhraseCycler.js';
+import { PHRASE_CHANGE_INTERVAL_MS } from './usePhraseCycler.js';
+import * as i18n from '../../i18n/index.js';
+
+const MOCK_WITTY_PHRASES = ['Phrase 1', 'Phrase 2', 'Phrase 3'];
 
 describe('useLoadingIndicator', () => {
   beforeEach(() => {
     vi.useFakeTimers();
+    vi.spyOn(i18n, 'ta').mockReturnValue(MOCK_WITTY_PHRASES);
+    vi.spyOn(i18n, 't').mockImplementation((key) => key);
   });
 
   afterEach(() => {
     vi.useRealTimers(); // Restore real timers after each test
     act(() => vi.runOnlyPendingTimers);
+    vi.restoreAllMocks();
   });
 
   it('should initialize with default values when Idle', () => {
@@ -28,9 +31,7 @@ describe('useLoadingIndicator', () => {
       useLoadingIndicator(StreamingState.Idle),
     );
     expect(result.current.elapsedTime).toBe(0);
-    expect(WITTY_LOADING_PHRASES).toContain(
-      result.current.currentLoadingPhrase,
-    );
+    expect(MOCK_WITTY_PHRASES).toContain(result.current.currentLoadingPhrase);
   });
 
   it('should reflect values when Responding', async () => {
@@ -40,18 +41,14 @@ describe('useLoadingIndicator', () => {
 
     // Initial state before timers advance
     expect(result.current.elapsedTime).toBe(0);
-    expect(WITTY_LOADING_PHRASES).toContain(
-      result.current.currentLoadingPhrase,
-    );
+    expect(MOCK_WITTY_PHRASES).toContain(result.current.currentLoadingPhrase);
 
     await act(async () => {
       await vi.advanceTimersByTimeAsync(PHRASE_CHANGE_INTERVAL_MS + 1);
     });
 
     // Phrase should cycle if PHRASE_CHANGE_INTERVAL_MS has passed
-    expect(WITTY_LOADING_PHRASES).toContain(
-      result.current.currentLoadingPhrase,
-    );
+    expect(MOCK_WITTY_PHRASES).toContain(result.current.currentLoadingPhrase);
   });
 
   it('should show waiting phrase and retain elapsedTime when WaitingForConfirmation', async () => {
@@ -104,9 +101,7 @@ describe('useLoadingIndicator', () => {
       rerender({ streamingState: StreamingState.Responding });
     });
     expect(result.current.elapsedTime).toBe(0); // Should reset
-    expect(WITTY_LOADING_PHRASES).toContain(
-      result.current.currentLoadingPhrase,
-    );
+    expect(MOCK_WITTY_PHRASES).toContain(result.current.currentLoadingPhrase);
 
     await act(async () => {
       await vi.advanceTimersByTimeAsync(1000);
@@ -130,9 +125,7 @@ describe('useLoadingIndicator', () => {
     });
 
     expect(result.current.elapsedTime).toBe(0);
-    expect(WITTY_LOADING_PHRASES).toContain(
-      result.current.currentLoadingPhrase,
-    );
+    expect(MOCK_WITTY_PHRASES).toContain(result.current.currentLoadingPhrase);
 
     // Timer should not advance
     await act(async () => {

packages/cli/src/ui/hooks/usePhraseCycler.test.ts

@@ -8,13 +8,17 @@ import { describe, it, expect, vi, beforeEach, afterEach } from 'vitest';
 import { renderHook, act } from '@testing-library/react';
 import {
   usePhraseCycler,
-  WITTY_LOADING_PHRASES,
   PHRASE_CHANGE_INTERVAL_MS,
 } from './usePhraseCycler.js';
+import * as i18n from '../../i18n/index.js';
+
+const MOCK_WITTY_PHRASES = ['Phrase 1', 'Phrase 2', 'Phrase 3'];
 
 describe('usePhraseCycler', () => {
   beforeEach(() => {
     vi.useFakeTimers();
+    vi.spyOn(i18n, 'ta').mockReturnValue(MOCK_WITTY_PHRASES);
+    vi.spyOn(i18n, 't').mockImplementation((key) => key);
   });
 
   afterEach(() => {
@@ -23,7 +27,7 @@ describe('usePhraseCycler', () => {
 
   it('should initialize with a witty phrase when not active and not waiting', () => {
     const { result } = renderHook(() => usePhraseCycler(false, false));
-    expect(WITTY_LOADING_PHRASES).toContain(result.current);
+    expect(MOCK_WITTY_PHRASES).toContain(result.current);
   });
 
   it('should show "Waiting for user confirmation..." when isWaiting is true', () => {
@@ -47,35 +51,30 @@ describe('usePhraseCycler', () => {
   it('should cycle through witty phrases when isActive is true and not waiting', () => {
     const { result } = renderHook(() => usePhraseCycler(true, false));
     // Initial phrase should be one of the witty phrases
-    expect(WITTY_LOADING_PHRASES).toContain(result.current);
+    expect(MOCK_WITTY_PHRASES).toContain(result.current);
     const _initialPhrase = result.current;
 
     act(() => {
       vi.advanceTimersByTime(PHRASE_CHANGE_INTERVAL_MS);
     });
     // Phrase should change and be one of the witty phrases
-    expect(WITTY_LOADING_PHRASES).toContain(result.current);
+    expect(MOCK_WITTY_PHRASES).toContain(result.current);
 
     const _secondPhrase = result.current;
     act(() => {
       vi.advanceTimersByTime(PHRASE_CHANGE_INTERVAL_MS);
     });
-    expect(WITTY_LOADING_PHRASES).toContain(result.current);
+    expect(MOCK_WITTY_PHRASES).toContain(result.current);
   });
 
   it('should reset to a witty phrase when isActive becomes true after being false (and not waiting)', () => {
-    // Ensure there are at least two phrases for this test to be meaningful.
-    if (WITTY_LOADING_PHRASES.length < 2) {
-      return;
-    }
-
     // Mock Math.random to make the test deterministic.
     let callCount = 0;
     vi.spyOn(Math, 'random').mockImplementation(() => {
       // Cycle through 0, 1, 0, 1, ...
       const val = callCount % 2;
       callCount++;
-      return val / WITTY_LOADING_PHRASES.length;
+      return val / MOCK_WITTY_PHRASES.length;
     });
 
     const { result, rerender } = renderHook(
@@ -86,9 +85,9 @@ describe('usePhraseCycler', () => {
     // Activate
     rerender({ isActive: true, isWaiting: false });
     const firstActivePhrase = result.current;
-    expect(WITTY_LOADING_PHRASES).toContain(firstActivePhrase);
+    expect(MOCK_WITTY_PHRASES).toContain(firstActivePhrase);
     // With our mock, this should be the first phrase.
-    expect(firstActivePhrase).toBe(WITTY_LOADING_PHRASES[0]);
+    expect(firstActivePhrase).toBe(MOCK_WITTY_PHRASES[0]);
 
     act(() => {
       vi.advanceTimersByTime(PHRASE_CHANGE_INTERVAL_MS);
@@ -96,18 +95,18 @@ describe('usePhraseCycler', () => {
 
     // Phrase should change to the second phrase.
     expect(result.current).not.toBe(firstActivePhrase);
-    expect(result.current).toBe(WITTY_LOADING_PHRASES[1]);
+    expect(result.current).toBe(MOCK_WITTY_PHRASES[1]);
 
     // Set to inactive - should reset to the default initial phrase
     rerender({ isActive: false, isWaiting: false });
-    expect(WITTY_LOADING_PHRASES).toContain(result.current);
+    expect(MOCK_WITTY_PHRASES).toContain(result.current);
 
     // Set back to active - should pick a random witty phrase (which our mock controls)
     act(() => {
       rerender({ isActive: true, isWaiting: false });
     });
     // The random mock will now return 0, so it should be the first phrase again.
-    expect(result.current).toBe(WITTY_LOADING_PHRASES[0]);
+    expect(result.current).toBe(MOCK_WITTY_PHRASES[0]);
   });
 
   it('should clear phrase interval on unmount when active', () => {
@@ -148,7 +147,7 @@ describe('usePhraseCycler', () => {
 
     rerender({ isActive: true, isWaiting: false, customPhrases: undefined });
 
-    expect(WITTY_LOADING_PHRASES).toContain(result.current);
+    expect(MOCK_WITTY_PHRASES).toContain(result.current);
   });
 
   it('should fall back to witty phrases if custom phrases are an empty array', () => {
@@ -164,7 +163,7 @@ describe('usePhraseCycler', () => {
       },
     );
 
-    expect(WITTY_LOADING_PHRASES).toContain(result.current);
+    expect(MOCK_WITTY_PHRASES).toContain(result.current);
   });
 
   it('should reset to a witty phrase when transitioning from waiting to active', () => {
@@ -174,16 +173,13 @@ describe('usePhraseCycler', () => {
     );
 
     const _initialPhrase = result.current;
-    expect(WITTY_LOADING_PHRASES).toContain(_initialPhrase);
+    expect(MOCK_WITTY_PHRASES).toContain(_initialPhrase);
 
     // Cycle to a different phrase (potentially)
     act(() => {
       vi.advanceTimersByTime(PHRASE_CHANGE_INTERVAL_MS);
     });
-    if (WITTY_LOADING_PHRASES.length > 1) {
-      // This check is probabilistic with random selection
-    }
-    expect(WITTY_LOADING_PHRASES).toContain(result.current);
+    expect(MOCK_WITTY_PHRASES).toContain(result.current);
 
     // Go to waiting state
     rerender({ isActive: false, isWaiting: true });
@@ -191,6 +187,6 @@ describe('usePhraseCycler', () => {
 
     // Go back to active cycling - should pick a random witty phrase
     rerender({ isActive: true, isWaiting: false });
-    expect(WITTY_LOADING_PHRASES).toContain(result.current);
+    expect(MOCK_WITTY_PHRASES).toContain(result.current);
   });
 });

packages/cli/src/ui/hooks/usePhraseCycler.ts

@@ -5,139 +5,9 @@
  */
 
 import { useState, useEffect, useRef, useMemo } from 'react';
-import { t } from '../../i18n/index.js';
+import { t, ta } from '../../i18n/index.js';
 
-export const WITTY_LOADING_PHRASES = [
-  "I'm Feeling Lucky",
-  'Shipping awesomeness... ',
-  'Painting the serifs back on...',
-  'Navigating the slime mold...',
-  'Consulting the digital spirits...',
-  'Reticulating splines...',
-  'Warming up the AI hamsters...',
-  'Asking the magic conch shell...',
-  'Generating witty retort...',
-  'Polishing the algorithms...',
-  "Don't rush perfection (or my code)...",
-  'Brewing fresh bytes...',
-  'Counting electrons...',
-  'Engaging cognitive processors...',
-  'Checking for syntax errors in the universe...',
-  'One moment, optimizing humor...',
-  'Shuffling punchlines...',
-  'Untangling neural nets...',
-  'Compiling brilliance...',
-  'Loading wit.exe...',
-  'Summoning the cloud of wisdom...',
-  'Preparing a witty response...',
-  "Just a sec, I'm debugging reality...",
-  'Confuzzling the options...',
-  'Tuning the cosmic frequencies...',
-  'Crafting a response worthy of your patience...',
-  'Compiling the 1s and 0s...',
-  'Resolving dependencies... and existential crises...',
-  'Defragmenting memories... both RAM and personal...',
-  'Rebooting the humor module...',
-  'Caching the essentials (mostly cat memes)...',
-  'Optimizing for ludicrous speed',
-  "Swapping bits... don't tell the bytes...",
-  'Garbage collecting... be right back...',
-  'Assembling the interwebs...',
-  'Converting coffee into code...',
-  'Updating the syntax for reality...',
-  'Rewiring the synapses...',
-  'Looking for a misplaced semicolon...',
-  "Greasin' the cogs of the machine...",
-  'Pre-heating the servers...',
-  'Calibrating the flux capacitor...',
-  'Engaging the improbability drive...',
-  'Channeling the Force...',
-  'Aligning the stars for optimal response...',
-  'So say we all...',
-  'Loading the next great idea...',
-  "Just a moment, I'm in the zone...",
-  'Preparing to dazzle you with brilliance...',
-  "Just a tick, I'm polishing my wit...",
-  "Hold tight, I'm crafting a masterpiece...",
-  "Just a jiffy, I'm debugging the universe...",
-  "Just a moment, I'm aligning the pixels...",
-  "Just a sec, I'm optimizing the humor...",
-  "Just a moment, I'm tuning the algorithms...",
-  'Warp speed engaged...',
-  'Mining for more Dilithium crystals...',
-  "Don't panic...",
-  'Following the white rabbit...',
-  'The truth is in here... somewhere...',
-  'Blowing on the cartridge...',
-  'Loading... Do a barrel roll!',
-  'Waiting for the respawn...',
-  'Finishing the Kessel Run in less than 12 parsecs...',
-  "The cake is not a lie, it's just still loading...",
-  'Fiddling with the character creation screen...',
-  "Just a moment, I'm finding the right meme...",
-  "Pressing 'A' to continue...",
-  'Herding digital cats...',
-  'Polishing the pixels...',
-  'Finding a suitable loading screen pun...',
-  'Distracting you with this witty phrase...',
-  'Almost there... probably...',
-  'Our hamsters are working as fast as they can...',
-  'Giving Cloudy a pat on the head...',
-  'Petting the cat...',
-  'Rickrolling my boss...',
-  'Never gonna give you up, never gonna let you down...',
-  'Slapping the bass...',
-  'Tasting the snozberries...',
-  "I'm going the distance, I'm going for speed...",
-  'Is this the real life? Is this just fantasy?...',
-  "I've got a good feeling about this...",
-  'Poking the bear...',
-  'Doing research on the latest memes...',
-  'Figuring out how to make this more witty...',
-  'Hmmm... let me think...',
-  'What do you call a fish with no eyes? A fsh...',
-  'Why did the computer go to therapy? It had too many bytes...',
-  "Why don't programmers like nature? It has too many bugs...",
-  'Why do programmers prefer dark mode? Because light attracts bugs...',
-  'Why did the developer go broke? Because they used up all their cache...',
-  "What can you do with a broken pencil? Nothing, it's pointless...",
-  'Applying percussive maintenance...',
-  'Searching for the correct USB orientation...',
-  'Ensuring the magic smoke stays inside the wires...',
-  'Rewriting in Rust for no particular reason...',
-  'Trying to exit Vim...',
-  'Spinning up the hamster wheel...',
-  "That's not a bug, it's an undocumented feature...",
-  'Engage.',
-  "I'll be back... with an answer.",
-  'My other process is a TARDIS...',
-  'Communing with the machine spirit...',
-  'Letting the thoughts marinate...',
-  'Just remembered where I put my keys...',
-  'Pondering the orb...',
-  "I've seen things you people wouldn't believe... like a user who reads loading messages.",
-  'Initiating thoughtful gaze...',
-  "What's a computer's favorite snack? Microchips.",
-  "Why do Java developers wear glasses? Because they don't C#.",
-  'Charging the laser... pew pew!',
-  'Dividing by zero... just kidding!',
-  'Looking for an adult superviso... I mean, processing.',
-  'Making it go beep boop.',
-  'Buffering... because even AIs need a moment.',
-  'Entangling quantum particles for a faster response...',
-  'Polishing the chrome... on the algorithms.',
-  'Are you not entertained? (Working on it!)',
-  'Summoning the code gremlins... to help, of course.',
-  'Just waiting for the dial-up tone to finish...',
-  'Recalibrating the humor-o-meter.',
-  'My other loading screen is even funnier.',
-  "Pretty sure there's a cat walking on the keyboard somewhere...",
-  'Enhancing... Enhancing... Still loading.',
-  "It's not a bug, it's a feature... of this loading screen.",
-  'Have you tried turning it off and on again? (The loading screen, not me.)',
-  'Constructing additional pylons...',
-  'New line? That’s Ctrl+J.',
-];
+export const WITTY_LOADING_PHRASES: string[] = ["I'm Feeling Lucky"];
 
 export const PHRASE_CHANGE_INTERVAL_MS = 15000;
 
@@ -152,14 +22,16 @@ export const usePhraseCycler = (
   isWaiting: boolean,
   customPhrases?: string[],
 ) => {
-  // Translate all phrases at once if using default phrases
-  const loadingPhrases = useMemo(
-    () =>
-      customPhrases && customPhrases.length > 0
-        ? customPhrases
-        : WITTY_LOADING_PHRASES.map((phrase) => t(phrase)),
-    [customPhrases],
-  );
+  // Get phrases from translations if available
+  const loadingPhrases = useMemo(() => {
+    if (customPhrases && customPhrases.length > 0) {
+      return customPhrases;
+    }
+    const translatedPhrases = ta('WITTY_LOADING_PHRASES');
+    return translatedPhrases.length > 0
+      ? translatedPhrases
+      : WITTY_LOADING_PHRASES;
+  }, [customPhrases]);
 
   const [currentLoadingPhrase, setCurrentLoadingPhrase] = useState(
     loadingPhrases[0],

packages/cli/src/ui/hooks/useQuotaAndFallback.test.ts

@@ -0,0 +1,391 @@
+/**
+ * @license
+ * Copyright 2025 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import {
+  vi,
+  describe,
+  it,
+  expect,
+  beforeEach,
+  afterEach,
+  type Mock,
+} from 'vitest';
+import { act, renderHook } from '@testing-library/react';
+import {
+  type Config,
+  type FallbackModelHandler,
+  UserTierId,
+  AuthType,
+  isGenericQuotaExceededError,
+  isProQuotaExceededError,
+  makeFakeConfig,
+} from '@qwen-code/qwen-code-core';
+import { useQuotaAndFallback } from './useQuotaAndFallback.js';
+import type { UseHistoryManagerReturn } from './useHistoryManager.js';
+import { AuthState, MessageType } from '../types.js';
+
+// Mock the error checking functions from the core package to control test scenarios
+vi.mock('@qwen-code/qwen-code-core', async (importOriginal) => {
+  const original =
+    await importOriginal<typeof import('@qwen-code/qwen-code-core')>();
+  return {
+    ...original,
+    isGenericQuotaExceededError: vi.fn(),
+    isProQuotaExceededError: vi.fn(),
+  };
+});
+
+// Use a type alias for SpyInstance as it's not directly exported
+type SpyInstance = ReturnType<typeof vi.spyOn>;
+
+describe('useQuotaAndFallback', () => {
+  let mockConfig: Config;
+  let mockHistoryManager: UseHistoryManagerReturn;
+  let mockSetAuthState: Mock;
+  let mockSetModelSwitchedFromQuotaError: Mock;
+  let setFallbackHandlerSpy: SpyInstance;
+
+  const mockedIsGenericQuotaExceededError = isGenericQuotaExceededError as Mock;
+  const mockedIsProQuotaExceededError = isProQuotaExceededError as Mock;
+
+  beforeEach(() => {
+    mockConfig = makeFakeConfig();
+
+    // Spy on the method that requires the private field and mock its return.
+    // This is cleaner than modifying the config class for tests.
+    vi.spyOn(mockConfig, 'getContentGeneratorConfig').mockReturnValue({
+      model: 'test-model',
+      authType: AuthType.LOGIN_WITH_GOOGLE,
+    });
+
+    mockHistoryManager = {
+      addItem: vi.fn(),
+      history: [],
+      updateItem: vi.fn(),
+      clearItems: vi.fn(),
+      loadHistory: vi.fn(),
+    };
+    mockSetAuthState = vi.fn();
+    mockSetModelSwitchedFromQuotaError = vi.fn();
+
+    setFallbackHandlerSpy = vi.spyOn(mockConfig, 'setFallbackModelHandler');
+    vi.spyOn(mockConfig, 'setQuotaErrorOccurred');
+
+    mockedIsGenericQuotaExceededError.mockReturnValue(false);
+    mockedIsProQuotaExceededError.mockReturnValue(false);
+  });
+
+  afterEach(() => {
+    vi.clearAllMocks();
+  });
+
+  it('should register a fallback handler on initialization', () => {
+    renderHook(() =>
+      useQuotaAndFallback({
+        config: mockConfig,
+        historyManager: mockHistoryManager,
+        userTier: UserTierId.FREE,
+        setAuthState: mockSetAuthState,
+        setModelSwitchedFromQuotaError: mockSetModelSwitchedFromQuotaError,
+      }),
+    );
+
+    expect(setFallbackHandlerSpy).toHaveBeenCalledTimes(1);
+    expect(setFallbackHandlerSpy.mock.calls[0][0]).toBeInstanceOf(Function);
+  });
+
+  describe('Fallback Handler Logic', () => {
+    // Helper function to render the hook and extract the registered handler
+    const getRegisteredHandler = (
+      userTier: UserTierId = UserTierId.FREE,
+    ): FallbackModelHandler => {
+      renderHook(
+        (props) =>
+          useQuotaAndFallback({
+            config: mockConfig,
+            historyManager: mockHistoryManager,
+            userTier: props.userTier,
+            setAuthState: mockSetAuthState,
+            setModelSwitchedFromQuotaError: mockSetModelSwitchedFromQuotaError,
+          }),
+        { initialProps: { userTier } },
+      );
+      return setFallbackHandlerSpy.mock.calls[0][0] as FallbackModelHandler;
+    };
+
+    it('should return null and take no action if already in fallback mode', async () => {
+      vi.spyOn(mockConfig, 'isInFallbackMode').mockReturnValue(true);
+      const handler = getRegisteredHandler();
+      const result = await handler('gemini-pro', 'gemini-flash', new Error());
+
+      expect(result).toBeNull();
+      expect(mockHistoryManager.addItem).not.toHaveBeenCalled();
+    });
+
+    it('should return null and take no action if authType is not LOGIN_WITH_GOOGLE', async () => {
+      // Override the default mock from beforeEach for this specific test
+      vi.spyOn(mockConfig, 'getContentGeneratorConfig').mockReturnValue({
+        model: 'test-model',
+        authType: AuthType.USE_GEMINI,
+      });
+
+      const handler = getRegisteredHandler();
+      const result = await handler('gemini-pro', 'gemini-flash', new Error());
+
+      expect(result).toBeNull();
+      expect(mockHistoryManager.addItem).not.toHaveBeenCalled();
+    });
+
+    describe('Automatic Fallback Scenarios', () => {
+      const testCases = [
+        {
+          errorType: 'generic',
+          tier: UserTierId.FREE,
+          expectedMessageSnippets: [
+            'Automatically switching from model-A to model-B',
+            'upgrade to a Gemini Code Assist Standard or Enterprise plan',
+          ],
+        },
+        {
+          errorType: 'generic',
+          tier: UserTierId.STANDARD, // Paid tier
+          expectedMessageSnippets: [
+            'Automatically switching from model-A to model-B',
+            'switch to using a paid API key from AI Studio',
+          ],
+        },
+        {
+          errorType: 'other',
+          tier: UserTierId.FREE,
+          expectedMessageSnippets: [
+            'Automatically switching from model-A to model-B for faster responses',
+            'upgrade to a Gemini Code Assist Standard or Enterprise plan',
+          ],
+        },
+        {
+          errorType: 'other',
+          tier: UserTierId.LEGACY, // Paid tier
+          expectedMessageSnippets: [
+            'Automatically switching from model-A to model-B for faster responses',
+            'switch to using a paid API key from AI Studio',
+          ],
+        },
+      ];
+
+      for (const { errorType, tier, expectedMessageSnippets } of testCases) {
+        it(`should handle ${errorType} error for ${tier} tier correctly`, async () => {
+          mockedIsGenericQuotaExceededError.mockReturnValue(
+            errorType === 'generic',
+          );
+
+          const handler = getRegisteredHandler(tier);
+          const result = await handler(
+            'model-A',
+            'model-B',
+            new Error('quota exceeded'),
+          );
+
+          // Automatic fallbacks should return 'stop'
+          expect(result).toBe('stop');
+
+          expect(mockHistoryManager.addItem).toHaveBeenCalledWith(
+            expect.objectContaining({ type: MessageType.INFO }),
+            expect.any(Number),
+          );
+
+          const message = (mockHistoryManager.addItem as Mock).mock.calls[0][0]
+            .text;
+          for (const snippet of expectedMessageSnippets) {
+            expect(message).toContain(snippet);
+          }
+
+          expect(mockSetModelSwitchedFromQuotaError).toHaveBeenCalledWith(true);
+          expect(mockConfig.setQuotaErrorOccurred).toHaveBeenCalledWith(true);
+        });
+      }
+    });
+
+    describe('Interactive Fallback (Pro Quota Error)', () => {
+      beforeEach(() => {
+        mockedIsProQuotaExceededError.mockReturnValue(true);
+      });
+
+      it('should set an interactive request and wait for user choice', async () => {
+        const { result } = renderHook(() =>
+          useQuotaAndFallback({
+            config: mockConfig,
+            historyManager: mockHistoryManager,
+            userTier: UserTierId.FREE,
+            setAuthState: mockSetAuthState,
+            setModelSwitchedFromQuotaError: mockSetModelSwitchedFromQuotaError,
+          }),
+        );
+
+        const handler = setFallbackHandlerSpy.mock
+          .calls[0][0] as FallbackModelHandler;
+
+        // Call the handler but do not await it, to check the intermediate state
+        const promise = handler(
+          'gemini-pro',
+          'gemini-flash',
+          new Error('pro quota'),
+        );
+
+        await act(async () => {});
+
+        // The hook should now have a pending request for the UI to handle
+        expect(result.current.proQuotaRequest).not.toBeNull();
+        expect(result.current.proQuotaRequest?.failedModel).toBe('gemini-pro');
+
+        // Simulate the user choosing to continue with the fallback model
+        act(() => {
+          result.current.handleProQuotaChoice('continue');
+        });
+
+        // The original promise from the handler should now resolve
+        const intent = await promise;
+        expect(intent).toBe('retry');
+
+        // The pending request should be cleared from the state
+        expect(result.current.proQuotaRequest).toBeNull();
+      });
+
+      it('should handle race conditions by stopping subsequent requests', async () => {
+        const { result } = renderHook(() =>
+          useQuotaAndFallback({
+            config: mockConfig,
+            historyManager: mockHistoryManager,
+            userTier: UserTierId.FREE,
+            setAuthState: mockSetAuthState,
+            setModelSwitchedFromQuotaError: mockSetModelSwitchedFromQuotaError,
+          }),
+        );
+
+        const handler = setFallbackHandlerSpy.mock
+          .calls[0][0] as FallbackModelHandler;
+
+        const promise1 = handler(
+          'gemini-pro',
+          'gemini-flash',
+          new Error('pro quota 1'),
+        );
+        await act(async () => {});
+
+        const firstRequest = result.current.proQuotaRequest;
+        expect(firstRequest).not.toBeNull();
+
+        const result2 = await handler(
+          'gemini-pro',
+          'gemini-flash',
+          new Error('pro quota 2'),
+        );
+
+        // The lock should have stopped the second request
+        expect(result2).toBe('stop');
+        expect(result.current.proQuotaRequest).toBe(firstRequest);
+
+        act(() => {
+          result.current.handleProQuotaChoice('continue');
+        });
+
+        const intent1 = await promise1;
+        expect(intent1).toBe('retry');
+        expect(result.current.proQuotaRequest).toBeNull();
+      });
+    });
+  });
+
+  describe('handleProQuotaChoice', () => {
+    beforeEach(() => {
+      mockedIsProQuotaExceededError.mockReturnValue(true);
+    });
+
+    it('should do nothing if there is no pending pro quota request', () => {
+      const { result } = renderHook(() =>
+        useQuotaAndFallback({
+          config: mockConfig,
+          historyManager: mockHistoryManager,
+          userTier: UserTierId.FREE,
+          setAuthState: mockSetAuthState,
+          setModelSwitchedFromQuotaError: mockSetModelSwitchedFromQuotaError,
+        }),
+      );
+
+      act(() => {
+        result.current.handleProQuotaChoice('auth');
+      });
+
+      expect(mockSetAuthState).not.toHaveBeenCalled();
+      expect(mockHistoryManager.addItem).not.toHaveBeenCalled();
+    });
+
+    it('should resolve intent to "auth" and trigger auth state update', async () => {
+      const { result } = renderHook(() =>
+        useQuotaAndFallback({
+          config: mockConfig,
+          historyManager: mockHistoryManager,
+          userTier: UserTierId.FREE,
+          setAuthState: mockSetAuthState,
+          setModelSwitchedFromQuotaError: mockSetModelSwitchedFromQuotaError,
+        }),
+      );
+
+      const handler = setFallbackHandlerSpy.mock
+        .calls[0][0] as FallbackModelHandler;
+      const promise = handler(
+        'gemini-pro',
+        'gemini-flash',
+        new Error('pro quota'),
+      );
+      await act(async () => {}); // Allow state to update
+
+      act(() => {
+        result.current.handleProQuotaChoice('auth');
+      });
+
+      const intent = await promise;
+      expect(intent).toBe('auth');
+      expect(mockSetAuthState).toHaveBeenCalledWith(AuthState.Updating);
+      expect(result.current.proQuotaRequest).toBeNull();
+    });
+
+    it('should resolve intent to "retry" and add info message on continue', async () => {
+      const { result } = renderHook(() =>
+        useQuotaAndFallback({
+          config: mockConfig,
+          historyManager: mockHistoryManager,
+          userTier: UserTierId.FREE,
+          setAuthState: mockSetAuthState,
+          setModelSwitchedFromQuotaError: mockSetModelSwitchedFromQuotaError,
+        }),
+      );
+
+      const handler = setFallbackHandlerSpy.mock
+        .calls[0][0] as FallbackModelHandler;
+      // The first `addItem` call is for the initial quota error message
+      const promise = handler(
+        'gemini-pro',
+        'gemini-flash',
+        new Error('pro quota'),
+      );
+      await act(async () => {}); // Allow state to update
+
+      act(() => {
+        result.current.handleProQuotaChoice('continue');
+      });
+
+      const intent = await promise;
+      expect(intent).toBe('retry');
+      expect(result.current.proQuotaRequest).toBeNull();
+
+      // Check for the second "Switched to fallback model" message
+      expect(mockHistoryManager.addItem).toHaveBeenCalledTimes(2);
+      const lastCall = (mockHistoryManager.addItem as Mock).mock.calls[1][0];
+      expect(lastCall.type).toBe(MessageType.INFO);
+      expect(lastCall.text).toContain('Switched to fallback model.');
+    });
+  });
+});

packages/cli/src/ui/hooks/useQuotaAndFallback.ts

@@ -0,0 +1,175 @@
+/**
+ * @license
+ * Copyright 2025 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import {
+  AuthType,
+  type Config,
+  type FallbackModelHandler,
+  type FallbackIntent,
+  isGenericQuotaExceededError,
+  isProQuotaExceededError,
+  UserTierId,
+} from '@qwen-code/qwen-code-core';
+import { useCallback, useEffect, useRef, useState } from 'react';
+import { type UseHistoryManagerReturn } from './useHistoryManager.js';
+import { AuthState, MessageType } from '../types.js';
+import { type ProQuotaDialogRequest } from '../contexts/UIStateContext.js';
+
+interface UseQuotaAndFallbackArgs {
+  config: Config;
+  historyManager: UseHistoryManagerReturn;
+  userTier: UserTierId | undefined;
+  setAuthState: (state: AuthState) => void;
+  setModelSwitchedFromQuotaError: (value: boolean) => void;
+}
+
+export function useQuotaAndFallback({
+  config,
+  historyManager,
+  userTier,
+  setAuthState,
+  setModelSwitchedFromQuotaError,
+}: UseQuotaAndFallbackArgs) {
+  const [proQuotaRequest, setProQuotaRequest] =
+    useState<ProQuotaDialogRequest | null>(null);
+  const isDialogPending = useRef(false);
+
+  // Set up Flash fallback handler
+  useEffect(() => {
+    const fallbackHandler: FallbackModelHandler = async (
+      failedModel,
+      fallbackModel,
+      error,
+    ): Promise<FallbackIntent | null> => {
+      if (config.isInFallbackMode()) {
+        return null;
+      }
+
+      // Fallbacks are currently only handled for OAuth users.
+      const contentGeneratorConfig = config.getContentGeneratorConfig();
+      if (
+        !contentGeneratorConfig ||
+        contentGeneratorConfig.authType !== AuthType.LOGIN_WITH_GOOGLE
+      ) {
+        return null;
+      }
+
+      // Use actual user tier if available; otherwise, default to FREE tier behavior (safe default)
+      const isPaidTier =
+        userTier === UserTierId.LEGACY || userTier === UserTierId.STANDARD;
+
+      let message: string;
+
+      if (error && isProQuotaExceededError(error)) {
+        // Pro Quota specific messages (Interactive)
+        if (isPaidTier) {
+          message = `⚡ You have reached your daily ${failedModel} quota limit.
+⚡ You can choose to authenticate with a paid API key or continue with the fallback model.
+⚡ To continue accessing the ${failedModel} model today, consider using /auth to switch to using a paid API key from AI Studio at https://aistudio.google.com/apikey`;
+        } else {
+          message = `⚡ You have reached your daily ${failedModel} quota limit.
+⚡ You can choose to authenticate with a paid API key or continue with the fallback model.
+⚡ To increase your limits, upgrade to a Gemini Code Assist Standard or Enterprise plan with higher limits at https://goo.gle/set-up-gemini-code-assist
+⚡ Or you can utilize a Gemini API Key. See: https://goo.gle/gemini-cli-docs-auth#gemini-api-key
+⚡ You can switch authentication methods by typing /auth`;
+        }
+      } else if (error && isGenericQuotaExceededError(error)) {
+        // Generic Quota (Automatic fallback)
+        const actionMessage = `⚡ You have reached your daily quota limit.\n⚡ Automatically switching from ${failedModel} to ${fallbackModel} for the remainder of this session.`;
+
+        if (isPaidTier) {
+          message = `${actionMessage}
+⚡ To continue accessing the ${failedModel} model today, consider using /auth to switch to using a paid API key from AI Studio at https://aistudio.google.com/apikey`;
+        } else {
+          message = `${actionMessage}
+⚡ To increase your limits, upgrade to a Gemini Code Assist Standard or Enterprise plan with higher limits at https://goo.gle/set-up-gemini-code-assist
+⚡ Or you can utilize a Gemini API Key. See: https://goo.gle/gemini-cli-docs-auth#gemini-api-key
+⚡ You can switch authentication methods by typing /auth`;
+        }
+      } else {
+        // Consecutive 429s or other errors (Automatic fallback)
+        const actionMessage = `⚡ Automatically switching from ${failedModel} to ${fallbackModel} for faster responses for the remainder of this session.`;
+
+        if (isPaidTier) {
+          message = `${actionMessage}
+⚡ Possible reasons for this are that you have received multiple consecutive capacity errors or you have reached your daily ${failedModel} quota limit
+⚡ To continue accessing the ${failedModel} model today, consider using /auth to switch to using a paid API key from AI Studio at https://aistudio.google.com/apikey`;
+        } else {
+          message = `${actionMessage}
+⚡ Possible reasons for this are that you have received multiple consecutive capacity errors or you have reached your daily ${failedModel} quota limit
+⚡ To increase your limits, upgrade to a Gemini Code Assist Standard or Enterprise plan with higher limits at https://goo.gle/set-up-gemini-code-assist
+⚡ Or you can utilize a Gemini API Key. See: https://goo.gle/gemini-cli-docs-auth#gemini-api-key
+⚡ You can switch authentication methods by typing /auth`;
+        }
+      }
+
+      // Add message to UI history
+      historyManager.addItem(
+        {
+          type: MessageType.INFO,
+          text: message,
+        },
+        Date.now(),
+      );
+
+      setModelSwitchedFromQuotaError(true);
+      config.setQuotaErrorOccurred(true);
+
+      // Interactive Fallback for Pro quota
+      if (error && isProQuotaExceededError(error)) {
+        if (isDialogPending.current) {
+          return 'stop'; // A dialog is already active, so just stop this request.
+        }
+        isDialogPending.current = true;
+
+        const intent: FallbackIntent = await new Promise<FallbackIntent>(
+          (resolve) => {
+            setProQuotaRequest({
+              failedModel,
+              fallbackModel,
+              resolve,
+            });
+          },
+        );
+
+        return intent;
+      }
+
+      return 'stop';
+    };
+
+    config.setFallbackModelHandler(fallbackHandler);
+  }, [config, historyManager, userTier, setModelSwitchedFromQuotaError]);
+
+  const handleProQuotaChoice = useCallback(
+    (choice: 'auth' | 'continue') => {
+      if (!proQuotaRequest) return;
+
+      const intent: FallbackIntent = choice === 'auth' ? 'auth' : 'retry';
+      proQuotaRequest.resolve(intent);
+      setProQuotaRequest(null);
+      isDialogPending.current = false; // Reset the flag here
+
+      if (choice === 'auth') {
+        setAuthState(AuthState.Updating);
+      } else {
+        historyManager.addItem(
+          {
+            type: MessageType.INFO,
+            text: 'Switched to fallback model. Tip: Press Ctrl+P (or Up Arrow) to recall your previous prompt and submit it again if you wish.',
+          },
+          Date.now(),
+        );
+      }
+    },
+    [proQuotaRequest, setAuthState, historyManager],
+  );
+
+  return {
+    proQuotaRequest,
+    handleProQuotaChoice,
+  };
+}

packages/cli/src/ui/hooks/useQwenAuth.test.ts

@@ -411,7 +411,7 @@ describe('useQwenAuth', () => {
     expect(geminiResult.current.qwenAuthState.authStatus).toBe('idle');
 
     const { result: oauthResult } = renderHook(() =>
-      useQwenAuth(AuthType.USE_OPENAI, true),
+      useQwenAuth(AuthType.LOGIN_WITH_GOOGLE, true),
     );
     expect(oauthResult.current.qwenAuthState.authStatus).toBe('idle');
   });

packages/cli/src/ui/hooks/useToolScheduler.test.ts

@@ -62,7 +62,7 @@ const mockConfig = {
   getAllowedTools: vi.fn(() => []),
   getContentGeneratorConfig: () => ({
     model: 'test-model',
-    authType: 'gemini-api-key',
+    authType: 'oauth-personal',
   }),
   getUseSmartEdit: () => false,
   getUseModelRouter: () => false,

packages/cli/src/validateNonInterActiveAuth.ts

@@ -21,13 +21,6 @@ function getAuthTypeFromEnv(): AuthType | undefined {
     return AuthType.QWEN_OAUTH;
   }
 
-  if (process.env['GEMINI_API_KEY']) {
-    return AuthType.USE_GEMINI;
-  }
-  if (process.env['GOOGLE_API_KEY']) {
-    return AuthType.USE_VERTEX_AI;
-  }
-
   return undefined;
 }
 

packages/core/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@qwen-code/qwen-code-core",
-  "version": "0.6.0-preview.0",
+  "version": "0.6.0",
   "description": "Qwen Code Core",
   "repository": {
     "type": "git",
@@ -23,8 +23,8 @@
     "scripts/postinstall.js"
   ],
   "dependencies": {
-    "@google/genai": "1.30.0",
-    "@modelcontextprotocol/sdk": "^1.25.1",
+    "@google/genai": "1.16.0",
+    "@modelcontextprotocol/sdk": "^1.11.0",
     "@opentelemetry/api": "^1.9.0",
     "async-mutex": "^0.5.0",
     "@opentelemetry/exporter-logs-otlp-grpc": "^0.203.0",
@@ -34,6 +34,7 @@
     "@opentelemetry/exporter-trace-otlp-grpc": "^0.203.0",
     "@opentelemetry/exporter-trace-otlp-http": "^0.203.0",
     "@opentelemetry/instrumentation-http": "^0.203.0",
+    "@opentelemetry/resource-detector-gcp": "^0.40.0",
     "@opentelemetry/sdk-node": "^0.203.0",
     "@types/html-to-text": "^9.0.4",
     "@xterm/headless": "5.5.0",
@@ -47,7 +48,7 @@
     "fdir": "^6.4.6",
     "fzf": "^0.5.2",
     "glob": "^10.5.0",
-    "google-auth-library": "^10.5.0",
+    "google-auth-library": "^9.11.0",
     "html-to-text": "^9.0.5",
     "https-proxy-agent": "^7.0.6",
     "ignore": "^7.0.0",

packages/core/src/code_assist/codeAssist.ts

@@ -0,0 +1,54 @@
+/**
+ * @license
+ * Copyright 2025 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import type { ContentGenerator } from '../core/contentGenerator.js';
+import { AuthType } from '../core/contentGenerator.js';
+import { getOauthClient } from './oauth2.js';
+import { setupUser } from './setup.js';
+import type { HttpOptions } from './server.js';
+import { CodeAssistServer } from './server.js';
+import type { Config } from '../config/config.js';
+import { LoggingContentGenerator } from '../core/loggingContentGenerator.js';
+
+export async function createCodeAssistContentGenerator(
+  httpOptions: HttpOptions,
+  authType: AuthType,
+  config: Config,
+  sessionId?: string,
+): Promise<ContentGenerator> {
+  if (
+    authType === AuthType.LOGIN_WITH_GOOGLE ||
+    authType === AuthType.CLOUD_SHELL
+  ) {
+    const authClient = await getOauthClient(authType, config);
+    const userData = await setupUser(authClient);
+    return new CodeAssistServer(
+      authClient,
+      userData.projectId,
+      httpOptions,
+      sessionId,
+      userData.userTier,
+    );
+  }
+
+  throw new Error(`Unsupported authType: ${authType}`);
+}
+
+export function getCodeAssistServer(
+  config: Config,
+): CodeAssistServer | undefined {
+  let server = config.getContentGenerator();
+
+  // Unwrap LoggingContentGenerator if present
+  if (server instanceof LoggingContentGenerator) {
+    server = server.getWrapped();
+  }
+
+  if (!(server instanceof CodeAssistServer)) {
+    return undefined;
+  }
+  return server;
+}

packages/core/src/code_assist/converter.test.ts

@@ -0,0 +1,456 @@
+/**
+ * @license
+ * Copyright 2025 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import { describe, it, expect } from 'vitest';
+import type { CaGenerateContentResponse } from './converter.js';
+import {
+  toGenerateContentRequest,
+  fromGenerateContentResponse,
+  toContents,
+} from './converter.js';
+import type {
+  ContentListUnion,
+  GenerateContentParameters,
+} from '@google/genai';
+import {
+  GenerateContentResponse,
+  FinishReason,
+  BlockedReason,
+  type Part,
+} from '@google/genai';
+
+describe('converter', () => {
+  describe('toCodeAssistRequest', () => {
+    it('should convert a simple request with project', () => {
+      const genaiReq: GenerateContentParameters = {
+        model: 'gemini-pro',
+        contents: [{ role: 'user', parts: [{ text: 'Hello' }] }],
+      };
+      const codeAssistReq = toGenerateContentRequest(
+        genaiReq,
+        'my-prompt',
+        'my-project',
+        'my-session',
+      );
+      expect(codeAssistReq).toEqual({
+        model: 'gemini-pro',
+        project: 'my-project',
+        request: {
+          contents: [{ role: 'user', parts: [{ text: 'Hello' }] }],
+          systemInstruction: undefined,
+          cachedContent: undefined,
+          tools: undefined,
+          toolConfig: undefined,
+          labels: undefined,
+          safetySettings: undefined,
+          generationConfig: undefined,
+          session_id: 'my-session',
+        },
+        user_prompt_id: 'my-prompt',
+      });
+    });
+
+    it('should convert a request without a project', () => {
+      const genaiReq: GenerateContentParameters = {
+        model: 'gemini-pro',
+        contents: [{ role: 'user', parts: [{ text: 'Hello' }] }],
+      };
+      const codeAssistReq = toGenerateContentRequest(
+        genaiReq,
+        'my-prompt',
+        undefined,
+        'my-session',
+      );
+      expect(codeAssistReq).toEqual({
+        model: 'gemini-pro',
+        project: undefined,
+        request: {
+          contents: [{ role: 'user', parts: [{ text: 'Hello' }] }],
+          systemInstruction: undefined,
+          cachedContent: undefined,
+          tools: undefined,
+          toolConfig: undefined,
+          labels: undefined,
+          safetySettings: undefined,
+          generationConfig: undefined,
+          session_id: 'my-session',
+        },
+        user_prompt_id: 'my-prompt',
+      });
+    });
+
+    it('should convert a request with sessionId', () => {
+      const genaiReq: GenerateContentParameters = {
+        model: 'gemini-pro',
+        contents: [{ role: 'user', parts: [{ text: 'Hello' }] }],
+      };
+      const codeAssistReq = toGenerateContentRequest(
+        genaiReq,
+        'my-prompt',
+        'my-project',
+        'session-123',
+      );
+      expect(codeAssistReq).toEqual({
+        model: 'gemini-pro',
+        project: 'my-project',
+        request: {
+          contents: [{ role: 'user', parts: [{ text: 'Hello' }] }],
+          systemInstruction: undefined,
+          cachedContent: undefined,
+          tools: undefined,
+          toolConfig: undefined,
+          labels: undefined,
+          safetySettings: undefined,
+          generationConfig: undefined,
+          session_id: 'session-123',
+        },
+        user_prompt_id: 'my-prompt',
+      });
+    });
+
+    it('should handle string content', () => {
+      const genaiReq: GenerateContentParameters = {
+        model: 'gemini-pro',
+        contents: 'Hello',
+      };
+      const codeAssistReq = toGenerateContentRequest(
+        genaiReq,
+        'my-prompt',
+        'my-project',
+        'my-session',
+      );
+      expect(codeAssistReq.request.contents).toEqual([
+        { role: 'user', parts: [{ text: 'Hello' }] },
+      ]);
+    });
+
+    it('should handle Part[] content', () => {
+      const genaiReq: GenerateContentParameters = {
+        model: 'gemini-pro',
+        contents: [{ text: 'Hello' }, { text: 'World' }],
+      };
+      const codeAssistReq = toGenerateContentRequest(
+        genaiReq,
+        'my-prompt',
+        'my-project',
+        'my-session',
+      );
+      expect(codeAssistReq.request.contents).toEqual([
+        { role: 'user', parts: [{ text: 'Hello' }] },
+        { role: 'user', parts: [{ text: 'World' }] },
+      ]);
+    });
+
+    it('should handle system instructions', () => {
+      const genaiReq: GenerateContentParameters = {
+        model: 'gemini-pro',
+        contents: 'Hello',
+        config: {
+          systemInstruction: 'You are a helpful assistant.',
+        },
+      };
+      const codeAssistReq = toGenerateContentRequest(
+        genaiReq,
+        'my-prompt',
+        'my-project',
+        'my-session',
+      );
+      expect(codeAssistReq.request.systemInstruction).toEqual({
+        role: 'user',
+        parts: [{ text: 'You are a helpful assistant.' }],
+      });
+    });
+
+    it('should handle generation config', () => {
+      const genaiReq: GenerateContentParameters = {
+        model: 'gemini-pro',
+        contents: 'Hello',
+        config: {
+          temperature: 0.8,
+          topK: 40,
+        },
+      };
+      const codeAssistReq = toGenerateContentRequest(
+        genaiReq,
+        'my-prompt',
+        'my-project',
+        'my-session',
+      );
+      expect(codeAssistReq.request.generationConfig).toEqual({
+        temperature: 0.8,
+        topK: 40,
+      });
+    });
+
+    it('should handle all generation config fields', () => {
+      const genaiReq: GenerateContentParameters = {
+        model: 'gemini-pro',
+        contents: 'Hello',
+        config: {
+          temperature: 0.1,
+          topP: 0.2,
+          topK: 3,
+          candidateCount: 4,
+          maxOutputTokens: 5,
+          stopSequences: ['a'],
+          responseLogprobs: true,
+          logprobs: 6,
+          presencePenalty: 0.7,
+          frequencyPenalty: 0.8,
+          seed: 9,
+          responseMimeType: 'application/json',
+        },
+      };
+      const codeAssistReq = toGenerateContentRequest(
+        genaiReq,
+        'my-prompt',
+        'my-project',
+        'my-session',
+      );
+      expect(codeAssistReq.request.generationConfig).toEqual({
+        temperature: 0.1,
+        topP: 0.2,
+        topK: 3,
+        candidateCount: 4,
+        maxOutputTokens: 5,
+        stopSequences: ['a'],
+        responseLogprobs: true,
+        logprobs: 6,
+        presencePenalty: 0.7,
+        frequencyPenalty: 0.8,
+        seed: 9,
+        responseMimeType: 'application/json',
+      });
+    });
+  });
+
+  describe('fromCodeAssistResponse', () => {
+    it('should convert a simple response', () => {
+      const codeAssistRes: CaGenerateContentResponse = {
+        response: {
+          candidates: [
+            {
+              index: 0,
+              content: {
+                role: 'model',
+                parts: [{ text: 'Hi there!' }],
+              },
+              finishReason: FinishReason.STOP,
+              safetyRatings: [],
+            },
+          ],
+        },
+      };
+      const genaiRes = fromGenerateContentResponse(codeAssistRes);
+      expect(genaiRes).toBeInstanceOf(GenerateContentResponse);
+      expect(genaiRes.candidates).toEqual(codeAssistRes.response.candidates);
+    });
+
+    it('should handle prompt feedback and usage metadata', () => {
+      const codeAssistRes: CaGenerateContentResponse = {
+        response: {
+          candidates: [],
+          promptFeedback: {
+            blockReason: BlockedReason.SAFETY,
+            safetyRatings: [],
+          },
+          usageMetadata: {
+            promptTokenCount: 10,
+            candidatesTokenCount: 20,
+            totalTokenCount: 30,
+          },
+        },
+      };
+      const genaiRes = fromGenerateContentResponse(codeAssistRes);
+      expect(genaiRes.promptFeedback).toEqual(
+        codeAssistRes.response.promptFeedback,
+      );
+      expect(genaiRes.usageMetadata).toEqual(
+        codeAssistRes.response.usageMetadata,
+      );
+    });
+
+    it('should handle automatic function calling history', () => {
+      const codeAssistRes: CaGenerateContentResponse = {
+        response: {
+          candidates: [],
+          automaticFunctionCallingHistory: [
+            {
+              role: 'model',
+              parts: [
+                {
+                  functionCall: {
+                    name: 'test_function',
+                    args: {
+                      foo: 'bar',
+                    },
+                  },
+                },
+              ],
+            },
+          ],
+        },
+      };
+      const genaiRes = fromGenerateContentResponse(codeAssistRes);
+      expect(genaiRes.automaticFunctionCallingHistory).toEqual(
+        codeAssistRes.response.automaticFunctionCallingHistory,
+      );
+    });
+
+    it('should handle modelVersion', () => {
+      const codeAssistRes: CaGenerateContentResponse = {
+        response: {
+          candidates: [],
+          modelVersion: 'qwen3-coder-plus',
+        },
+      };
+      const genaiRes = fromGenerateContentResponse(codeAssistRes);
+      expect(genaiRes.modelVersion).toEqual('qwen3-coder-plus');
+    });
+  });
+
+  describe('toContents', () => {
+    it('should handle Content', () => {
+      const content: ContentListUnion = {
+        role: 'user',
+        parts: [{ text: 'hello' }],
+      };
+      expect(toContents(content)).toEqual([
+        { role: 'user', parts: [{ text: 'hello' }] },
+      ]);
+    });
+
+    it('should handle array of Contents', () => {
+      const contents: ContentListUnion = [
+        { role: 'user', parts: [{ text: 'hello' }] },
+        { role: 'model', parts: [{ text: 'hi' }] },
+      ];
+      expect(toContents(contents)).toEqual([
+        { role: 'user', parts: [{ text: 'hello' }] },
+        { role: 'model', parts: [{ text: 'hi' }] },
+      ]);
+    });
+
+    it('should handle Part', () => {
+      const part: ContentListUnion = { text: 'a part' };
+      expect(toContents(part)).toEqual([
+        { role: 'user', parts: [{ text: 'a part' }] },
+      ]);
+    });
+
+    it('should handle array of Parts', () => {
+      const parts = [{ text: 'part 1' }, 'part 2'];
+      expect(toContents(parts)).toEqual([
+        { role: 'user', parts: [{ text: 'part 1' }] },
+        { role: 'user', parts: [{ text: 'part 2' }] },
+      ]);
+    });
+
+    it('should handle string', () => {
+      const str: ContentListUnion = 'a string';
+      expect(toContents(str)).toEqual([
+        { role: 'user', parts: [{ text: 'a string' }] },
+      ]);
+    });
+
+    it('should handle array of strings', () => {
+      const strings: ContentListUnion = ['string 1', 'string 2'];
+      expect(toContents(strings)).toEqual([
+        { role: 'user', parts: [{ text: 'string 1' }] },
+        { role: 'user', parts: [{ text: 'string 2' }] },
+      ]);
+    });
+
+    it('should convert thought parts to text parts for API compatibility', () => {
+      const contentWithThought: ContentListUnion = {
+        role: 'model',
+        parts: [
+          { text: 'regular text' },
+          { thought: 'thinking about the problem' } as Part & {
+            thought: string;
+          },
+          { text: 'more text' },
+        ],
+      };
+      expect(toContents(contentWithThought)).toEqual([
+        {
+          role: 'model',
+          parts: [
+            { text: 'regular text' },
+            { text: '[Thought: thinking about the problem]' },
+            { text: 'more text' },
+          ],
+        },
+      ]);
+    });
+
+    it('should combine text and thought for text parts with thoughts', () => {
+      const contentWithTextAndThought: ContentListUnion = {
+        role: 'model',
+        parts: [
+          {
+            text: 'Here is my response',
+            thought: 'I need to be careful here',
+          } as Part & { thought: string },
+        ],
+      };
+      expect(toContents(contentWithTextAndThought)).toEqual([
+        {
+          role: 'model',
+          parts: [
+            {
+              text: 'Here is my response\n[Thought: I need to be careful here]',
+            },
+          ],
+        },
+      ]);
+    });
+
+    it('should preserve non-thought properties while removing thought', () => {
+      const contentWithComplexPart: ContentListUnion = {
+        role: 'model',
+        parts: [
+          {
+            functionCall: { name: 'calculate', args: { x: 5, y: 10 } },
+            thought: 'Performing calculation',
+          } as Part & { thought: string },
+        ],
+      };
+      expect(toContents(contentWithComplexPart)).toEqual([
+        {
+          role: 'model',
+          parts: [
+            {
+              functionCall: { name: 'calculate', args: { x: 5, y: 10 } },
+            },
+          ],
+        },
+      ]);
+    });
+
+    it('should convert invalid text content to valid text part with thought', () => {
+      const contentWithInvalidText: ContentListUnion = {
+        role: 'model',
+        parts: [
+          {
+            text: 123, // Invalid - should be string
+            thought: 'Processing number',
+          } as Part & { thought: string; text: number },
+        ],
+      };
+      expect(toContents(contentWithInvalidText)).toEqual([
+        {
+          role: 'model',
+          parts: [
+            {
+              text: '123\n[Thought: Processing number]',
+            },
+          ],
+        },
+      ]);
+    });
+  });
+});

packages/core/src/code_assist/converter.ts

@@ -0,0 +1,285 @@
+/**
+ * @license
+ * Copyright 2025 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import type {
+  Content,
+  ContentListUnion,
+  ContentUnion,
+  GenerateContentConfig,
+  GenerateContentParameters,
+  CountTokensParameters,
+  CountTokensResponse,
+  GenerationConfigRoutingConfig,
+  MediaResolution,
+  Candidate,
+  ModelSelectionConfig,
+  GenerateContentResponsePromptFeedback,
+  GenerateContentResponseUsageMetadata,
+  Part,
+  SafetySetting,
+  PartUnion,
+  SpeechConfigUnion,
+  ThinkingConfig,
+  ToolListUnion,
+  ToolConfig,
+} from '@google/genai';
+import { GenerateContentResponse } from '@google/genai';
+
+export interface CAGenerateContentRequest {
+  model: string;
+  project?: string;
+  user_prompt_id?: string;
+  request: VertexGenerateContentRequest;
+}
+
+interface VertexGenerateContentRequest {
+  contents: Content[];
+  systemInstruction?: Content;
+  cachedContent?: string;
+  tools?: ToolListUnion;
+  toolConfig?: ToolConfig;
+  labels?: Record<string, string>;
+  safetySettings?: SafetySetting[];
+  generationConfig?: VertexGenerationConfig;
+  session_id?: string;
+}
+
+interface VertexGenerationConfig {
+  temperature?: number;
+  topP?: number;
+  topK?: number;
+  candidateCount?: number;
+  maxOutputTokens?: number;
+  stopSequences?: string[];
+  responseLogprobs?: boolean;
+  logprobs?: number;
+  presencePenalty?: number;
+  frequencyPenalty?: number;
+  seed?: number;
+  responseMimeType?: string;
+  responseJsonSchema?: unknown;
+  responseSchema?: unknown;
+  routingConfig?: GenerationConfigRoutingConfig;
+  modelSelectionConfig?: ModelSelectionConfig;
+  responseModalities?: string[];
+  mediaResolution?: MediaResolution;
+  speechConfig?: SpeechConfigUnion;
+  audioTimestamp?: boolean;
+  thinkingConfig?: ThinkingConfig;
+}
+
+export interface CaGenerateContentResponse {
+  response: VertexGenerateContentResponse;
+}
+
+interface VertexGenerateContentResponse {
+  candidates: Candidate[];
+  automaticFunctionCallingHistory?: Content[];
+  promptFeedback?: GenerateContentResponsePromptFeedback;
+  usageMetadata?: GenerateContentResponseUsageMetadata;
+  modelVersion?: string;
+}
+
+export interface CaCountTokenRequest {
+  request: VertexCountTokenRequest;
+}
+
+interface VertexCountTokenRequest {
+  model: string;
+  contents: Content[];
+}
+
+export interface CaCountTokenResponse {
+  totalTokens: number;
+}
+
+export function toCountTokenRequest(
+  req: CountTokensParameters,
+): CaCountTokenRequest {
+  return {
+    request: {
+      model: 'models/' + req.model,
+      contents: toContents(req.contents),
+    },
+  };
+}
+
+export function fromCountTokenResponse(
+  res: CaCountTokenResponse,
+): CountTokensResponse {
+  return {
+    totalTokens: res.totalTokens,
+  };
+}
+
+export function toGenerateContentRequest(
+  req: GenerateContentParameters,
+  userPromptId: string,
+  project?: string,
+  sessionId?: string,
+): CAGenerateContentRequest {
+  return {
+    model: req.model,
+    project,
+    user_prompt_id: userPromptId,
+    request: toVertexGenerateContentRequest(req, sessionId),
+  };
+}
+
+export function fromGenerateContentResponse(
+  res: CaGenerateContentResponse,
+): GenerateContentResponse {
+  const inres = res.response;
+  const out = new GenerateContentResponse();
+  out.candidates = inres.candidates;
+  out.automaticFunctionCallingHistory = inres.automaticFunctionCallingHistory;
+  out.promptFeedback = inres.promptFeedback;
+  out.usageMetadata = inres.usageMetadata;
+  out.modelVersion = inres.modelVersion;
+  return out;
+}
+
+function toVertexGenerateContentRequest(
+  req: GenerateContentParameters,
+  sessionId?: string,
+): VertexGenerateContentRequest {
+  return {
+    contents: toContents(req.contents),
+    systemInstruction: maybeToContent(req.config?.systemInstruction),
+    cachedContent: req.config?.cachedContent,
+    tools: req.config?.tools,
+    toolConfig: req.config?.toolConfig,
+    labels: req.config?.labels,
+    safetySettings: req.config?.safetySettings,
+    generationConfig: toVertexGenerationConfig(req.config),
+    session_id: sessionId,
+  };
+}
+
+export function toContents(contents: ContentListUnion): Content[] {
+  if (Array.isArray(contents)) {
+    // it's a Content[] or a PartsUnion[]
+    return contents.map(toContent);
+  }
+  // it's a Content or a PartsUnion
+  return [toContent(contents)];
+}
+
+function maybeToContent(content?: ContentUnion): Content | undefined {
+  if (!content) {
+    return undefined;
+  }
+  return toContent(content);
+}
+
+function toContent(content: ContentUnion): Content {
+  if (Array.isArray(content)) {
+    // it's a PartsUnion[]
+    return {
+      role: 'user',
+      parts: toParts(content),
+    };
+  }
+  if (typeof content === 'string') {
+    // it's a string
+    return {
+      role: 'user',
+      parts: [{ text: content }],
+    };
+  }
+  if ('parts' in content) {
+    // it's a Content - process parts to handle thought filtering
+    return {
+      ...content,
+      parts: content.parts
+        ? toParts(content.parts.filter((p) => p != null))
+        : [],
+    };
+  }
+  // it's a Part
+  return {
+    role: 'user',
+    parts: [toPart(content as Part)],
+  };
+}
+
+export function toParts(parts: PartUnion[]): Part[] {
+  return parts.map(toPart);
+}
+
+function toPart(part: PartUnion): Part {
+  if (typeof part === 'string') {
+    // it's a string
+    return { text: part };
+  }
+
+  // Handle thought parts for CountToken API compatibility
+  // The CountToken API expects parts to have certain required "oneof" fields initialized,
+  // but thought parts don't conform to this schema and cause API failures
+  if ('thought' in part && part.thought) {
+    const thoughtText = `[Thought: ${part.thought}]`;
+
+    const newPart = { ...part };
+    delete (newPart as Record<string, unknown>)['thought'];
+
+    const hasApiContent =
+      'functionCall' in newPart ||
+      'functionResponse' in newPart ||
+      'inlineData' in newPart ||
+      'fileData' in newPart;
+
+    if (hasApiContent) {
+      // It's a functionCall or other non-text part. Just strip the thought.
+      return newPart;
+    }
+
+    // If no other valid API content, this must be a text part.
+    // Combine existing text (if any) with the thought, preserving other properties.
+    const text = (newPart as { text?: unknown }).text;
+    const existingText = text ? String(text) : '';
+    const combinedText = existingText
+      ? `${existingText}\n${thoughtText}`
+      : thoughtText;
+
+    return {
+      ...newPart,
+      text: combinedText,
+    };
+  }
+
+  return part;
+}
+
+function toVertexGenerationConfig(
+  config?: GenerateContentConfig,
+): VertexGenerationConfig | undefined {
+  if (!config) {
+    return undefined;
+  }
+  return {
+    temperature: config.temperature,
+    topP: config.topP,
+    topK: config.topK,
+    candidateCount: config.candidateCount,
+    maxOutputTokens: config.maxOutputTokens,
+    stopSequences: config.stopSequences,
+    responseLogprobs: config.responseLogprobs,
+    logprobs: config.logprobs,
+    presencePenalty: config.presencePenalty,
+    frequencyPenalty: config.frequencyPenalty,
+    seed: config.seed,
+    responseMimeType: config.responseMimeType,
+    responseSchema: config.responseSchema,
+    responseJsonSchema: config.responseJsonSchema,
+    routingConfig: config.routingConfig,
+    modelSelectionConfig: config.modelSelectionConfig,
+    responseModalities: config.responseModalities,
+    mediaResolution: config.mediaResolution,
+    speechConfig: config.speechConfig,
+    audioTimestamp: config.audioTimestamp,
+    thinkingConfig: config.thinkingConfig,
+  };
+}

packages/core/src/code_assist/oauth-credential-storage.test.ts

@@ -0,0 +1,217 @@
+/**
+ * @license
+ * Copyright 2025 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import { type Credentials } from 'google-auth-library';
+import { describe, it, expect, vi, beforeEach, afterEach } from 'vitest';
+import { OAuthCredentialStorage } from './oauth-credential-storage.js';
+import type { OAuthCredentials } from '../mcp/token-storage/types.js';
+
+import * as path from 'node:path';
+import * as os from 'node:os';
+import { promises as fs } from 'node:fs';
+
+// Mock external dependencies
+const mockHybridTokenStorage = vi.hoisted(() => ({
+  getCredentials: vi.fn(),
+  setCredentials: vi.fn(),
+  deleteCredentials: vi.fn(),
+}));
+vi.mock('../mcp/token-storage/hybrid-token-storage.js', () => ({
+  HybridTokenStorage: vi.fn(() => mockHybridTokenStorage),
+}));
+vi.mock('node:fs', () => ({
+  promises: {
+    readFile: vi.fn(),
+    rm: vi.fn(),
+  },
+}));
+vi.mock('node:os');
+vi.mock('node:path');
+
+describe('OAuthCredentialStorage', () => {
+  const mockCredentials: Credentials = {
+    access_token: 'mock_access_token',
+    refresh_token: 'mock_refresh_token',
+    expiry_date: Date.now() + 3600 * 1000,
+    token_type: 'Bearer',
+    scope: 'email profile',
+  };
+
+  const mockMcpCredentials: OAuthCredentials = {
+    serverName: 'main-account',
+    token: {
+      accessToken: 'mock_access_token',
+      refreshToken: 'mock_refresh_token',
+      tokenType: 'Bearer',
+      scope: 'email profile',
+      expiresAt: mockCredentials.expiry_date!,
+    },
+    updatedAt: expect.any(Number),
+  };
+
+  const oldFilePath = '/mock/home/.qwen/oauth.json';
+
+  beforeEach(() => {
+    vi.spyOn(mockHybridTokenStorage, 'getCredentials').mockResolvedValue(null);
+    vi.spyOn(mockHybridTokenStorage, 'setCredentials').mockResolvedValue(
+      undefined,
+    );
+    vi.spyOn(mockHybridTokenStorage, 'deleteCredentials').mockResolvedValue(
+      undefined,
+    );
+
+    vi.spyOn(fs, 'readFile').mockRejectedValue(new Error('File not found'));
+    vi.spyOn(fs, 'rm').mockResolvedValue(undefined);
+
+    vi.spyOn(os, 'homedir').mockReturnValue('/mock/home');
+    vi.spyOn(path, 'join').mockReturnValue(oldFilePath);
+  });
+
+  afterEach(() => {
+    vi.restoreAllMocks();
+  });
+
+  describe('loadCredentials', () => {
+    it('should load credentials from HybridTokenStorage if available', async () => {
+      vi.spyOn(mockHybridTokenStorage, 'getCredentials').mockResolvedValue(
+        mockMcpCredentials,
+      );
+
+      const result = await OAuthCredentialStorage.loadCredentials();
+
+      expect(mockHybridTokenStorage.getCredentials).toHaveBeenCalledWith(
+        'main-account',
+      );
+      expect(result).toEqual(mockCredentials);
+    });
+
+    it('should fallback to migrateFromFileStorage if no credentials in HybridTokenStorage', async () => {
+      vi.spyOn(mockHybridTokenStorage, 'getCredentials').mockResolvedValue(
+        null,
+      );
+      vi.spyOn(fs, 'readFile').mockResolvedValue(
+        JSON.stringify(mockCredentials),
+      );
+
+      const result = await OAuthCredentialStorage.loadCredentials();
+
+      expect(mockHybridTokenStorage.getCredentials).toHaveBeenCalledWith(
+        'main-account',
+      );
+      expect(fs.readFile).toHaveBeenCalledWith(oldFilePath, 'utf-8');
+      expect(mockHybridTokenStorage.setCredentials).toHaveBeenCalled(); // Verify credentials were saved
+      expect(fs.rm).toHaveBeenCalledWith(oldFilePath, { force: true }); // Verify old file was removed
+      expect(result).toEqual(mockCredentials);
+    });
+
+    it('should return null if no credentials found and no old file to migrate', async () => {
+      vi.spyOn(fs, 'readFile').mockRejectedValue({
+        message: 'File not found',
+        code: 'ENOENT',
+      });
+
+      const result = await OAuthCredentialStorage.loadCredentials();
+
+      expect(result).toBeNull();
+    });
+
+    it('should throw an error if loading fails', async () => {
+      vi.spyOn(mockHybridTokenStorage, 'getCredentials').mockRejectedValue(
+        new Error('Loading error'),
+      );
+
+      await expect(OAuthCredentialStorage.loadCredentials()).rejects.toThrow(
+        'Failed to load OAuth credentials',
+      );
+    });
+
+    it('should throw an error if read file fails', async () => {
+      vi.spyOn(mockHybridTokenStorage, 'getCredentials').mockResolvedValue(
+        null,
+      );
+      vi.spyOn(fs, 'readFile').mockRejectedValue(
+        new Error('Permission denied'),
+      );
+
+      await expect(OAuthCredentialStorage.loadCredentials()).rejects.toThrow(
+        'Failed to load OAuth credentials',
+      );
+    });
+
+    it('should not throw error if migration file removal failed', async () => {
+      vi.spyOn(mockHybridTokenStorage, 'getCredentials').mockResolvedValue(
+        null,
+      );
+      vi.spyOn(fs, 'readFile').mockResolvedValue(
+        JSON.stringify(mockCredentials),
+      );
+      vi.spyOn(OAuthCredentialStorage, 'saveCredentials').mockResolvedValue(
+        undefined,
+      );
+      vi.spyOn(fs, 'rm').mockRejectedValue(new Error('Deletion failed'));
+
+      const result = await OAuthCredentialStorage.loadCredentials();
+
+      expect(result).toEqual(mockCredentials);
+    });
+  });
+
+  describe('saveCredentials', () => {
+    it('should save credentials to HybridTokenStorage', async () => {
+      await OAuthCredentialStorage.saveCredentials(mockCredentials);
+
+      expect(mockHybridTokenStorage.setCredentials).toHaveBeenCalledWith(
+        mockMcpCredentials,
+      );
+    });
+
+    it('should throw an error if access_token is missing', async () => {
+      const invalidCredentials: Credentials = {
+        ...mockCredentials,
+        access_token: undefined,
+      };
+      await expect(
+        OAuthCredentialStorage.saveCredentials(invalidCredentials),
+      ).rejects.toThrow(
+        'Attempted to save credentials without an access token.',
+      );
+    });
+  });
+
+  describe('clearCredentials', () => {
+    it('should delete credentials from HybridTokenStorage', async () => {
+      await OAuthCredentialStorage.clearCredentials();
+
+      expect(mockHybridTokenStorage.deleteCredentials).toHaveBeenCalledWith(
+        'main-account',
+      );
+    });
+
+    it('should attempt to remove the old file-based storage', async () => {
+      await OAuthCredentialStorage.clearCredentials();
+
+      expect(fs.rm).toHaveBeenCalledWith(oldFilePath, { force: true });
+    });
+
+    it('should not throw an error if deleting old file fails', async () => {
+      vi.spyOn(fs, 'rm').mockRejectedValue(new Error('File deletion failed'));
+
+      await expect(
+        OAuthCredentialStorage.clearCredentials(),
+      ).resolves.toBeUndefined();
+    });
+
+    it('should throw an error if clearing from HybridTokenStorage fails', async () => {
+      vi.spyOn(mockHybridTokenStorage, 'deleteCredentials').mockRejectedValue(
+        new Error('Deletion error'),
+      );
+
+      await expect(OAuthCredentialStorage.clearCredentials()).rejects.toThrow(
+        'Failed to clear OAuth credentials',
+      );
+    });
+  });
+});

packages/core/src/code_assist/oauth-credential-storage.ts

@@ -0,0 +1,130 @@
+/**
+ * @license
+ * Copyright 2025 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import { type Credentials } from 'google-auth-library';
+import { HybridTokenStorage } from '../mcp/token-storage/hybrid-token-storage.js';
+import { OAUTH_FILE } from '../config/storage.js';
+import type { OAuthCredentials } from '../mcp/token-storage/types.js';
+import * as path from 'node:path';
+import * as os from 'node:os';
+import { promises as fs } from 'node:fs';
+
+const QWEN_DIR = '.qwen';
+const KEYCHAIN_SERVICE_NAME = 'qwen-code-oauth';
+const MAIN_ACCOUNT_KEY = 'main-account';
+
+export class OAuthCredentialStorage {
+  private static storage: HybridTokenStorage = new HybridTokenStorage(
+    KEYCHAIN_SERVICE_NAME,
+  );
+
+  /**
+   * Load cached OAuth credentials
+   */
+  static async loadCredentials(): Promise<Credentials | null> {
+    try {
+      const credentials = await this.storage.getCredentials(MAIN_ACCOUNT_KEY);
+
+      if (credentials?.token) {
+        const { accessToken, refreshToken, expiresAt, tokenType, scope } =
+          credentials.token;
+        // Convert from OAuthCredentials format to Google Credentials format
+        const googleCreds: Credentials = {
+          access_token: accessToken,
+          refresh_token: refreshToken || undefined,
+          token_type: tokenType || undefined,
+          scope: scope || undefined,
+        };
+
+        if (expiresAt) {
+          googleCreds.expiry_date = expiresAt;
+        }
+
+        return googleCreds;
+      }
+
+      // Fallback: Try to migrate from old file-based storage
+      return await this.migrateFromFileStorage();
+    } catch (error: unknown) {
+      console.error(error);
+      throw new Error('Failed to load OAuth credentials');
+    }
+  }
+
+  /**
+   * Save OAuth credentials
+   */
+  static async saveCredentials(credentials: Credentials): Promise<void> {
+    if (!credentials.access_token) {
+      throw new Error('Attempted to save credentials without an access token.');
+    }
+
+    // Convert Google Credentials to OAuthCredentials format
+    const mcpCredentials: OAuthCredentials = {
+      serverName: MAIN_ACCOUNT_KEY,
+      token: {
+        accessToken: credentials.access_token,
+        refreshToken: credentials.refresh_token || undefined,
+        tokenType: credentials.token_type || 'Bearer',
+        scope: credentials.scope || undefined,
+        expiresAt: credentials.expiry_date || undefined,
+      },
+      updatedAt: Date.now(),
+    };
+
+    await this.storage.setCredentials(mcpCredentials);
+  }
+
+  /**
+   * Clear cached OAuth credentials
+   */
+  static async clearCredentials(): Promise<void> {
+    try {
+      await this.storage.deleteCredentials(MAIN_ACCOUNT_KEY);
+
+      // Also try to remove the old file if it exists
+      const oldFilePath = path.join(os.homedir(), QWEN_DIR, OAUTH_FILE);
+      await fs.rm(oldFilePath, { force: true }).catch(() => {});
+    } catch (error: unknown) {
+      console.error(error);
+      throw new Error('Failed to clear OAuth credentials');
+    }
+  }
+
+  /**
+   * Migrate credentials from old file-based storage to keychain
+   */
+  private static async migrateFromFileStorage(): Promise<Credentials | null> {
+    const oldFilePath = path.join(os.homedir(), QWEN_DIR, OAUTH_FILE);
+
+    let credsJson: string;
+    try {
+      credsJson = await fs.readFile(oldFilePath, 'utf-8');
+    } catch (error: unknown) {
+      if (
+        typeof error === 'object' &&
+        error !== null &&
+        'code' in error &&
+        error.code === 'ENOENT'
+      ) {
+        // File doesn't exist, so no migration.
+        return null;
+      }
+      // Other read errors should propagate.
+      throw error;
+    }
+
+    const credentials = JSON.parse(credsJson) as Credentials;
+
+    // Save to new storage
+    await this.saveCredentials(credentials);
+
+    // Remove old file after successful migration
+    await fs.rm(oldFilePath, { force: true }).catch(() => {});
+
+    return credentials;
+  }
+}

packages/core/src/code_assist/oauth2.ts

@@ -0,0 +1,563 @@
+/**
+ * @license
+ * Copyright 2025 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import type { Credentials } from 'google-auth-library';
+import {
+  CodeChallengeMethod,
+  Compute,
+  OAuth2Client,
+} from 'google-auth-library';
+import crypto from 'node:crypto';
+import { promises as fs } from 'node:fs';
+import * as http from 'node:http';
+import * as net from 'node:net';
+import path from 'node:path';
+import readline from 'node:readline';
+import url from 'node:url';
+import open from 'open';
+import type { Config } from '../config/config.js';
+import { Storage } from '../config/storage.js';
+import { AuthType } from '../core/contentGenerator.js';
+import { FatalAuthenticationError, getErrorMessage } from '../utils/errors.js';
+import { UserAccountManager } from '../utils/userAccountManager.js';
+import { OAuthCredentialStorage } from './oauth-credential-storage.js';
+import { FORCE_ENCRYPTED_FILE_ENV_VAR } from '../mcp/token-storage/index.js';
+
+const userAccountManager = new UserAccountManager();
+
+//  OAuth Client ID used to initiate OAuth2Client class.
+const OAUTH_CLIENT_ID =
+  '681255809395-oo8ft2oprdrnp9e3aqf6av3hmdib135j.apps.googleusercontent.com';
+
+// OAuth Secret value used to initiate OAuth2Client class.
+// Note: It's ok to save this in git because this is an installed application
+// as described here: https://developers.google.com/identity/protocols/oauth2#installed
+// "The process results in a client ID and, in some cases, a client secret,
+// which you embed in the source code of your application. (In this context,
+// the client secret is obviously not treated as a secret.)"
+const OAUTH_CLIENT_SECRET = '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';
+
+// OAuth Scopes for Cloud Code authorization.
+const OAUTH_SCOPE = [
+  'https://www.googleapis.com/auth/cloud-platform',
+  'https://www.googleapis.com/auth/userinfo.email',
+  'https://www.googleapis.com/auth/userinfo.profile',
+];
+
+const HTTP_REDIRECT = 301;
+const SIGN_IN_SUCCESS_URL =
+  'https://developers.google.com/gemini-code-assist/auth_success_gemini';
+const SIGN_IN_FAILURE_URL =
+  'https://developers.google.com/gemini-code-assist/auth_failure_gemini';
+
+/**
+ * An Authentication URL for updating the credentials of a Oauth2Client
+ * as well as a promise that will resolve when the credentials have
+ * been refreshed (or which throws error when refreshing credentials failed).
+ */
+export interface OauthWebLogin {
+  authUrl: string;
+  loginCompletePromise: Promise<void>;
+}
+
+const oauthClientPromises = new Map<AuthType, Promise<OAuth2Client>>();
+
+function getUseEncryptedStorageFlag() {
+  return process.env[FORCE_ENCRYPTED_FILE_ENV_VAR] === 'true';
+}
+
+async function initOauthClient(
+  authType: AuthType,
+  config: Config,
+): Promise<OAuth2Client> {
+  const client = new OAuth2Client({
+    clientId: OAUTH_CLIENT_ID,
+    clientSecret: OAUTH_CLIENT_SECRET,
+    transporterOptions: {
+      proxy: config.getProxy(),
+    },
+  });
+  const useEncryptedStorage = getUseEncryptedStorageFlag();
+
+  if (
+    process.env['GOOGLE_GENAI_USE_GCA'] &&
+    process.env['GOOGLE_CLOUD_ACCESS_TOKEN']
+  ) {
+    client.setCredentials({
+      access_token: process.env['GOOGLE_CLOUD_ACCESS_TOKEN'],
+    });
+    await fetchAndCacheUserInfo(client);
+    return client;
+  }
+
+  client.on('tokens', async (tokens: Credentials) => {
+    if (useEncryptedStorage) {
+      await OAuthCredentialStorage.saveCredentials(tokens);
+    } else {
+      await cacheCredentials(tokens);
+    }
+  });
+
+  // If there are cached creds on disk, they always take precedence
+  if (await loadCachedCredentials(client)) {
+    // Found valid cached credentials.
+    // Check if we need to retrieve Google Account ID or Email
+    if (!userAccountManager.getCachedGoogleAccount()) {
+      try {
+        await fetchAndCacheUserInfo(client);
+      } catch (error) {
+        // Non-fatal, continue with existing auth.
+        console.warn('Failed to fetch user info:', getErrorMessage(error));
+      }
+    }
+    console.log('Loaded cached credentials.');
+    return client;
+  }
+
+  // In Google Cloud Shell, we can use Application Default Credentials (ADC)
+  // provided via its metadata server to authenticate non-interactively using
+  // the identity of the user logged into Cloud Shell.
+  if (authType === AuthType.CLOUD_SHELL) {
+    try {
+      console.log("Attempting to authenticate via Cloud Shell VM's ADC.");
+      const computeClient = new Compute({
+        // We can leave this empty, since the metadata server will provide
+        // the service account email.
+      });
+      await computeClient.getAccessToken();
+      console.log('Authentication successful.');
+
+      // Do not cache creds in this case; note that Compute client will handle its own refresh
+      return computeClient;
+    } catch (e) {
+      throw new Error(
+        `Could not authenticate using Cloud Shell credentials. Please select a different authentication method or ensure you are in a properly configured environment. Error: ${getErrorMessage(
+          e,
+        )}`,
+      );
+    }
+  }
+
+  if (config.isBrowserLaunchSuppressed()) {
+    let success = false;
+    const maxRetries = 2;
+    for (let i = 0; !success && i < maxRetries; i++) {
+      success = await authWithUserCode(client);
+      if (!success) {
+        console.error(
+          '\nFailed to authenticate with user code.',
+          i === maxRetries - 1 ? '' : 'Retrying...\n',
+        );
+      }
+    }
+    if (!success) {
+      throw new FatalAuthenticationError(
+        'Failed to authenticate with user code.',
+      );
+    }
+  } else {
+    const webLogin = await authWithWeb(client);
+
+    console.log(
+      `\n\nCode Assist login required.\n` +
+        `Attempting to open authentication page in your browser.\n` +
+        `Otherwise navigate to:\n\n${webLogin.authUrl}\n\n`,
+    );
+    try {
+      // Attempt to open the authentication URL in the default browser.
+      // We do not use the `wait` option here because the main script's execution
+      // is already paused by `loginCompletePromise`, which awaits the server callback.
+      const childProcess = await open(webLogin.authUrl);
+
+      // IMPORTANT: Attach an error handler to the returned child process.
+      // Without this, if `open` fails to spawn a process (e.g., `xdg-open` is not found
+      // in a minimal Docker container), it will emit an unhandled 'error' event,
+      // causing the entire Node.js process to crash.
+      childProcess.on('error', (error) => {
+        console.error(
+          'Failed to open browser automatically. Please try running again with NO_BROWSER=true set.',
+        );
+        console.error('Browser error details:', getErrorMessage(error));
+      });
+    } catch (err) {
+      console.error(
+        'An unexpected error occurred while trying to open the browser:',
+        getErrorMessage(err),
+        '\nThis might be due to browser compatibility issues or system configuration.',
+        '\nPlease try running again with NO_BROWSER=true set for manual authentication.',
+      );
+      throw new FatalAuthenticationError(
+        `Failed to open browser: ${getErrorMessage(err)}`,
+      );
+    }
+    console.log('Waiting for authentication...');
+
+    // Add timeout to prevent infinite waiting when browser tab gets stuck
+    const authTimeout = 5 * 60 * 1000; // 5 minutes timeout
+    const timeoutPromise = new Promise<never>((_, reject) => {
+      setTimeout(() => {
+        reject(
+          new FatalAuthenticationError(
+            'Authentication timed out after 5 minutes. The browser tab may have gotten stuck in a loading state. ' +
+              'Please try again or use NO_BROWSER=true for manual authentication.',
+          ),
+        );
+      }, authTimeout);
+    });
+
+    await Promise.race([webLogin.loginCompletePromise, timeoutPromise]);
+  }
+
+  return client;
+}
+
+export async function getOauthClient(
+  authType: AuthType,
+  config: Config,
+): Promise<OAuth2Client> {
+  if (!oauthClientPromises.has(authType)) {
+    oauthClientPromises.set(authType, initOauthClient(authType, config));
+  }
+  return oauthClientPromises.get(authType)!;
+}
+
+async function authWithUserCode(client: OAuth2Client): Promise<boolean> {
+  const redirectUri = 'https://codeassist.google.com/authcode';
+  const codeVerifier = await client.generateCodeVerifierAsync();
+  const state = crypto.randomBytes(32).toString('hex');
+  const authUrl: string = client.generateAuthUrl({
+    redirect_uri: redirectUri,
+    access_type: 'offline',
+    scope: OAUTH_SCOPE,
+    code_challenge_method: CodeChallengeMethod.S256,
+    code_challenge: codeVerifier.codeChallenge,
+    state,
+  });
+  console.log('Please visit the following URL to authorize the application:');
+  console.log('');
+  console.log(authUrl);
+  console.log('');
+
+  const code = await new Promise<string>((resolve) => {
+    const rl = readline.createInterface({
+      input: process.stdin,
+      output: process.stdout,
+    });
+    rl.question('Enter the authorization code: ', (code) => {
+      rl.close();
+      resolve(code.trim());
+    });
+  });
+
+  if (!code) {
+    console.error('Authorization code is required.');
+    return false;
+  }
+
+  try {
+    const { tokens } = await client.getToken({
+      code,
+      codeVerifier: codeVerifier.codeVerifier,
+      redirect_uri: redirectUri,
+    });
+    client.setCredentials(tokens);
+  } catch (error) {
+    console.error(
+      'Failed to authenticate with authorization code:',
+      getErrorMessage(error),
+    );
+    return false;
+  }
+  return true;
+}
+
+async function authWithWeb(client: OAuth2Client): Promise<OauthWebLogin> {
+  const port = await getAvailablePort();
+  // The hostname used for the HTTP server binding (e.g., '0.0.0.0' in Docker).
+  const host = process.env['OAUTH_CALLBACK_HOST'] || 'localhost';
+  // The `redirectUri` sent to Google's authorization server MUST use a loopback IP literal
+  // (i.e., 'localhost' or '127.0.0.1'). This is a strict security policy for credentials of
+  // type 'Desktop app' or 'Web application' (when using loopback flow) to mitigate
+  // authorization code interception attacks.
+  const redirectUri = `http://localhost:${port}/oauth2callback`;
+  const state = crypto.randomBytes(32).toString('hex');
+  const authUrl = client.generateAuthUrl({
+    redirect_uri: redirectUri,
+    access_type: 'offline',
+    scope: OAUTH_SCOPE,
+    state,
+  });
+
+  const loginCompletePromise = new Promise<void>((resolve, reject) => {
+    const server = http.createServer(async (req, res) => {
+      try {
+        if (req.url!.indexOf('/oauth2callback') === -1) {
+          res.writeHead(HTTP_REDIRECT, { Location: SIGN_IN_FAILURE_URL });
+          res.end();
+          reject(
+            new FatalAuthenticationError(
+              'OAuth callback not received. Unexpected request: ' + req.url,
+            ),
+          );
+        }
+        // acquire the code from the querystring, and close the web server.
+        const qs = new url.URL(req.url!, 'http://localhost:3000').searchParams;
+        if (qs.get('error')) {
+          res.writeHead(HTTP_REDIRECT, { Location: SIGN_IN_FAILURE_URL });
+          res.end();
+
+          const errorCode = qs.get('error');
+          const errorDescription =
+            qs.get('error_description') || 'No additional details provided';
+          reject(
+            new FatalAuthenticationError(
+              `Google OAuth error: ${errorCode}. ${errorDescription}`,
+            ),
+          );
+        } else if (qs.get('state') !== state) {
+          res.end('State mismatch. Possible CSRF attack');
+
+          reject(
+            new FatalAuthenticationError(
+              'OAuth state mismatch. Possible CSRF attack or browser session issue.',
+            ),
+          );
+        } else if (qs.get('code')) {
+          try {
+            const { tokens } = await client.getToken({
+              code: qs.get('code')!,
+              redirect_uri: redirectUri,
+            });
+            client.setCredentials(tokens);
+
+            // Retrieve and cache Google Account ID during authentication
+            try {
+              await fetchAndCacheUserInfo(client);
+            } catch (error) {
+              console.warn(
+                'Failed to retrieve Google Account ID during authentication:',
+                getErrorMessage(error),
+              );
+              // Don't fail the auth flow if Google Account ID retrieval fails
+            }
+
+            res.writeHead(HTTP_REDIRECT, { Location: SIGN_IN_SUCCESS_URL });
+            res.end();
+            resolve();
+          } catch (error) {
+            res.writeHead(HTTP_REDIRECT, { Location: SIGN_IN_FAILURE_URL });
+            res.end();
+            reject(
+              new FatalAuthenticationError(
+                `Failed to exchange authorization code for tokens: ${getErrorMessage(error)}`,
+              ),
+            );
+          }
+        } else {
+          reject(
+            new FatalAuthenticationError(
+              'No authorization code received from Google OAuth. Please try authenticating again.',
+            ),
+          );
+        }
+      } catch (e) {
+        // Provide more specific error message for unexpected errors during OAuth flow
+        if (e instanceof FatalAuthenticationError) {
+          reject(e);
+        } else {
+          reject(
+            new FatalAuthenticationError(
+              `Unexpected error during OAuth authentication: ${getErrorMessage(e)}`,
+            ),
+          );
+        }
+      } finally {
+        server.close();
+      }
+    });
+
+    server.listen(port, host, () => {
+      // Server started successfully
+    });
+
+    server.on('error', (err) => {
+      reject(
+        new FatalAuthenticationError(
+          `OAuth callback server error: ${getErrorMessage(err)}`,
+        ),
+      );
+    });
+  });
+
+  return {
+    authUrl,
+    loginCompletePromise,
+  };
+}
+
+export function getAvailablePort(): Promise<number> {
+  return new Promise((resolve, reject) => {
+    let port = 0;
+    try {
+      const portStr = process.env['OAUTH_CALLBACK_PORT'];
+      if (portStr) {
+        port = parseInt(portStr, 10);
+        if (isNaN(port) || port <= 0 || port > 65535) {
+          return reject(
+            new Error(`Invalid value for OAUTH_CALLBACK_PORT: "${portStr}"`),
+          );
+        }
+        return resolve(port);
+      }
+      const server = net.createServer();
+      server.listen(0, () => {
+        const address = server.address()! as net.AddressInfo;
+        port = address.port;
+      });
+      server.on('listening', () => {
+        server.close();
+        server.unref();
+      });
+      server.on('error', (e) => reject(e));
+      server.on('close', () => resolve(port));
+    } catch (e) {
+      reject(e);
+    }
+  });
+}
+
+async function loadCachedCredentials(client: OAuth2Client): Promise<boolean> {
+  const useEncryptedStorage = getUseEncryptedStorageFlag();
+  if (useEncryptedStorage) {
+    const credentials = await OAuthCredentialStorage.loadCredentials();
+    if (credentials) {
+      client.setCredentials(credentials);
+      return true;
+    }
+    return false;
+  }
+
+  const pathsToTry = [
+    Storage.getOAuthCredsPath(),
+    process.env['GOOGLE_APPLICATION_CREDENTIALS'],
+  ].filter((p): p is string => !!p);
+
+  for (const keyFile of pathsToTry) {
+    try {
+      const creds = await fs.readFile(keyFile, 'utf-8');
+      client.setCredentials(JSON.parse(creds));
+
+      // This will verify locally that the credentials look good.
+      const { token } = await client.getAccessToken();
+      if (!token) {
+        continue;
+      }
+
+      // This will check with the server to see if it hasn't been revoked.
+      await client.getTokenInfo(token);
+
+      return true;
+    } catch (error) {
+      // Log specific error for debugging, but continue trying other paths
+      console.debug(
+        `Failed to load credentials from ${keyFile}:`,
+        getErrorMessage(error),
+      );
+    }
+  }
+
+  return false;
+}
+
+async function cacheCredentials(credentials: Credentials) {
+  const filePath = Storage.getOAuthCredsPath();
+  await fs.mkdir(path.dirname(filePath), { recursive: true });
+
+  const credString = JSON.stringify(credentials, null, 2);
+  await fs.writeFile(filePath, credString, { mode: 0o600 });
+  try {
+    await fs.chmod(filePath, 0o600);
+  } catch {
+    /* empty */
+  }
+}
+
+export function clearOauthClientCache() {
+  oauthClientPromises.clear();
+}
+
+export async function clearCachedCredentialFile() {
+  try {
+    const useEncryptedStorage = getUseEncryptedStorageFlag();
+    if (useEncryptedStorage) {
+      await OAuthCredentialStorage.clearCredentials();
+    } else {
+      await fs.rm(Storage.getOAuthCredsPath(), { force: true });
+    }
+    // Clear the Google Account ID cache when credentials are cleared
+    await userAccountManager.clearCachedGoogleAccount();
+    // Clear the in-memory OAuth client cache to force re-authentication
+    clearOauthClientCache();
+
+    /**
+     * Also clear Qwen SharedTokenManager cache and credentials file to prevent stale credentials
+     * when switching between auth types
+     * TODO: We do not depend on code_assist, we'll have to build an independent auth-cleaning procedure.
+     */
+    try {
+      const { SharedTokenManager } = await import(
+        '../qwen/sharedTokenManager.js'
+      );
+      const { clearQwenCredentials } = await import('../qwen/qwenOAuth2.js');
+
+      const sharedManager = SharedTokenManager.getInstance();
+      sharedManager.clearCache();
+
+      await clearQwenCredentials();
+    } catch (qwenError) {
+      console.debug('Could not clear Qwen credentials:', qwenError);
+    }
+  } catch (e) {
+    console.error('Failed to clear cached credentials:', e);
+  }
+}
+
+async function fetchAndCacheUserInfo(client: OAuth2Client): Promise<void> {
+  try {
+    const { token } = await client.getAccessToken();
+    if (!token) {
+      return;
+    }
+
+    const response = await fetch(
+      'https://www.googleapis.com/oauth2/v2/userinfo',
+      {
+        headers: {
+          Authorization: `Bearer ${token}`,
+        },
+      },
+    );
+
+    if (!response.ok) {
+      console.error(
+        'Failed to fetch user info:',
+        response.status,
+        response.statusText,
+      );
+      return;
+    }
+
+    const userInfo = await response.json();
+    await userAccountManager.cacheGoogleAccount(userInfo.email);
+  } catch (error) {
+    console.error('Error retrieving user info:', error);
+  }
+}
+
+// Helper to ensure test isolation
+export function resetOauthClientForTesting() {
+  oauthClientPromises.clear();
+}

packages/core/src/code_assist/server.test.ts

@@ -0,0 +1,255 @@
+/**
+ * @license
+ * Copyright 2025 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import { beforeEach, describe, it, expect, vi } from 'vitest';
+import { CodeAssistServer } from './server.js';
+import { OAuth2Client } from 'google-auth-library';
+import { UserTierId } from './types.js';
+
+vi.mock('google-auth-library');
+
+describe('CodeAssistServer', () => {
+  beforeEach(() => {
+    vi.resetAllMocks();
+  });
+
+  it('should be able to be constructed', () => {
+    const auth = new OAuth2Client();
+    const server = new CodeAssistServer(
+      auth,
+      'test-project',
+      {},
+      'test-session',
+      UserTierId.FREE,
+    );
+    expect(server).toBeInstanceOf(CodeAssistServer);
+  });
+
+  it('should call the generateContent endpoint', async () => {
+    const client = new OAuth2Client();
+    const server = new CodeAssistServer(
+      client,
+      'test-project',
+      {},
+      'test-session',
+      UserTierId.FREE,
+    );
+    const mockResponse = {
+      response: {
+        candidates: [
+          {
+            index: 0,
+            content: {
+              role: 'model',
+              parts: [{ text: 'response' }],
+            },
+            finishReason: 'STOP',
+            safetyRatings: [],
+          },
+        ],
+      },
+    };
+    vi.spyOn(server, 'requestPost').mockResolvedValue(mockResponse);
+
+    const response = await server.generateContent(
+      {
+        model: 'test-model',
+        contents: [{ role: 'user', parts: [{ text: 'request' }] }],
+      },
+      'user-prompt-id',
+    );
+
+    expect(server.requestPost).toHaveBeenCalledWith(
+      'generateContent',
+      expect.any(Object),
+      undefined,
+    );
+    expect(response.candidates?.[0]?.content?.parts?.[0]?.text).toBe(
+      'response',
+    );
+  });
+
+  it('should call the generateContentStream endpoint', async () => {
+    const client = new OAuth2Client();
+    const server = new CodeAssistServer(
+      client,
+      'test-project',
+      {},
+      'test-session',
+      UserTierId.FREE,
+    );
+    const mockResponse = (async function* () {
+      yield {
+        response: {
+          candidates: [
+            {
+              index: 0,
+              content: {
+                role: 'model',
+                parts: [{ text: 'response' }],
+              },
+              finishReason: 'STOP',
+              safetyRatings: [],
+            },
+          ],
+        },
+      };
+    })();
+    vi.spyOn(server, 'requestStreamingPost').mockResolvedValue(mockResponse);
+
+    const stream = await server.generateContentStream(
+      {
+        model: 'test-model',
+        contents: [{ role: 'user', parts: [{ text: 'request' }] }],
+      },
+      'user-prompt-id',
+    );
+
+    for await (const res of stream) {
+      expect(server.requestStreamingPost).toHaveBeenCalledWith(
+        'streamGenerateContent',
+        expect.any(Object),
+        undefined,
+      );
+      expect(res.candidates?.[0]?.content?.parts?.[0]?.text).toBe('response');
+    }
+  });
+
+  it('should call the onboardUser endpoint', async () => {
+    const client = new OAuth2Client();
+    const server = new CodeAssistServer(
+      client,
+      'test-project',
+      {},
+      'test-session',
+      UserTierId.FREE,
+    );
+    const mockResponse = {
+      name: 'operations/123',
+      done: true,
+    };
+    vi.spyOn(server, 'requestPost').mockResolvedValue(mockResponse);
+
+    const response = await server.onboardUser({
+      tierId: 'test-tier',
+      cloudaicompanionProject: 'test-project',
+      metadata: {},
+    });
+
+    expect(server.requestPost).toHaveBeenCalledWith(
+      'onboardUser',
+      expect.any(Object),
+    );
+    expect(response.name).toBe('operations/123');
+  });
+
+  it('should call the loadCodeAssist endpoint', async () => {
+    const client = new OAuth2Client();
+    const server = new CodeAssistServer(
+      client,
+      'test-project',
+      {},
+      'test-session',
+      UserTierId.FREE,
+    );
+    const mockResponse = {
+      currentTier: {
+        id: UserTierId.FREE,
+        name: 'Free',
+        description: 'free tier',
+      },
+      allowedTiers: [],
+      ineligibleTiers: [],
+      cloudaicompanionProject: 'projects/test',
+    };
+    vi.spyOn(server, 'requestPost').mockResolvedValue(mockResponse);
+
+    const response = await server.loadCodeAssist({
+      metadata: {},
+    });
+
+    expect(server.requestPost).toHaveBeenCalledWith(
+      'loadCodeAssist',
+      expect.any(Object),
+    );
+    expect(response).toEqual(mockResponse);
+  });
+
+  it('should return 0 for countTokens', async () => {
+    const client = new OAuth2Client();
+    const server = new CodeAssistServer(
+      client,
+      'test-project',
+      {},
+      'test-session',
+      UserTierId.FREE,
+    );
+    const mockResponse = {
+      totalTokens: 100,
+    };
+    vi.spyOn(server, 'requestPost').mockResolvedValue(mockResponse);
+
+    const response = await server.countTokens({
+      model: 'test-model',
+      contents: [{ role: 'user', parts: [{ text: 'request' }] }],
+    });
+    expect(response.totalTokens).toBe(100);
+  });
+
+  it('should throw an error for embedContent', async () => {
+    const client = new OAuth2Client();
+    const server = new CodeAssistServer(
+      client,
+      'test-project',
+      {},
+      'test-session',
+      UserTierId.FREE,
+    );
+    await expect(
+      server.embedContent({
+        model: 'test-model',
+        contents: [{ role: 'user', parts: [{ text: 'request' }] }],
+      }),
+    ).rejects.toThrow();
+  });
+
+  it('should handle VPC-SC errors when calling loadCodeAssist', async () => {
+    const client = new OAuth2Client();
+    const server = new CodeAssistServer(
+      client,
+      'test-project',
+      {},
+      'test-session',
+      UserTierId.FREE,
+    );
+    const mockVpcScError = {
+      response: {
+        data: {
+          error: {
+            details: [
+              {
+                reason: 'SECURITY_POLICY_VIOLATED',
+              },
+            ],
+          },
+        },
+      },
+    };
+    vi.spyOn(server, 'requestPost').mockRejectedValue(mockVpcScError);
+
+    const response = await server.loadCodeAssist({
+      metadata: {},
+    });
+
+    expect(server.requestPost).toHaveBeenCalledWith(
+      'loadCodeAssist',
+      expect.any(Object),
+    );
+    expect(response).toEqual({
+      currentTier: { id: UserTierId.STANDARD },
+    });
+  });
+});

packages/core/src/code_assist/server.ts

@@ -0,0 +1,253 @@
+/**
+ * @license
+ * Copyright 2025 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import type { OAuth2Client } from 'google-auth-library';
+import type {
+  CodeAssistGlobalUserSettingResponse,
+  GoogleRpcResponse,
+  LoadCodeAssistRequest,
+  LoadCodeAssistResponse,
+  LongRunningOperationResponse,
+  OnboardUserRequest,
+  SetCodeAssistGlobalUserSettingRequest,
+} from './types.js';
+import type {
+  CountTokensParameters,
+  CountTokensResponse,
+  EmbedContentParameters,
+  EmbedContentResponse,
+  GenerateContentParameters,
+  GenerateContentResponse,
+} from '@google/genai';
+import * as readline from 'node:readline';
+import type { ContentGenerator } from '../core/contentGenerator.js';
+import { UserTierId } from './types.js';
+import type {
+  CaCountTokenResponse,
+  CaGenerateContentResponse,
+} from './converter.js';
+import {
+  fromCountTokenResponse,
+  fromGenerateContentResponse,
+  toCountTokenRequest,
+  toGenerateContentRequest,
+} from './converter.js';
+
+/** HTTP options to be used in each of the requests. */
+export interface HttpOptions {
+  /** Additional HTTP headers to be sent with the request. */
+  headers?: Record<string, string>;
+}
+
+export const CODE_ASSIST_ENDPOINT = 'https://localhost:0'; // Disable Google Code Assist API Request
+export const CODE_ASSIST_API_VERSION = 'v1internal';
+
+export class CodeAssistServer implements ContentGenerator {
+  constructor(
+    readonly client: OAuth2Client,
+    readonly projectId?: string,
+    readonly httpOptions: HttpOptions = {},
+    readonly sessionId?: string,
+    readonly userTier?: UserTierId,
+  ) {}
+
+  async generateContentStream(
+    req: GenerateContentParameters,
+    userPromptId: string,
+  ): Promise<AsyncGenerator<GenerateContentResponse>> {
+    const resps = await this.requestStreamingPost<CaGenerateContentResponse>(
+      'streamGenerateContent',
+      toGenerateContentRequest(
+        req,
+        userPromptId,
+        this.projectId,
+        this.sessionId,
+      ),
+      req.config?.abortSignal,
+    );
+    return (async function* (): AsyncGenerator<GenerateContentResponse> {
+      for await (const resp of resps) {
+        yield fromGenerateContentResponse(resp);
+      }
+    })();
+  }
+
+  async generateContent(
+    req: GenerateContentParameters,
+    userPromptId: string,
+  ): Promise<GenerateContentResponse> {
+    const resp = await this.requestPost<CaGenerateContentResponse>(
+      'generateContent',
+      toGenerateContentRequest(
+        req,
+        userPromptId,
+        this.projectId,
+        this.sessionId,
+      ),
+      req.config?.abortSignal,
+    );
+    return fromGenerateContentResponse(resp);
+  }
+
+  async onboardUser(
+    req: OnboardUserRequest,
+  ): Promise<LongRunningOperationResponse> {
+    return await this.requestPost<LongRunningOperationResponse>(
+      'onboardUser',
+      req,
+    );
+  }
+
+  async loadCodeAssist(
+    req: LoadCodeAssistRequest,
+  ): Promise<LoadCodeAssistResponse> {
+    try {
+      return await this.requestPost<LoadCodeAssistResponse>(
+        'loadCodeAssist',
+        req,
+      );
+    } catch (e) {
+      if (isVpcScAffectedUser(e)) {
+        return {
+          currentTier: { id: UserTierId.STANDARD },
+        };
+      } else {
+        throw e;
+      }
+    }
+  }
+
+  async getCodeAssistGlobalUserSetting(): Promise<CodeAssistGlobalUserSettingResponse> {
+    return await this.requestGet<CodeAssistGlobalUserSettingResponse>(
+      'getCodeAssistGlobalUserSetting',
+    );
+  }
+
+  async setCodeAssistGlobalUserSetting(
+    req: SetCodeAssistGlobalUserSettingRequest,
+  ): Promise<CodeAssistGlobalUserSettingResponse> {
+    return await this.requestPost<CodeAssistGlobalUserSettingResponse>(
+      'setCodeAssistGlobalUserSetting',
+      req,
+    );
+  }
+
+  async countTokens(req: CountTokensParameters): Promise<CountTokensResponse> {
+    const resp = await this.requestPost<CaCountTokenResponse>(
+      'countTokens',
+      toCountTokenRequest(req),
+    );
+    return fromCountTokenResponse(resp);
+  }
+
+  async embedContent(
+    _req: EmbedContentParameters,
+  ): Promise<EmbedContentResponse> {
+    throw Error();
+  }
+
+  async requestPost<T>(
+    method: string,
+    req: object,
+    signal?: AbortSignal,
+  ): Promise<T> {
+    const res = await this.client.request({
+      url: this.getMethodUrl(method),
+      method: 'POST',
+      headers: {
+        'Content-Type': 'application/json',
+        ...this.httpOptions.headers,
+      },
+      responseType: 'json',
+      body: JSON.stringify(req),
+      signal,
+    });
+    return res.data as T;
+  }
+
+  async requestGet<T>(method: string, signal?: AbortSignal): Promise<T> {
+    const res = await this.client.request({
+      url: this.getMethodUrl(method),
+      method: 'GET',
+      headers: {
+        'Content-Type': 'application/json',
+        ...this.httpOptions.headers,
+      },
+      responseType: 'json',
+      signal,
+    });
+    return res.data as T;
+  }
+
+  async requestStreamingPost<T>(
+    method: string,
+    req: object,
+    signal?: AbortSignal,
+  ): Promise<AsyncGenerator<T>> {
+    const res = await this.client.request({
+      url: this.getMethodUrl(method),
+      method: 'POST',
+      params: {
+        alt: 'sse',
+      },
+      headers: {
+        'Content-Type': 'application/json',
+        ...this.httpOptions.headers,
+      },
+      responseType: 'stream',
+      body: JSON.stringify(req),
+      signal,
+    });
+
+    return (async function* (): AsyncGenerator<T> {
+      const rl = readline.createInterface({
+        input: res.data as NodeJS.ReadableStream,
+        crlfDelay: Infinity, // Recognizes '\r\n' and '\n' as line breaks
+      });
+
+      let bufferedLines: string[] = [];
+      for await (const line of rl) {
+        // blank lines are used to separate JSON objects in the stream
+        if (line === '') {
+          if (bufferedLines.length === 0) {
+            continue; // no data to yield
+          }
+          yield JSON.parse(bufferedLines.join('\n')) as T;
+          bufferedLines = []; // Reset the buffer after yielding
+        } else if (line.startsWith('data: ')) {
+          bufferedLines.push(line.slice(6).trim());
+        } else {
+          throw new Error(`Unexpected line format in response: ${line}`);
+        }
+      }
+    })();
+  }
+
+  getMethodUrl(method: string): string {
+    const endpoint =
+      process.env['CODE_ASSIST_ENDPOINT'] ?? CODE_ASSIST_ENDPOINT;
+    return `${endpoint}/${CODE_ASSIST_API_VERSION}:${method}`;
+  }
+}
+
+function isVpcScAffectedUser(error: unknown): boolean {
+  if (error && typeof error === 'object' && 'response' in error) {
+    const gaxiosError = error as {
+      response?: {
+        data?: unknown;
+      };
+    };
+    const response = gaxiosError.response?.data as
+      | GoogleRpcResponse
+      | undefined;
+    if (Array.isArray(response?.error?.details)) {
+      return response.error.details.some(
+        (detail) => detail.reason === 'SECURITY_POLICY_VIOLATED',
+      );
+    }
+  }
+  return false;
+}

packages/core/src/code_assist/setup.test.ts

@@ -0,0 +1,224 @@
+/**
+ * @license
+ * Copyright 2025 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import { describe, it, expect, vi, beforeEach, afterEach } from 'vitest';
+import { setupUser, ProjectIdRequiredError } from './setup.js';
+import { CodeAssistServer } from '../code_assist/server.js';
+import type { OAuth2Client } from 'google-auth-library';
+import type { GeminiUserTier } from './types.js';
+import { UserTierId } from './types.js';
+
+vi.mock('../code_assist/server.js');
+
+const mockPaidTier: GeminiUserTier = {
+  id: UserTierId.STANDARD,
+  name: 'paid',
+  description: 'Paid tier',
+  isDefault: true,
+};
+
+const mockFreeTier: GeminiUserTier = {
+  id: UserTierId.FREE,
+  name: 'free',
+  description: 'Free tier',
+  isDefault: true,
+};
+
+describe('setupUser for existing user', () => {
+  let mockLoad: ReturnType<typeof vi.fn>;
+  let mockOnboardUser: ReturnType<typeof vi.fn>;
+
+  beforeEach(() => {
+    vi.resetAllMocks();
+    mockLoad = vi.fn();
+    mockOnboardUser = vi.fn().mockResolvedValue({
+      done: true,
+      response: {
+        cloudaicompanionProject: {
+          id: 'server-project',
+        },
+      },
+    });
+    vi.mocked(CodeAssistServer).mockImplementation(
+      () =>
+        ({
+          loadCodeAssist: mockLoad,
+          onboardUser: mockOnboardUser,
+        }) as unknown as CodeAssistServer,
+    );
+  });
+
+  afterEach(() => {
+    vi.unstubAllEnvs();
+  });
+
+  it('should use GOOGLE_CLOUD_PROJECT when set and project from server is undefined', async () => {
+    vi.stubEnv('GOOGLE_CLOUD_PROJECT', 'test-project');
+    mockLoad.mockResolvedValue({
+      currentTier: mockPaidTier,
+    });
+    await setupUser({} as OAuth2Client);
+    expect(CodeAssistServer).toHaveBeenCalledWith(
+      {},
+      'test-project',
+      {},
+      '',
+      undefined,
+    );
+  });
+
+  it('should ignore GOOGLE_CLOUD_PROJECT when project from server is set', async () => {
+    vi.stubEnv('GOOGLE_CLOUD_PROJECT', 'test-project');
+    mockLoad.mockResolvedValue({
+      cloudaicompanionProject: 'server-project',
+      currentTier: mockPaidTier,
+    });
+    const projectId = await setupUser({} as OAuth2Client);
+    expect(CodeAssistServer).toHaveBeenCalledWith(
+      {},
+      'test-project',
+      {},
+      '',
+      undefined,
+    );
+    expect(projectId).toEqual({
+      projectId: 'server-project',
+      userTier: 'standard-tier',
+    });
+  });
+
+  it('should throw ProjectIdRequiredError when no project ID is available', async () => {
+    vi.stubEnv('GOOGLE_CLOUD_PROJECT', '');
+    // And the server itself requires a project ID internally
+    vi.mocked(CodeAssistServer).mockImplementation(() => {
+      throw new ProjectIdRequiredError();
+    });
+
+    await expect(setupUser({} as OAuth2Client)).rejects.toThrow(
+      ProjectIdRequiredError,
+    );
+  });
+});
+
+describe('setupUser for new user', () => {
+  let mockLoad: ReturnType<typeof vi.fn>;
+  let mockOnboardUser: ReturnType<typeof vi.fn>;
+
+  beforeEach(() => {
+    vi.resetAllMocks();
+    mockLoad = vi.fn();
+    mockOnboardUser = vi.fn().mockResolvedValue({
+      done: true,
+      response: {
+        cloudaicompanionProject: {
+          id: 'server-project',
+        },
+      },
+    });
+    vi.mocked(CodeAssistServer).mockImplementation(
+      () =>
+        ({
+          loadCodeAssist: mockLoad,
+          onboardUser: mockOnboardUser,
+        }) as unknown as CodeAssistServer,
+    );
+  });
+
+  afterEach(() => {
+    vi.unstubAllEnvs();
+  });
+
+  it('should use GOOGLE_CLOUD_PROJECT when set and onboard a new paid user', async () => {
+    vi.stubEnv('GOOGLE_CLOUD_PROJECT', 'test-project');
+    mockLoad.mockResolvedValue({
+      allowedTiers: [mockPaidTier],
+    });
+    const userData = await setupUser({} as OAuth2Client);
+    expect(CodeAssistServer).toHaveBeenCalledWith(
+      {},
+      'test-project',
+      {},
+      '',
+      undefined,
+    );
+    expect(mockLoad).toHaveBeenCalled();
+    expect(mockOnboardUser).toHaveBeenCalledWith({
+      tierId: 'standard-tier',
+      cloudaicompanionProject: 'test-project',
+      metadata: {
+        ideType: 'IDE_UNSPECIFIED',
+        platform: 'PLATFORM_UNSPECIFIED',
+        pluginType: 'GEMINI',
+        duetProject: 'test-project',
+      },
+    });
+    expect(userData).toEqual({
+      projectId: 'server-project',
+      userTier: 'standard-tier',
+    });
+  });
+
+  it('should onboard a new free user when GOOGLE_CLOUD_PROJECT is not set', async () => {
+    vi.stubEnv('GOOGLE_CLOUD_PROJECT', '');
+    mockLoad.mockResolvedValue({
+      allowedTiers: [mockFreeTier],
+    });
+    const userData = await setupUser({} as OAuth2Client);
+    expect(CodeAssistServer).toHaveBeenCalledWith(
+      {},
+      undefined,
+      {},
+      '',
+      undefined,
+    );
+    expect(mockLoad).toHaveBeenCalled();
+    expect(mockOnboardUser).toHaveBeenCalledWith({
+      tierId: 'free-tier',
+      cloudaicompanionProject: undefined,
+      metadata: {
+        ideType: 'IDE_UNSPECIFIED',
+        platform: 'PLATFORM_UNSPECIFIED',
+        pluginType: 'GEMINI',
+      },
+    });
+    expect(userData).toEqual({
+      projectId: 'server-project',
+      userTier: 'free-tier',
+    });
+  });
+
+  it('should use GOOGLE_CLOUD_PROJECT when onboard response has no project ID', async () => {
+    vi.stubEnv('GOOGLE_CLOUD_PROJECT', 'test-project');
+    mockLoad.mockResolvedValue({
+      allowedTiers: [mockPaidTier],
+    });
+    mockOnboardUser.mockResolvedValue({
+      done: true,
+      response: {
+        cloudaicompanionProject: undefined,
+      },
+    });
+    const userData = await setupUser({} as OAuth2Client);
+    expect(userData).toEqual({
+      projectId: 'test-project',
+      userTier: 'standard-tier',
+    });
+  });
+
+  it('should throw ProjectIdRequiredError when no project ID is available', async () => {
+    vi.stubEnv('GOOGLE_CLOUD_PROJECT', '');
+    mockLoad.mockResolvedValue({
+      allowedTiers: [mockPaidTier],
+    });
+    mockOnboardUser.mockResolvedValue({
+      done: true,
+      response: {},
+    });
+    await expect(setupUser({} as OAuth2Client)).rejects.toThrow(
+      ProjectIdRequiredError,
+    );
+  });
+});

packages/core/src/code_assist/setup.ts

@@ -0,0 +1,124 @@
+/**
+ * @license
+ * Copyright 2025 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import type {
+  ClientMetadata,
+  GeminiUserTier,
+  LoadCodeAssistResponse,
+  OnboardUserRequest,
+} from './types.js';
+import { UserTierId } from './types.js';
+import { CodeAssistServer } from './server.js';
+import type { OAuth2Client } from 'google-auth-library';
+
+export class ProjectIdRequiredError extends Error {
+  constructor() {
+    super(
+      'This account requires setting the GOOGLE_CLOUD_PROJECT env var. See https://goo.gle/gemini-cli-auth-docs#workspace-gca',
+    );
+  }
+}
+
+export interface UserData {
+  projectId: string;
+  userTier: UserTierId;
+}
+
+/**
+ *
+ * @param projectId the user's project id, if any
+ * @returns the user's actual project id
+ */
+export async function setupUser(client: OAuth2Client): Promise<UserData> {
+  const projectId = process.env['GOOGLE_CLOUD_PROJECT'] || undefined;
+  const caServer = new CodeAssistServer(client, projectId, {}, '', undefined);
+  const coreClientMetadata: ClientMetadata = {
+    ideType: 'IDE_UNSPECIFIED',
+    platform: 'PLATFORM_UNSPECIFIED',
+    pluginType: 'GEMINI',
+  };
+
+  const loadRes = await caServer.loadCodeAssist({
+    cloudaicompanionProject: projectId,
+    metadata: {
+      ...coreClientMetadata,
+      duetProject: projectId,
+    },
+  });
+
+  if (loadRes.currentTier) {
+    if (!loadRes.cloudaicompanionProject) {
+      if (projectId) {
+        return {
+          projectId,
+          userTier: loadRes.currentTier.id,
+        };
+      }
+      throw new ProjectIdRequiredError();
+    }
+    return {
+      projectId: loadRes.cloudaicompanionProject,
+      userTier: loadRes.currentTier.id,
+    };
+  }
+
+  const tier = getOnboardTier(loadRes);
+
+  let onboardReq: OnboardUserRequest;
+  if (tier.id === UserTierId.FREE) {
+    // The free tier uses a managed google cloud project. Setting a project in the `onboardUser` request causes a `Precondition Failed` error.
+    onboardReq = {
+      tierId: tier.id,
+      cloudaicompanionProject: undefined,
+      metadata: coreClientMetadata,
+    };
+  } else {
+    onboardReq = {
+      tierId: tier.id,
+      cloudaicompanionProject: projectId,
+      metadata: {
+        ...coreClientMetadata,
+        duetProject: projectId,
+      },
+    };
+  }
+
+  // Poll onboardUser until long running operation is complete.
+  let lroRes = await caServer.onboardUser(onboardReq);
+  while (!lroRes.done) {
+    await new Promise((f) => setTimeout(f, 5000));
+    lroRes = await caServer.onboardUser(onboardReq);
+  }
+
+  if (!lroRes.response?.cloudaicompanionProject?.id) {
+    if (projectId) {
+      return {
+        projectId,
+        userTier: tier.id,
+      };
+    }
+    throw new ProjectIdRequiredError();
+  }
+
+  return {
+    projectId: lroRes.response.cloudaicompanionProject.id,
+    userTier: tier.id,
+  };
+}
+
+function getOnboardTier(res: LoadCodeAssistResponse): GeminiUserTier {
+  for (const tier of res.allowedTiers || []) {
+    if (tier.isDefault) {
+      return tier;
+    }
+  }
+  return {
+    name: '',
+    description: '',
+    id: UserTierId.LEGACY,
+    userDefinedCloudaicompanionProject: true,
+  };
+}

packages/core/src/code_assist/types.ts

@@ -0,0 +1,201 @@
+/**
+ * @license
+ * Copyright 2025 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+export interface ClientMetadata {
+  ideType?: ClientMetadataIdeType;
+  ideVersion?: string;
+  pluginVersion?: string;
+  platform?: ClientMetadataPlatform;
+  updateChannel?: string;
+  duetProject?: string;
+  pluginType?: ClientMetadataPluginType;
+  ideName?: string;
+}
+
+export type ClientMetadataIdeType =
+  | 'IDE_UNSPECIFIED'
+  | 'VSCODE'
+  | 'INTELLIJ'
+  | 'VSCODE_CLOUD_WORKSTATION'
+  | 'INTELLIJ_CLOUD_WORKSTATION'
+  | 'CLOUD_SHELL';
+export type ClientMetadataPlatform =
+  | 'PLATFORM_UNSPECIFIED'
+  | 'DARWIN_AMD64'
+  | 'DARWIN_ARM64'
+  | 'LINUX_AMD64'
+  | 'LINUX_ARM64'
+  | 'WINDOWS_AMD64';
+export type ClientMetadataPluginType =
+  | 'PLUGIN_UNSPECIFIED'
+  | 'CLOUD_CODE'
+  | 'GEMINI'
+  | 'AIPLUGIN_INTELLIJ'
+  | 'AIPLUGIN_STUDIO';
+
+export interface LoadCodeAssistRequest {
+  cloudaicompanionProject?: string;
+  metadata: ClientMetadata;
+}
+
+/**
+ * Represents LoadCodeAssistResponse proto json field
+ * http://google3/google/internal/cloud/code/v1internal/cloudcode.proto;l=224
+ */
+export interface LoadCodeAssistResponse {
+  currentTier?: GeminiUserTier | null;
+  allowedTiers?: GeminiUserTier[] | null;
+  ineligibleTiers?: IneligibleTier[] | null;
+  cloudaicompanionProject?: string | null;
+}
+
+/**
+ * GeminiUserTier reflects the structure received from the CodeAssist when calling LoadCodeAssist.
+ */
+export interface GeminiUserTier {
+  id: UserTierId;
+  name?: string;
+  description?: string;
+  // This value is used to declare whether a given tier requires the user to configure the project setting on the IDE settings or not.
+  userDefinedCloudaicompanionProject?: boolean | null;
+  isDefault?: boolean;
+  privacyNotice?: PrivacyNotice;
+  hasAcceptedTos?: boolean;
+  hasOnboardedPreviously?: boolean;
+}
+
+/**
+ * Includes information specifying the reasons for a user's ineligibility for a specific tier.
+ * @param reasonCode mnemonic code representing the reason for in-eligibility.
+ * @param reasonMessage message to display to the user.
+ * @param tierId id of the tier.
+ * @param tierName name of the tier.
+ */
+export interface IneligibleTier {
+  reasonCode: IneligibleTierReasonCode;
+  reasonMessage: string;
+  tierId: UserTierId;
+  tierName: string;
+}
+
+/**
+ * List of predefined reason codes when a tier is blocked from a specific tier.
+ * https://source.corp.google.com/piper///depot/google3/google/internal/cloud/code/v1internal/cloudcode.proto;l=378
+ */
+export enum IneligibleTierReasonCode {
+  // go/keep-sorted start
+  DASHER_USER = 'DASHER_USER',
+  INELIGIBLE_ACCOUNT = 'INELIGIBLE_ACCOUNT',
+  NON_USER_ACCOUNT = 'NON_USER_ACCOUNT',
+  RESTRICTED_AGE = 'RESTRICTED_AGE',
+  RESTRICTED_NETWORK = 'RESTRICTED_NETWORK',
+  UNKNOWN = 'UNKNOWN',
+  UNKNOWN_LOCATION = 'UNKNOWN_LOCATION',
+  UNSUPPORTED_LOCATION = 'UNSUPPORTED_LOCATION',
+  // go/keep-sorted end
+}
+/**
+ * UserTierId represents IDs returned from the Cloud Code Private API representing a user's tier
+ *
+ * //depot/google3/cloud/developer_experience/cloudcode/pa/service/usertier.go;l=16
+ */
+export enum UserTierId {
+  FREE = 'free-tier',
+  LEGACY = 'legacy-tier',
+  STANDARD = 'standard-tier',
+}
+
+/**
+ * PrivacyNotice reflects the structure received from the CodeAssist in regards to a tier
+ * privacy notice.
+ */
+export interface PrivacyNotice {
+  showNotice: boolean;
+  noticeText?: string;
+}
+
+/**
+ * Proto signature of OnboardUserRequest as payload to OnboardUser call
+ */
+export interface OnboardUserRequest {
+  tierId: string | undefined;
+  cloudaicompanionProject: string | undefined;
+  metadata: ClientMetadata | undefined;
+}
+
+/**
+ * Represents LongRunningOperation proto
+ * http://google3/google/longrunning/operations.proto;rcl=698857719;l=107
+ */
+export interface LongRunningOperationResponse {
+  name: string;
+  done?: boolean;
+  response?: OnboardUserResponse;
+}
+
+/**
+ * Represents OnboardUserResponse proto
+ * http://google3/google/internal/cloud/code/v1internal/cloudcode.proto;l=215
+ */
+export interface OnboardUserResponse {
+  // tslint:disable-next-line:enforce-name-casing This is the name of the field in the proto.
+  cloudaicompanionProject?: {
+    id: string;
+    name: string;
+  };
+}
+
+/**
+ * Status code of user license status
+ * it does not strictly correspond to the proto
+ * Error value is an additional value assigned to error responses from OnboardUser
+ */
+export enum OnboardUserStatusCode {
+  Default = 'DEFAULT',
+  Notice = 'NOTICE',
+  Warning = 'WARNING',
+  Error = 'ERROR',
+}
+
+/**
+ * Status of user onboarded to gemini
+ */
+export interface OnboardUserStatus {
+  statusCode: OnboardUserStatusCode;
+  displayMessage: string;
+  helpLink: HelpLinkUrl | undefined;
+}
+
+export interface HelpLinkUrl {
+  description: string;
+  url: string;
+}
+
+export interface SetCodeAssistGlobalUserSettingRequest {
+  cloudaicompanionProject?: string;
+  freeTierDataCollectionOptin: boolean;
+}
+
+export interface CodeAssistGlobalUserSettingResponse {
+  cloudaicompanionProject?: string;
+  freeTierDataCollectionOptin: boolean;
+}
+
+/**
+ * Relevant fields that can be returned from a Google RPC response
+ */
+export interface GoogleRpcResponse {
+  error?: {
+    details?: GoogleRpcErrorInfo[];
+  };
+}
+
+/**
+ * Relevant fields that can be returned in the details of an error returned from GoogleRPCs
+ */
+interface GoogleRpcErrorInfo {
+  reason?: string;
+}

packages/core/src/config/config.test.ts

@@ -283,6 +283,23 @@ describe('Server Config (config.ts)', () => {
       expect(config.isInFallbackMode()).toBe(false);
     });
 
+    it('should strip thoughts when switching from GenAI to Vertex', async () => {
+      const config = new Config(baseParams);
+
+      vi.mocked(createContentGeneratorConfig).mockImplementation(
+        (_: Config, authType: AuthType | undefined) =>
+          ({ authType }) as unknown as ContentGeneratorConfig,
+      );
+
+      await config.refreshAuth(AuthType.USE_GEMINI);
+
+      await config.refreshAuth(AuthType.LOGIN_WITH_GOOGLE);
+
+      expect(
+        config.getGeminiClient().stripThoughtsFromHistory,
+      ).toHaveBeenCalledWith();
+    });
+
     it('should not strip thoughts when switching from Vertex to GenAI', async () => {
       const config = new Config(baseParams);
 

packages/core/src/config/config.ts

@@ -16,7 +16,6 @@ import { ProxyAgent, setGlobalDispatcher } from 'undici';
 import type {
   ContentGenerator,
   ContentGeneratorConfig,
-  AuthType,
 } from '../core/contentGenerator.js';
 import type { FallbackModelHandler } from '../fallback/types.js';
 import type { MCPOAuthConfig } from '../mcp/oauth-provider.js';
@@ -27,6 +26,7 @@ import type { AnyToolInvocation } from '../tools/tools.js';
 import { BaseLlmClient } from '../core/baseLlmClient.js';
 import { GeminiClient } from '../core/client.js';
 import {
+  AuthType,
   createContentGenerator,
   createContentGeneratorConfig,
 } from '../core/contentGenerator.js';
@@ -684,6 +684,16 @@ export class Config {
   }
 
   async refreshAuth(authMethod: AuthType, isInitialAuth?: boolean) {
+    // Vertex and Genai have incompatible encryption and sending history with
+    // throughtSignature from Genai to Vertex will fail, we need to strip them
+    if (
+      this.contentGeneratorConfig?.authType === AuthType.USE_GEMINI &&
+      authMethod === AuthType.LOGIN_WITH_GOOGLE
+    ) {
+      // Restore the conversation history to the new client
+      this.geminiClient.stripThoughtsFromHistory();
+    }
+
     const newContentGeneratorConfig = createContentGeneratorConfig(
       this,
       authMethod,

packages/core/src/config/flashFallback.test.ts

@@ -31,7 +31,7 @@ describe('Flash Model Fallback Configuration', () => {
       config as unknown as { contentGeneratorConfig: unknown }
     ).contentGeneratorConfig = {
       model: DEFAULT_GEMINI_MODEL,
-      authType: 'gemini-api-key',
+      authType: 'oauth-personal',
     };
   });
 

packages/core/src/core/__tests__/openaiTimeoutHandling.test.ts

@@ -73,7 +73,6 @@ describe('OpenAIContentGenerator Timeout Handling', () => {
       }),
       buildClient: vi.fn().mockReturnValue(mockOpenAIClient),
       buildRequest: vi.fn().mockImplementation((req) => req),
-      getDefaultGenerationConfig: vi.fn().mockReturnValue({}),
     };
 
     // Create generator instance
@@ -300,7 +299,6 @@ describe('OpenAIContentGenerator Timeout Handling', () => {
         }),
         buildClient: vi.fn().mockReturnValue(mockOpenAIClient),
         buildRequest: vi.fn().mockImplementation((req) => req),
-        getDefaultGenerationConfig: vi.fn().mockReturnValue({}),
       };
 
       new OpenAIContentGenerator(
@@ -335,7 +333,6 @@ describe('OpenAIContentGenerator Timeout Handling', () => {
         }),
         buildClient: vi.fn().mockReturnValue(mockOpenAIClient),
         buildRequest: vi.fn().mockImplementation((req) => req),
-        getDefaultGenerationConfig: vi.fn().mockReturnValue({}),
       };
 
       new OpenAIContentGenerator(

packages/core/src/core/baseLlmClient.test.ts

@@ -146,11 +146,12 @@ describe('BaseLlmClient', () => {
       // Validate the parameters passed to the underlying generator
       expect(mockGenerateContent).toHaveBeenCalledTimes(1);
       expect(mockGenerateContent).toHaveBeenCalledWith(
-        expect.objectContaining({
+        {
           model: 'test-model',
           contents: defaultOptions.contents,
-          config: expect.objectContaining({
+          config: {
             abortSignal: defaultOptions.abortSignal,
+            topP: 0.8,
             tools: [
               {
                 functionDeclarations: [
@@ -162,8 +163,9 @@ describe('BaseLlmClient', () => {
                 ],
               },
             ],
-          }),
-        }),
+            // Crucial: systemInstruction should NOT be in the config object if not provided
+          },
+        },
         'test-prompt-id',
       );
     });
@@ -186,6 +188,7 @@ describe('BaseLlmClient', () => {
         expect.objectContaining({
           config: expect.objectContaining({
             temperature: 0.8,
+            topP: 0.8, // Default should remain if not overridden
             topK: 10,
             tools: expect.any(Array),
           }),

packages/core/src/core/baseLlmClient.ts

@@ -64,6 +64,11 @@ export interface GenerateJsonOptions {
  * A client dedicated to stateless, utility-focused LLM calls.
  */
 export class BaseLlmClient {
+  // Default configuration for utility tasks
+  private readonly defaultUtilityConfig: GenerateContentConfig = {
+    topP: 0.8,
+  };
+
   constructor(
     private readonly contentGenerator: ContentGenerator,
     private readonly config: Config,
@@ -84,6 +89,7 @@ export class BaseLlmClient {
 
     const requestConfig: GenerateContentConfig = {
       abortSignal,
+      ...this.defaultUtilityConfig,
       ...options.config,
       ...(systemInstruction && { systemInstruction }),
     };

packages/core/src/core/client.test.ts

@@ -15,7 +15,11 @@ import {
 } from 'vitest';
 
 import type { Content, GenerateContentResponse, Part } from '@google/genai';
-import { GeminiClient } from './client.js';
+import {
+  isThinkingDefault,
+  isThinkingSupported,
+  GeminiClient,
+} from './client.js';
 import { findCompressSplitPoint } from '../services/chatCompressionService.js';
 import {
   AuthType,
@@ -243,6 +247,40 @@ describe('findCompressSplitPoint', () => {
   });
 });
 
+describe('isThinkingSupported', () => {
+  it('should return true for gemini-2.5', () => {
+    expect(isThinkingSupported('gemini-2.5')).toBe(true);
+  });
+
+  it('should return true for gemini-2.5-pro', () => {
+    expect(isThinkingSupported('gemini-2.5-pro')).toBe(true);
+  });
+
+  it('should return false for other models', () => {
+    expect(isThinkingSupported('gemini-1.5-flash')).toBe(false);
+    expect(isThinkingSupported('some-other-model')).toBe(false);
+  });
+});
+
+describe('isThinkingDefault', () => {
+  it('should return false for gemini-2.5-flash-lite', () => {
+    expect(isThinkingDefault('gemini-2.5-flash-lite')).toBe(false);
+  });
+
+  it('should return true for gemini-2.5', () => {
+    expect(isThinkingDefault('gemini-2.5')).toBe(true);
+  });
+
+  it('should return true for gemini-2.5-pro', () => {
+    expect(isThinkingDefault('gemini-2.5-pro')).toBe(true);
+  });
+
+  it('should return false for other models', () => {
+    expect(isThinkingDefault('gemini-1.5-flash')).toBe(false);
+    expect(isThinkingDefault('some-other-model')).toBe(false);
+  });
+});
+
 describe('Gemini Client (client.ts)', () => {
   let mockContentGenerator: ContentGenerator;
   let mockConfig: Config;
@@ -2266,15 +2304,16 @@ ${JSON.stringify(
       );
 
       expect(mockContentGenerator.generateContent).toHaveBeenCalledWith(
-        expect.objectContaining({
+        {
           model: DEFAULT_GEMINI_FLASH_MODEL,
-          config: expect.objectContaining({
+          config: {
             abortSignal,
             systemInstruction: getCoreSystemPrompt(''),
             temperature: 0.5,
-          }),
+            topP: 0.8,
+          },
           contents,
-        }),
+        },
         'test-session-id',
       );
     });

packages/core/src/core/client.ts

@@ -15,7 +15,11 @@ import type {
 
 // Config
 import { ApprovalMode, type Config } from '../config/config.js';
-import { DEFAULT_GEMINI_FLASH_MODEL } from '../config/models.js';
+import {
+  DEFAULT_GEMINI_FLASH_MODEL,
+  DEFAULT_GEMINI_MODEL_AUTO,
+  DEFAULT_THINKING_MODE,
+} from '../config/models.js';
 
 // Core modules
 import type { ContentGenerator } from './contentGenerator.js';
@@ -74,10 +78,24 @@ import { type File, type IdeContext } from '../ide/types.js';
 // Fallback handling
 import { handleFallback } from '../fallback/handler.js';
 
+export function isThinkingSupported(model: string) {
+  return model.startsWith('gemini-2.5') || model === DEFAULT_GEMINI_MODEL_AUTO;
+}
+
+export function isThinkingDefault(model: string) {
+  if (model.startsWith('gemini-2.5-flash-lite')) {
+    return false;
+  }
+  return model.startsWith('gemini-2.5') || model === DEFAULT_GEMINI_MODEL_AUTO;
+}
+
 const MAX_TURNS = 100;
 
 export class GeminiClient {
   private chat?: GeminiChat;
+  private readonly generateContentConfig: GenerateContentConfig = {
+    topP: 0.8,
+  };
   private sessionTurnCount = 0;
 
   private readonly loopDetector: LoopDetectionService;
@@ -189,10 +207,20 @@ export class GeminiClient {
       const model = this.config.getModel();
       const systemInstruction = getCoreSystemPrompt(userMemory, model);
 
+      const config: GenerateContentConfig = { ...this.generateContentConfig };
+
+      if (isThinkingSupported(model)) {
+        config.thinkingConfig = {
+          includeThoughts: true,
+          thinkingBudget: DEFAULT_THINKING_MODE,
+        };
+      }
+
       return new GeminiChat(
         this.config,
         {
           systemInstruction,
+          ...config,
           tools,
         },
         history,
@@ -589,6 +617,11 @@ export class GeminiClient {
   ): Promise<GenerateContentResponse> {
     let currentAttemptModel: string = model;
 
+    const configToUse: GenerateContentConfig = {
+      ...this.generateContentConfig,
+      ...generationConfig,
+    };
+
     try {
       const userMemory = this.config.getUserMemory();
       const finalSystemInstruction = generationConfig.systemInstruction
@@ -597,7 +630,7 @@ export class GeminiClient {
 
       const requestConfig: GenerateContentConfig = {
         abortSignal,
-        ...generationConfig,
+        ...configToUse,
         systemInstruction: finalSystemInstruction,
       };
 
@@ -638,7 +671,7 @@ export class GeminiClient {
         `Error generating content via API with model ${currentAttemptModel}.`,
         {
           requestContents: contents,
-          requestConfig: generationConfig,
+          requestConfig: configToUse,
         },
         'generateContent-api',
       );

packages/core/src/core/contentGenerator.test.ts

@@ -5,19 +5,42 @@
  */
 
 import { describe, it, expect, vi } from 'vitest';
+import type { ContentGenerator } from './contentGenerator.js';
 import { createContentGenerator, AuthType } from './contentGenerator.js';
+import { createCodeAssistContentGenerator } from '../code_assist/codeAssist.js';
 import { GoogleGenAI } from '@google/genai';
 import type { Config } from '../config/config.js';
-import { LoggingContentGenerator } from './geminiContentGenerator/loggingContentGenerator.js';
+import { LoggingContentGenerator } from './loggingContentGenerator.js';
 
+vi.mock('../code_assist/codeAssist.js');
 vi.mock('@google/genai');
 
+const mockConfig = {
+  getCliVersion: vi.fn().mockReturnValue('1.0.0'),
+} as unknown as Config;
+
 describe('createContentGenerator', () => {
-  it('should create a Gemini content generator', async () => {
+  it('should create a CodeAssistContentGenerator', async () => {
+    const mockGenerator = {} as unknown as ContentGenerator;
+    vi.mocked(createCodeAssistContentGenerator).mockResolvedValue(
+      mockGenerator as never,
+    );
+    const generator = await createContentGenerator(
+      {
+        model: 'test-model',
+        authType: AuthType.LOGIN_WITH_GOOGLE,
+      },
+      mockConfig,
+    );
+    expect(createCodeAssistContentGenerator).toHaveBeenCalled();
+    expect(generator).toEqual(
+      new LoggingContentGenerator(mockGenerator, mockConfig),
+    );
+  });
+
+  it('should create a GoogleGenAI content generator', async () => {
     const mockConfig = {
       getUsageStatisticsEnabled: () => true,
-      getContentGeneratorConfig: () => ({}),
-      getCliVersion: () => '1.0.0',
     } as unknown as Config;
 
     const mockGenerator = {
@@ -42,17 +65,17 @@ describe('createContentGenerator', () => {
         },
       },
     });
-    // We expect it to be a LoggingContentGenerator wrapping a GeminiContentGenerator
-    expect(generator).toBeInstanceOf(LoggingContentGenerator);
-    const wrapped = (generator as LoggingContentGenerator).getWrapped();
-    expect(wrapped).toBeDefined();
+    expect(generator).toEqual(
+      new LoggingContentGenerator(
+        (mockGenerator as GoogleGenAI).models,
+        mockConfig,
+      ),
+    );
   });
 
-  it('should create a Gemini content generator with client install id logging disabled', async () => {
+  it('should create a GoogleGenAI content generator with client install id logging disabled', async () => {
     const mockConfig = {
       getUsageStatisticsEnabled: () => false,
-      getContentGeneratorConfig: () => ({}),
-      getCliVersion: () => '1.0.0',
     } as unknown as Config;
     const mockGenerator = {
       models: {},
@@ -75,6 +98,11 @@ describe('createContentGenerator', () => {
         },
       },
     });
-    expect(generator).toBeInstanceOf(LoggingContentGenerator);
+    expect(generator).toEqual(
+      new LoggingContentGenerator(
+        (mockGenerator as GoogleGenAI).models,
+        mockConfig,
+      ),
+    );
   });
 });

packages/core/src/core/contentGenerator.ts

@@ -12,9 +12,15 @@ import type {
   GenerateContentParameters,
   GenerateContentResponse,
 } from '@google/genai';
+import { GoogleGenAI } from '@google/genai';
+import { createCodeAssistContentGenerator } from '../code_assist/codeAssist.js';
 import { DEFAULT_QWEN_MODEL } from '../config/models.js';
 import type { Config } from '../config/config.js';
 
+import type { UserTierId } from '../code_assist/types.js';
+import { InstallationManager } from '../utils/installationManager.js';
+import { LoggingContentGenerator } from './loggingContentGenerator.js';
+
 /**
  * Interface abstracting the core functionalities for generating content and counting tokens.
  */
@@ -33,12 +39,14 @@ export interface ContentGenerator {
 
   embedContent(request: EmbedContentParameters): Promise<EmbedContentResponse>;
 
-  useSummarizedThinking(): boolean;
+  userTier?: UserTierId;
 }
 
 export enum AuthType {
+  LOGIN_WITH_GOOGLE = 'oauth-personal',
   USE_GEMINI = 'gemini-api-key',
   USE_VERTEX_AI = 'vertex-ai',
+  CLOUD_SHELL = 'cloud-shell',
   USE_OPENAI = 'openai',
   QWEN_OAUTH = 'qwen-oauth',
 }
@@ -51,9 +59,12 @@ export type ContentGeneratorConfig = {
   authType?: AuthType | undefined;
   enableOpenAILogging?: boolean;
   openAILoggingDir?: string;
-  timeout?: number; // Timeout configuration in milliseconds
-  maxRetries?: number; // Maximum retries for failed requests
-  disableCacheControl?: boolean; // Disable cache control for DashScope providers
+  // Timeout configuration in milliseconds
+  timeout?: number;
+  // Maximum retries for failed requests
+  maxRetries?: number;
+  // Disable cache control for DashScope providers
+  disableCacheControl?: boolean;
   samplingParams?: {
     top_p?: number;
     top_k?: number;
@@ -63,9 +74,6 @@ export type ContentGeneratorConfig = {
     temperature?: number;
     max_tokens?: number;
   };
-  reasoning?: {
-    effort?: 'low' | 'medium' | 'high';
-  };
   proxy?: string | undefined;
   userAgent?: string;
   // Schema compliance mode for tool definitions
@@ -115,14 +123,48 @@ export async function createContentGenerator(
   gcConfig: Config,
   isInitialAuth?: boolean,
 ): Promise<ContentGenerator> {
+  const version = process.env['CLI_VERSION'] || process.version;
+  const userAgent = `QwenCode/${version} (${process.platform}; ${process.arch})`;
+  const baseHeaders: Record<string, string> = {
+    'User-Agent': userAgent,
+  };
+
+  if (
+    config.authType === AuthType.LOGIN_WITH_GOOGLE ||
+    config.authType === AuthType.CLOUD_SHELL
+  ) {
+    const httpOptions = { headers: baseHeaders };
+    return new LoggingContentGenerator(
+      await createCodeAssistContentGenerator(
+        httpOptions,
+        config.authType,
+        gcConfig,
+      ),
+      gcConfig,
+    );
+  }
+
   if (
     config.authType === AuthType.USE_GEMINI ||
     config.authType === AuthType.USE_VERTEX_AI
   ) {
-    const { createGeminiContentGenerator } = await import(
-      './geminiContentGenerator/index.js'
-    );
-    return createGeminiContentGenerator(config, gcConfig);
+    let headers: Record<string, string> = { ...baseHeaders };
+    if (gcConfig?.getUsageStatisticsEnabled()) {
+      const installationManager = new InstallationManager();
+      const installationId = installationManager.getInstallationId();
+      headers = {
+        ...headers,
+        'x-gemini-api-privileged-user-id': `${installationId}`,
+      };
+    }
+    const httpOptions = { headers };
+
+    const googleGenAI = new GoogleGenAI({
+      apiKey: config.apiKey === '' ? undefined : config.apiKey,
+      vertexai: config.vertexai,
+      httpOptions,
+    });
+    return new LoggingContentGenerator(googleGenAI.models, gcConfig);
   }
 
   if (config.authType === AuthType.USE_OPENAI) {

packages/core/src/core/coreToolScheduler.test.ts

@@ -240,7 +240,7 @@ describe('CoreToolScheduler', () => {
       getAllowedTools: () => [],
       getContentGeneratorConfig: () => ({
         model: 'test-model',
-        authType: 'gemini-api-key',
+        authType: 'oauth-personal',
       }),
       getShellExecutionConfig: () => ({
         terminalWidth: 90,
@@ -318,7 +318,7 @@ describe('CoreToolScheduler', () => {
       getAllowedTools: () => [],
       getContentGeneratorConfig: () => ({
         model: 'test-model',
-        authType: 'gemini-api-key',
+        authType: 'oauth-personal',
       }),
       getShellExecutionConfig: () => ({
         terminalWidth: 90,
@@ -497,7 +497,7 @@ describe('CoreToolScheduler', () => {
         getExcludeTools: () => ['write_file', 'edit', 'run_shell_command'],
         getContentGeneratorConfig: () => ({
           model: 'test-model',
-          authType: 'gemini-api-key',
+          authType: 'oauth-personal',
         }),
         getShellExecutionConfig: () => ({
           terminalWidth: 90,
@@ -584,7 +584,7 @@ describe('CoreToolScheduler', () => {
         getExcludeTools: () => ['write_file', 'edit'], // Different excluded tools
         getContentGeneratorConfig: () => ({
           model: 'test-model',
-          authType: 'gemini-api-key',
+          authType: 'oauth-personal',
         }),
         getShellExecutionConfig: () => ({
           terminalWidth: 90,
@@ -674,7 +674,7 @@ describe('CoreToolScheduler with payload', () => {
       getAllowedTools: () => [],
       getContentGeneratorConfig: () => ({
         model: 'test-model',
-        authType: 'gemini-api-key',
+        authType: 'oauth-personal',
       }),
       getShellExecutionConfig: () => ({
         terminalWidth: 90,
@@ -1001,7 +1001,7 @@ describe('CoreToolScheduler edit cancellation', () => {
       getAllowedTools: () => [],
       getContentGeneratorConfig: () => ({
         model: 'test-model',
-        authType: 'gemini-api-key',
+        authType: 'oauth-personal',
       }),
       getShellExecutionConfig: () => ({
         terminalWidth: 90,
@@ -1108,7 +1108,7 @@ describe('CoreToolScheduler YOLO mode', () => {
       getAllowedTools: () => [],
       getContentGeneratorConfig: () => ({
         model: 'test-model',
-        authType: 'gemini-api-key',
+        authType: 'oauth-personal',
       }),
       getShellExecutionConfig: () => ({
         terminalWidth: 90,
@@ -1258,7 +1258,7 @@ describe('CoreToolScheduler cancellation during executing with live output', ()
       getApprovalMode: () => ApprovalMode.DEFAULT,
       getContentGeneratorConfig: () => ({
         model: 'test-model',
-        authType: 'gemini-api-key',
+        authType: 'oauth-personal',
       }),
       getToolRegistry: () => mockToolRegistry,
       getShellExecutionConfig: () => ({
@@ -1350,7 +1350,7 @@ describe('CoreToolScheduler request queueing', () => {
       getAllowedTools: () => [],
       getContentGeneratorConfig: () => ({
         model: 'test-model',
-        authType: 'gemini-api-key',
+        authType: 'oauth-personal',
       }),
       getShellExecutionConfig: () => ({
         terminalWidth: 90,
@@ -1482,7 +1482,7 @@ describe('CoreToolScheduler request queueing', () => {
       getToolRegistry: () => toolRegistry,
       getContentGeneratorConfig: () => ({
         model: 'test-model',
-        authType: 'gemini-api-key',
+        authType: 'oauth-personal',
       }),
       getShellExecutionConfig: () => ({
         terminalWidth: 80,
@@ -1586,7 +1586,7 @@ describe('CoreToolScheduler request queueing', () => {
       getAllowedTools: () => [],
       getContentGeneratorConfig: () => ({
         model: 'test-model',
-        authType: 'gemini-api-key',
+        authType: 'oauth-personal',
       }),
       getShellExecutionConfig: () => ({
         terminalWidth: 90,
@@ -1854,7 +1854,7 @@ describe('CoreToolScheduler Sequential Execution', () => {
       getAllowedTools: () => [],
       getContentGeneratorConfig: () => ({
         model: 'test-model',
-        authType: 'gemini-api-key',
+        authType: 'oauth-personal',
       }),
       getShellExecutionConfig: () => ({
         terminalWidth: 90,
@@ -1975,7 +1975,7 @@ describe('CoreToolScheduler Sequential Execution', () => {
       getAllowedTools: () => [],
       getContentGeneratorConfig: () => ({
         model: 'test-model',
-        authType: 'gemini-api-key',
+        authType: 'oauth-personal',
       }),
       getShellExecutionConfig: () => ({
         terminalWidth: 90,

packages/core/src/core/geminiChat.test.ts

@@ -100,7 +100,6 @@ describe('GeminiChat', () => {
       countTokens: vi.fn(),
       embedContent: vi.fn(),
       batchEmbedContents: vi.fn(),
-      useSummarizedThinking: vi.fn().mockReturnValue(false),
     } as unknown as ContentGenerator;
 
     mockHandleFallback.mockClear();
@@ -112,7 +111,7 @@ describe('GeminiChat', () => {
       getUsageStatisticsEnabled: () => true,
       getDebugMode: () => false,
       getContentGeneratorConfig: vi.fn().mockReturnValue({
-        authType: 'gemini-api-key', // Ensure this is set for fallback tests
+        authType: 'oauth-personal', // Ensure this is set for fallback tests
         model: 'test-model',
       }),
       getModel: vi.fn().mockReturnValue('gemini-pro'),
@@ -719,99 +718,6 @@ describe('GeminiChat', () => {
         1,
       );
     });
-
-    it('should handle summarized thinking by conditionally including thoughts in history', async () => {
-      // Case 1: useSummarizedThinking is true -> thoughts NOT in history
-      vi.mocked(mockContentGenerator.useSummarizedThinking).mockReturnValue(
-        true,
-      );
-      const stream1 = (async function* () {
-        yield {
-          candidates: [
-            {
-              content: {
-                role: 'model',
-                parts: [{ thought: true, text: 'T1' }, { text: 'A1' }],
-              },
-              finishReason: 'STOP',
-            },
-          ],
-        } as unknown as GenerateContentResponse;
-      })();
-      vi.mocked(mockContentGenerator.generateContentStream).mockResolvedValue(
-        stream1,
-      );
-
-      const res1 = await chat.sendMessageStream('m1', { message: 'h1' }, 'p1');
-      for await (const _ of res1);
-
-      const history1 = chat.getHistory();
-      expect(history1[1].parts).toEqual([{ text: 'A1' }]);
-
-      // Case 2: useSummarizedThinking is false -> thoughts ARE in history
-      chat.clearHistory();
-      vi.mocked(mockContentGenerator.useSummarizedThinking).mockReturnValue(
-        false,
-      );
-      const stream2 = (async function* () {
-        yield {
-          candidates: [
-            {
-              content: {
-                role: 'model',
-                parts: [{ thought: true, text: 'T2' }, { text: 'A2' }],
-              },
-              finishReason: 'STOP',
-            },
-          ],
-        } as unknown as GenerateContentResponse;
-      })();
-      vi.mocked(mockContentGenerator.generateContentStream).mockResolvedValue(
-        stream2,
-      );
-
-      const res2 = await chat.sendMessageStream('m1', { message: 'h1' }, 'p2');
-      for await (const _ of res2);
-
-      const history2 = chat.getHistory();
-      expect(history2[1].parts).toEqual([
-        { text: 'T2', thought: true },
-        { text: 'A2' },
-      ]);
-    });
-
-    it('should keep parts with thoughtSignature when consolidating history', async () => {
-      const stream = (async function* () {
-        yield {
-          candidates: [
-            {
-              content: {
-                role: 'model',
-                parts: [
-                  {
-                    text: 'p1',
-                    thoughtSignature: 's1',
-                  } as unknown as { text: string; thoughtSignature: string },
-                ],
-              },
-              finishReason: 'STOP',
-            },
-          ],
-        } as unknown as GenerateContentResponse;
-      })();
-      vi.mocked(mockContentGenerator.generateContentStream).mockResolvedValue(
-        stream,
-      );
-
-      const res = await chat.sendMessageStream('m1', { message: 'h1' }, 'p1');
-      for await (const _ of res);
-
-      const history = chat.getHistory();
-      expect(history[1].parts![0]).toEqual({
-        text: 'p1',
-        thoughtSignature: 's1',
-      });
-    });
   });
 
   describe('addHistory', () => {
@@ -1476,7 +1382,7 @@ describe('GeminiChat', () => {
     });
 
     it('should call handleFallback with the specific failed model and retry if handler returns true', async () => {
-      const authType = AuthType.USE_GEMINI;
+      const authType = AuthType.LOGIN_WITH_GOOGLE;
       vi.mocked(mockConfig.getContentGeneratorConfig).mockReturnValue({
         model: 'test-model',
         authType,
@@ -1626,7 +1532,7 @@ describe('GeminiChat', () => {
   });
 
   describe('stripThoughtsFromHistory', () => {
-    it('should strip thoughts and thought signatures, and remove empty content objects', () => {
+    it('should strip thought signatures', () => {
       chat.setHistory([
         {
           role: 'user',
@@ -1638,15 +1544,10 @@ describe('GeminiChat', () => {
             { text: 'thinking...', thought: true },
             { text: 'hi' },
             {
-              text: 'hidden metadata',
-              thoughtSignature: 'abc',
-            } as unknown as { text: string; thoughtSignature: string },
+              functionCall: { name: 'test', args: {} },
+            },
           ],
         },
-        {
-          role: 'model',
-          parts: [{ text: 'only thinking', thought: true }],
-        },
       ]);
 
       chat.stripThoughtsFromHistory();
@@ -1658,7 +1559,7 @@ describe('GeminiChat', () => {
         },
         {
           role: 'model',
-          parts: [{ text: 'hi' }, { text: 'hidden metadata' }],
+          parts: [{ text: 'hi' }, { functionCall: { name: 'test', args: {} } }],
         },
       ]);
     });

packages/core/src/core/geminiChat.ts

@@ -92,7 +92,6 @@ export function isValidNonThoughtTextPart(part: Part): boolean {
   return (
     typeof part.text === 'string' &&
     !part.thought &&
-    !part.thoughtSignature &&
     // Technically, the model should never generate parts that have text and
     //  any of these but we don't trust them so check anyways.
     !part.functionCall &&
@@ -110,24 +109,18 @@ function isValidContent(content: Content): boolean {
     if (part === undefined || Object.keys(part).length === 0) {
       return false;
     }
-    if (!isValidContentPart(part)) {
+    if (
+      !part.thought &&
+      part.text !== undefined &&
+      part.text === '' &&
+      part.functionCall === undefined
+    ) {
       return false;
     }
   }
   return true;
 }
 
-function isValidContentPart(part: Part): boolean {
-  const isInvalid =
-    !part.thought &&
-    !part.thoughtSignature &&
-    part.text !== undefined &&
-    part.text === '' &&
-    part.functionCall === undefined;
-
-  return !isInvalid;
-}
-
 /**
  * Validates the history contains the correct roles.
  *
@@ -455,29 +448,15 @@ export class GeminiChat {
         if (!content.parts) return content;
 
         // Filter out thought parts entirely
-        const filteredParts = content.parts
-          .filter(
-            (part) =>
-              !(
-                part &&
-                typeof part === 'object' &&
-                'thought' in part &&
-                part.thought
-              ),
-          )
-          .map((part) => {
-            if (
+        const filteredParts = content.parts.filter(
+          (part) =>
+            !(
               part &&
               typeof part === 'object' &&
-              'thoughtSignature' in part
-            ) {
-              const newPart = { ...part };
-              delete (newPart as { thoughtSignature?: string })
-                .thoughtSignature;
-              return newPart;
-            }
-            return part;
-          });
+              'thought' in part &&
+              part.thought
+            ),
+        );
 
         return {
           ...content,
@@ -559,15 +538,11 @@ export class GeminiChat {
       yield chunk; // Yield every chunk to the UI immediately.
     }
 
-    let thoughtText = '';
-    // Only include thoughts if not using summarized thinking.
-    if (!this.config.getContentGenerator().useSummarizedThinking()) {
-      thoughtText = allModelParts
-        .filter((part) => part.thought)
-        .map((part) => part.text)
-        .join('')
-        .trim();
-    }
+    const thoughtParts = allModelParts.filter((part) => part.thought);
+    const thoughtText = thoughtParts
+      .map((part) => part.text)
+      .join('')
+      .trim();
 
     const contentParts = allModelParts.filter((part) => !part.thought);
     const consolidatedHistoryParts: Part[] = [];
@@ -580,7 +555,7 @@ export class GeminiChat {
         isValidNonThoughtTextPart(part)
       ) {
         lastPart.text += part.text;
-      } else if (isValidContentPart(part)) {
+      } else {
         consolidatedHistoryParts.push(part);
       }
     }

packages/core/src/core/geminiContentGenerator/geminiContentGenerator.test.ts

@@ -1,173 +0,0 @@
-/**
- * @license
- * Copyright 2025 Google LLC
- * SPDX-License-Identifier: Apache-2.0
- */
-
-import { describe, it, expect, vi, beforeEach } from 'vitest';
-import { GeminiContentGenerator } from './geminiContentGenerator.js';
-import { GoogleGenAI } from '@google/genai';
-
-vi.mock('@google/genai', () => {
-  const mockGenerateContent = vi.fn();
-  const mockGenerateContentStream = vi.fn();
-  const mockCountTokens = vi.fn();
-  const mockEmbedContent = vi.fn();
-
-  return {
-    GoogleGenAI: vi.fn().mockImplementation(() => ({
-      models: {
-        generateContent: mockGenerateContent,
-        generateContentStream: mockGenerateContentStream,
-        countTokens: mockCountTokens,
-        embedContent: mockEmbedContent,
-      },
-    })),
-  };
-});
-
-describe('GeminiContentGenerator', () => {
-  let generator: GeminiContentGenerator;
-  // eslint-disable-next-line @typescript-eslint/no-explicit-any
-  let mockGoogleGenAI: any;
-
-  beforeEach(() => {
-    vi.clearAllMocks();
-    generator = new GeminiContentGenerator({
-      apiKey: 'test-api-key',
-    });
-    mockGoogleGenAI = vi.mocked(GoogleGenAI).mock.results[0].value;
-  });
-
-  it('should call generateContent on the underlying model', async () => {
-    const request = { model: 'gemini-1.5-flash', contents: [] };
-    const expectedResponse = { responseId: 'test-id' };
-    mockGoogleGenAI.models.generateContent.mockResolvedValue(expectedResponse);
-
-    const response = await generator.generateContent(request, 'prompt-id');
-
-    expect(mockGoogleGenAI.models.generateContent).toHaveBeenCalledWith(
-      expect.objectContaining({
-        ...request,
-        config: expect.objectContaining({
-          temperature: 1,
-          topP: 0.95,
-          thinkingConfig: {
-            includeThoughts: true,
-            thinkingLevel: 'THINKING_LEVEL_UNSPECIFIED',
-          },
-        }),
-      }),
-    );
-    expect(response).toBe(expectedResponse);
-  });
-
-  it('should call generateContentStream on the underlying model', async () => {
-    const request = { model: 'gemini-1.5-flash', contents: [] };
-    const mockStream = (async function* () {
-      yield { responseId: '1' };
-    })();
-    mockGoogleGenAI.models.generateContentStream.mockResolvedValue(mockStream);
-
-    const stream = await generator.generateContentStream(request, 'prompt-id');
-
-    expect(mockGoogleGenAI.models.generateContentStream).toHaveBeenCalledWith(
-      expect.objectContaining({
-        ...request,
-        config: expect.objectContaining({
-          temperature: 1,
-          topP: 0.95,
-          thinkingConfig: {
-            includeThoughts: true,
-            thinkingLevel: 'THINKING_LEVEL_UNSPECIFIED',
-          },
-        }),
-      }),
-    );
-    expect(stream).toBe(mockStream);
-  });
-
-  it('should call countTokens on the underlying model', async () => {
-    const request = { model: 'gemini-1.5-flash', contents: [] };
-    const expectedResponse = { totalTokens: 10 };
-    mockGoogleGenAI.models.countTokens.mockResolvedValue(expectedResponse);
-
-    const response = await generator.countTokens(request);
-
-    expect(mockGoogleGenAI.models.countTokens).toHaveBeenCalledWith(request);
-    expect(response).toBe(expectedResponse);
-  });
-
-  it('should call embedContent on the underlying model', async () => {
-    const request = { model: 'embedding-model', contents: [] };
-    const expectedResponse = { embeddings: [] };
-    mockGoogleGenAI.models.embedContent.mockResolvedValue(expectedResponse);
-
-    const response = await generator.embedContent(request);
-
-    expect(mockGoogleGenAI.models.embedContent).toHaveBeenCalledWith(request);
-    expect(response).toBe(expectedResponse);
-  });
-
-  it('should prioritize contentGeneratorConfig samplingParams over request config', async () => {
-    const generatorWithParams = new GeminiContentGenerator({ apiKey: 'test' }, {
-      model: 'gemini-1.5-flash',
-      samplingParams: {
-        temperature: 0.1,
-        top_p: 0.2,
-      },
-      // eslint-disable-next-line @typescript-eslint/no-explicit-any
-    } as any);
-
-    const request = {
-      model: 'gemini-1.5-flash',
-      contents: [],
-      config: {
-        temperature: 0.9,
-        topP: 0.9,
-      },
-    };
-
-    await generatorWithParams.generateContent(request, 'prompt-id');
-
-    expect(mockGoogleGenAI.models.generateContent).toHaveBeenCalledWith(
-      expect.objectContaining({
-        config: expect.objectContaining({
-          temperature: 0.1,
-          topP: 0.2,
-        }),
-      }),
-    );
-  });
-
-  it('should map reasoning effort to thinkingConfig', async () => {
-    const generatorWithReasoning = new GeminiContentGenerator(
-      { apiKey: 'test' },
-      {
-        model: 'gemini-2.5-pro',
-        reasoning: {
-          effort: 'high',
-        },
-        // eslint-disable-next-line @typescript-eslint/no-explicit-any
-      } as any,
-    );
-
-    const request = {
-      model: 'gemini-2.5-pro',
-      contents: [],
-    };
-
-    await generatorWithReasoning.generateContent(request, 'prompt-id');
-
-    expect(mockGoogleGenAI.models.generateContent).toHaveBeenCalledWith(
-      expect.objectContaining({
-        config: expect.objectContaining({
-          thinkingConfig: {
-            includeThoughts: true,
-            thinkingLevel: 'HIGH',
-          },
-        }),
-      }),
-    );
-  });
-});

packages/core/src/core/geminiContentGenerator/geminiContentGenerator.ts

@@ -1,144 +0,0 @@
-/**
- * @license
- * Copyright 2025 Google LLC
- * SPDX-License-Identifier: Apache-2.0
- */
-
-import type {
-  CountTokensParameters,
-  CountTokensResponse,
-  EmbedContentParameters,
-  EmbedContentResponse,
-  GenerateContentParameters,
-  GenerateContentResponse,
-  GenerateContentConfig,
-  ThinkingLevel,
-} from '@google/genai';
-import { GoogleGenAI } from '@google/genai';
-import type {
-  ContentGenerator,
-  ContentGeneratorConfig,
-} from '../contentGenerator.js';
-
-/**
- * A wrapper for GoogleGenAI that implements the ContentGenerator interface.
- */
-export class GeminiContentGenerator implements ContentGenerator {
-  private readonly googleGenAI: GoogleGenAI;
-  private readonly contentGeneratorConfig?: ContentGeneratorConfig;
-
-  constructor(
-    options: {
-      apiKey?: string;
-      vertexai?: boolean;
-      httpOptions?: { headers: Record<string, string> };
-    },
-    contentGeneratorConfig?: ContentGeneratorConfig,
-  ) {
-    this.googleGenAI = new GoogleGenAI(options);
-    this.contentGeneratorConfig = contentGeneratorConfig;
-  }
-
-  private buildSamplingParameters(
-    request: GenerateContentParameters,
-  ): GenerateContentConfig {
-    const configSamplingParams = this.contentGeneratorConfig?.samplingParams;
-    const requestConfig = request.config || {};
-
-    // Helper function to get parameter value with priority: config > request > default
-    const getParameterValue = <T>(
-      configValue: T | undefined,
-      requestKey: keyof GenerateContentConfig,
-      defaultValue?: T,
-    ): T | undefined => {
-      const requestValue = requestConfig[requestKey] as T | undefined;
-
-      if (configValue !== undefined) return configValue;
-      if (requestValue !== undefined) return requestValue;
-      return defaultValue;
-    };
-
-    return {
-      ...requestConfig,
-      temperature: getParameterValue<number>(
-        configSamplingParams?.temperature,
-        'temperature',
-        1,
-      ),
-      topP: getParameterValue<number>(
-        configSamplingParams?.top_p,
-        'topP',
-        0.95,
-      ),
-      topK: getParameterValue<number>(configSamplingParams?.top_k, 'topK', 64),
-      maxOutputTokens: getParameterValue<number>(
-        configSamplingParams?.max_tokens,
-        'maxOutputTokens',
-      ),
-      presencePenalty: getParameterValue<number>(
-        configSamplingParams?.presence_penalty,
-        'presencePenalty',
-      ),
-      frequencyPenalty: getParameterValue<number>(
-        configSamplingParams?.frequency_penalty,
-        'frequencyPenalty',
-      ),
-      thinkingConfig: getParameterValue(
-        this.contentGeneratorConfig?.reasoning
-          ? {
-              includeThoughts: true,
-              thinkingLevel: (this.contentGeneratorConfig.reasoning.effort ===
-              'low'
-                ? 'LOW'
-                : this.contentGeneratorConfig.reasoning.effort === 'high'
-                  ? 'HIGH'
-                  : 'THINKING_LEVEL_UNSPECIFIED') as ThinkingLevel,
-            }
-          : undefined,
-        'thinkingConfig',
-        {
-          includeThoughts: true,
-          thinkingLevel: 'THINKING_LEVEL_UNSPECIFIED' as ThinkingLevel,
-        },
-      ),
-    };
-  }
-
-  async generateContent(
-    request: GenerateContentParameters,
-    _userPromptId: string,
-  ): Promise<GenerateContentResponse> {
-    const finalRequest = {
-      ...request,
-      config: this.buildSamplingParameters(request),
-    };
-    return this.googleGenAI.models.generateContent(finalRequest);
-  }
-
-  async generateContentStream(
-    request: GenerateContentParameters,
-    _userPromptId: string,
-  ): Promise<AsyncGenerator<GenerateContentResponse>> {
-    const finalRequest = {
-      ...request,
-      config: this.buildSamplingParameters(request),
-    };
-    return this.googleGenAI.models.generateContentStream(finalRequest);
-  }
-
-  async countTokens(
-    request: CountTokensParameters,
-  ): Promise<CountTokensResponse> {
-    return this.googleGenAI.models.countTokens(request);
-  }
-
-  async embedContent(
-    request: EmbedContentParameters,
-  ): Promise<EmbedContentResponse> {
-    return this.googleGenAI.models.embedContent(request);
-  }
-
-  useSummarizedThinking(): boolean {
-    return true;
-  }
-}

packages/core/src/core/geminiContentGenerator/index.test.ts

@@ -1,47 +0,0 @@
-/**
- * @license
- * Copyright 2025 Google LLC
- * SPDX-License-Identifier: Apache-2.0
- */
-
-import { describe, it, expect, vi, beforeEach } from 'vitest';
-import { createGeminiContentGenerator } from './index.js';
-import { GeminiContentGenerator } from './geminiContentGenerator.js';
-import { LoggingContentGenerator } from './loggingContentGenerator.js';
-import type { Config } from '../../config/config.js';
-import { AuthType } from '../contentGenerator.js';
-
-vi.mock('./geminiContentGenerator.js', () => ({
-  GeminiContentGenerator: vi.fn().mockImplementation(() => ({})),
-}));
-
-vi.mock('./loggingContentGenerator.js', () => ({
-  LoggingContentGenerator: vi.fn().mockImplementation((wrapped) => wrapped),
-}));
-
-describe('createGeminiContentGenerator', () => {
-  let mockConfig: Config;
-
-  beforeEach(() => {
-    vi.clearAllMocks();
-    mockConfig = {
-      getUsageStatisticsEnabled: vi.fn().mockReturnValue(false),
-      getContentGeneratorConfig: vi.fn().mockReturnValue({}),
-      getCliVersion: vi.fn().mockReturnValue('1.0.0'),
-    } as unknown as Config;
-  });
-
-  it('should create a GeminiContentGenerator wrapped in LoggingContentGenerator', () => {
-    const config = {
-      model: 'gemini-1.5-flash',
-      apiKey: 'test-key',
-      authType: AuthType.USE_GEMINI,
-    };
-
-    const generator = createGeminiContentGenerator(config, mockConfig);
-
-    expect(GeminiContentGenerator).toHaveBeenCalled();
-    expect(LoggingContentGenerator).toHaveBeenCalled();
-    expect(generator).toBeDefined();
-  });
-});

packages/core/src/core/geminiContentGenerator/index.ts

@@ -1,55 +0,0 @@
-/**
- * @license
- * Copyright 2025 Google LLC
- * SPDX-License-Identifier: Apache-2.0
- */
-
-import { GeminiContentGenerator } from './geminiContentGenerator.js';
-import type {
-  ContentGenerator,
-  ContentGeneratorConfig,
-} from '../contentGenerator.js';
-import type { Config } from '../../config/config.js';
-import { InstallationManager } from '../../utils/installationManager.js';
-import { LoggingContentGenerator } from './loggingContentGenerator.js';
-
-export { GeminiContentGenerator } from './geminiContentGenerator.js';
-export { LoggingContentGenerator } from './loggingContentGenerator.js';
-
-/**
- * Create a Gemini content generator.
- */
-export function createGeminiContentGenerator(
-  config: ContentGeneratorConfig,
-  gcConfig: Config,
-): ContentGenerator {
-  const version = process.env['CLI_VERSION'] || process.version;
-  const userAgent =
-    config.userAgent ||
-    `QwenCode/${version} (${process.platform}; ${process.arch})`;
-  const baseHeaders: Record<string, string> = {
-    'User-Agent': userAgent,
-  };
-
-  let headers: Record<string, string> = { ...baseHeaders };
-  if (gcConfig?.getUsageStatisticsEnabled()) {
-    const installationManager = new InstallationManager();
-    const installationId = installationManager.getInstallationId();
-    headers = {
-      ...headers,
-      'x-gemini-api-privileged-user-id': `${installationId}`,
-    };
-  }
-  const httpOptions = { headers };
-
-  const geminiContentGenerator = new GeminiContentGenerator(
-    {
-      apiKey: config.apiKey === '' ? undefined : config.apiKey,
-      vertexai: config.vertexai,
-      httpOptions,
-    },
-    config,
-  );
-
-  return new LoggingContentGenerator(geminiContentGenerator, gcConfig);
-}

packages/core/src/core/loggingContentGenerator.ts

@@ -13,24 +13,21 @@ import type {
   GenerateContentParameters,
   GenerateContentResponseUsageMetadata,
   GenerateContentResponse,
-  ContentListUnion,
-  ContentUnion,
-  Part,
-  PartUnion,
 } from '@google/genai';
 import {
   ApiRequestEvent,
   ApiResponseEvent,
   ApiErrorEvent,
-} from '../../telemetry/types.js';
-import type { Config } from '../../config/config.js';
+} from '../telemetry/types.js';
+import type { Config } from '../config/config.js';
 import {
   logApiError,
   logApiRequest,
   logApiResponse,
-} from '../../telemetry/loggers.js';
-import type { ContentGenerator } from '../contentGenerator.js';
-import { isStructuredError } from '../../utils/quotaErrorDetection.js';
+} from '../telemetry/loggers.js';
+import type { ContentGenerator } from './contentGenerator.js';
+import { toContents } from '../code_assist/converter.js';
+import { isStructuredError } from '../utils/quotaErrorDetection.js';
 
 interface StructuredError {
   status: number;
@@ -115,7 +112,7 @@ export class LoggingContentGenerator implements ContentGenerator {
     userPromptId: string,
   ): Promise<GenerateContentResponse> {
     const startTime = Date.now();
-    this.logApiRequest(this.toContents(req.contents), req.model, userPromptId);
+    this.logApiRequest(toContents(req.contents), req.model, userPromptId);
     try {
       const response = await this.wrapped.generateContent(req, userPromptId);
       const durationMs = Date.now() - startTime;
@@ -140,7 +137,7 @@ export class LoggingContentGenerator implements ContentGenerator {
     userPromptId: string,
   ): Promise<AsyncGenerator<GenerateContentResponse>> {
     const startTime = Date.now();
-    this.logApiRequest(this.toContents(req.contents), req.model, userPromptId);
+    this.logApiRequest(toContents(req.contents), req.model, userPromptId);
 
     let stream: AsyncGenerator<GenerateContentResponse>;
     try {
@@ -208,95 +205,4 @@ export class LoggingContentGenerator implements ContentGenerator {
   ): Promise<EmbedContentResponse> {
     return this.wrapped.embedContent(req);
   }
-
-  useSummarizedThinking(): boolean {
-    return this.wrapped.useSummarizedThinking();
-  }
-
-  private toContents(contents: ContentListUnion): Content[] {
-    if (Array.isArray(contents)) {
-      // it's a Content[] or a PartsUnion[]
-      return contents.map((c) => this.toContent(c));
-    }
-    // it's a Content or a PartsUnion
-    return [this.toContent(contents)];
-  }
-
-  private toContent(content: ContentUnion): Content {
-    if (Array.isArray(content)) {
-      // it's a PartsUnion[]
-      return {
-        role: 'user',
-        parts: this.toParts(content),
-      };
-    }
-    if (typeof content === 'string') {
-      // it's a string
-      return {
-        role: 'user',
-        parts: [{ text: content }],
-      };
-    }
-    if ('parts' in content) {
-      // it's a Content - process parts to handle thought filtering
-      return {
-        ...content,
-        parts: content.parts
-          ? this.toParts(content.parts.filter((p) => p != null))
-          : [],
-      };
-    }
-    // it's a Part
-    return {
-      role: 'user',
-      parts: [this.toPart(content as Part)],
-    };
-  }
-
-  private toParts(parts: PartUnion[]): Part[] {
-    return parts.map((p) => this.toPart(p));
-  }
-
-  private toPart(part: PartUnion): Part {
-    if (typeof part === 'string') {
-      // it's a string
-      return { text: part };
-    }
-
-    // Handle thought parts for CountToken API compatibility
-    // The CountToken API expects parts to have certain required "oneof" fields initialized,
-    // but thought parts don't conform to this schema and cause API failures
-    if ('thought' in part && part.thought) {
-      const thoughtText = `[Thought: ${part.thought}]`;
-
-      const newPart = { ...part };
-      delete (newPart as Record<string, unknown>)['thought'];
-
-      const hasApiContent =
-        'functionCall' in newPart ||
-        'functionResponse' in newPart ||
-        'inlineData' in newPart ||
-        'fileData' in newPart;
-
-      if (hasApiContent) {
-        // It's a functionCall or other non-text part. Just strip the thought.
-        return newPart;
-      }
-
-      // If no other valid API content, this must be a text part.
-      // Combine existing text (if any) with the thought, preserving other properties.
-      const text = (newPart as { text?: unknown }).text;
-      const existingText = text ? String(text) : '';
-      const combinedText = existingText
-        ? `${existingText}\n${thoughtText}`
-        : thoughtText;
-
-      return {
-        ...newPart,
-        text: combinedText,
-      };
-    }
-
-    return part;
-  }
 }

packages/core/src/core/nonInteractiveToolExecutor.test.ts

@@ -47,7 +47,7 @@ describe('executeToolCall', () => {
       getDebugMode: () => false,
       getContentGeneratorConfig: () => ({
         model: 'test-model',
-        authType: 'gemini-api-key',
+        authType: 'oauth-personal',
       }),
       getShellExecutionConfig: () => ({
         terminalWidth: 90,

packages/core/src/core/openaiContentGenerator/openaiContentGenerator.test.ts

@@ -99,7 +99,6 @@ describe('OpenAIContentGenerator (Refactored)', () => {
         },
       } as unknown as OpenAI),
       buildRequest: vi.fn().mockImplementation((req) => req),
-      getDefaultGenerationConfig: vi.fn().mockReturnValue({}),
     };
 
     generator = new OpenAIContentGenerator(
@@ -212,7 +211,6 @@ describe('OpenAIContentGenerator (Refactored)', () => {
           },
         } as unknown as OpenAI),
         buildRequest: vi.fn().mockImplementation((req) => req),
-        getDefaultGenerationConfig: vi.fn().mockReturnValue({}),
       };
 
       const testGenerator = new TestGenerator(
@@ -279,7 +277,6 @@ describe('OpenAIContentGenerator (Refactored)', () => {
           },
         } as unknown as OpenAI),
         buildRequest: vi.fn().mockImplementation((req) => req),
-        getDefaultGenerationConfig: vi.fn().mockReturnValue({}),
       };
 
       const testGenerator = new TestGenerator(

packages/core/src/core/openaiContentGenerator/openaiContentGenerator.ts

@@ -154,8 +154,4 @@ export class OpenAIContentGenerator implements ContentGenerator {
       );
     }
   }
-
-  useSummarizedThinking(): boolean {
-    return false;
-  }
 }

packages/core/src/core/openaiContentGenerator/pipeline.test.ts

@@ -60,7 +60,6 @@ describe('ContentGenerationPipeline', () => {
       buildClient: vi.fn().mockReturnValue(mockClient),
       buildRequest: vi.fn().mockImplementation((req) => req),
       buildHeaders: vi.fn().mockReturnValue({}),
-      getDefaultGenerationConfig: vi.fn().mockReturnValue({}),
     };
 
     // Mock telemetry service

packages/core/src/core/openaiContentGenerator/pipeline.ts

@@ -283,22 +283,16 @@ export class ContentGenerationPipeline {
   private buildSamplingParameters(
     request: GenerateContentParameters,
   ): Record<string, unknown> {
-    const defaultSamplingParams =
-      this.config.provider.getDefaultGenerationConfig();
     const configSamplingParams = this.contentGeneratorConfig.samplingParams;
 
     // Helper function to get parameter value with priority: config > request > default
     const getParameterValue = <T>(
       configKey: keyof NonNullable<typeof configSamplingParams>,
-      requestKey?: keyof NonNullable<typeof request.config>,
+      requestKey: keyof NonNullable<typeof request.config>,
+      defaultValue?: T,
     ): T | undefined => {
       const configValue = configSamplingParams?.[configKey] as T | undefined;
-      const requestValue = requestKey
-        ? (request.config?.[requestKey] as T | undefined)
-        : undefined;
-      const defaultValue = requestKey
-        ? (defaultSamplingParams[requestKey] as T)
-        : undefined;
+      const requestValue = request.config?.[requestKey] as T | undefined;
 
       if (configValue !== undefined) return configValue;
       if (requestValue !== undefined) return requestValue;
@@ -310,8 +304,12 @@ export class ContentGenerationPipeline {
       key: string,
       configKey: keyof NonNullable<typeof configSamplingParams>,
       requestKey?: keyof NonNullable<typeof request.config>,
-    ): Record<string, T | undefined> => {
-      const value = getParameterValue<T>(configKey, requestKey);
+      defaultValue?: T,
+    ): Record<string, T> | Record<string, never> => {
+      const value = requestKey
+        ? getParameterValue(configKey, requestKey, defaultValue)
+        : ((configSamplingParams?.[configKey] as T | undefined) ??
+          defaultValue);
 
       return value !== undefined ? { [key]: value } : {};
     };
@@ -325,18 +323,10 @@ export class ContentGenerationPipeline {
       ...addParameterIfDefined('max_tokens', 'max_tokens', 'maxOutputTokens'),
 
       // Config-only parameters (no request fallback)
-      ...addParameterIfDefined('top_k', 'top_k', 'topK'),
+      ...addParameterIfDefined('top_k', 'top_k'),
       ...addParameterIfDefined('repetition_penalty', 'repetition_penalty'),
-      ...addParameterIfDefined(
-        'presence_penalty',
-        'presence_penalty',
-        'presencePenalty',
-      ),
-      ...addParameterIfDefined(
-        'frequency_penalty',
-        'frequency_penalty',
-        'frequencyPenalty',
-      ),
+      ...addParameterIfDefined('presence_penalty', 'presence_penalty'),
+      ...addParameterIfDefined('frequency_penalty', 'frequency_penalty'),
     };
 
     return params;

packages/core/src/core/openaiContentGenerator/provider/dashscope.ts

@@ -1,5 +1,4 @@
 import OpenAI from 'openai';
-import type { GenerateContentConfig } from '@google/genai';
 import type { Config } from '../../../config/config.js';
 import type { ContentGeneratorConfig } from '../../contentGenerator.js';
 import { AuthType } from '../../contentGenerator.js';
@@ -142,14 +141,6 @@ export class DashScopeOpenAICompatibleProvider
     };
   }
 
-  getDefaultGenerationConfig(): GenerateContentConfig {
-    return {
-      temperature: 0.7,
-      topP: 0.8,
-      topK: 20,
-    };
-  }
-
   /**
    * Add cache control flag to specified message(s) for DashScope providers
    */

packages/core/src/core/openaiContentGenerator/provider/deepseek.ts

@@ -8,7 +8,6 @@ import type OpenAI from 'openai';
 import type { Config } from '../../../config/config.js';
 import type { ContentGeneratorConfig } from '../../contentGenerator.js';
 import { DefaultOpenAICompatibleProvider } from './default.js';
-import type { GenerateContentConfig } from '@google/genai';
 
 export class DeepSeekOpenAICompatibleProvider extends DefaultOpenAICompatibleProvider {
   constructor(
@@ -77,10 +76,4 @@ export class DeepSeekOpenAICompatibleProvider extends DefaultOpenAICompatiblePro
       messages,
     };
   }
-
-  override getDefaultGenerationConfig(): GenerateContentConfig {
-    return {
-      temperature: 0,
-    };
-  }
 }

packages/core/src/core/openaiContentGenerator/provider/default.ts

@@ -1,5 +1,4 @@
 import OpenAI from 'openai';
-import type { GenerateContentConfig } from '@google/genai';
 import type { Config } from '../../../config/config.js';
 import type { ContentGeneratorConfig } from '../../contentGenerator.js';
 import { DEFAULT_TIMEOUT, DEFAULT_MAX_RETRIES } from '../constants.js';
@@ -56,10 +55,4 @@ export class DefaultOpenAICompatibleProvider
       ...request, // Preserve all original parameters including sampling params
     };
   }
-
-  getDefaultGenerationConfig(): GenerateContentConfig {
-    return {
-      topP: 0.95,
-    };
-  }
 }

packages/core/src/core/openaiContentGenerator/provider/types.ts

@@ -1,4 +1,3 @@
-import type { GenerateContentConfig } from '@google/genai';
 import type OpenAI from 'openai';
 
 // Extended types to support cache_control for DashScope
@@ -23,7 +22,6 @@ export interface OpenAICompatibleProvider {
     request: OpenAI.Chat.ChatCompletionCreateParams,
     userPromptId: string,
   ): OpenAI.Chat.ChatCompletionCreateParams;
-  getDefaultGenerationConfig(): GenerateContentConfig;
 }
 
 export type DashScopeRequestMetadata = {

packages/core/src/core/turn.test.ts

@@ -36,6 +36,13 @@ vi.mock('../utils/errorReporting', () => ({
   reportError: vi.fn(),
 }));
 
+// Use the actual implementation from partUtils now that it's provided.
+vi.mock('../utils/generateContentResponseUtilities', () => ({
+  getResponseText: (resp: GenerateContentResponse) =>
+    resp.candidates?.[0]?.content?.parts?.map((part) => part.text).join('') ||
+    undefined,
+}));
+
 describe('Turn', () => {
   let turn: Turn;
   // Define a type for the mocked Chat instance for clarity
@@ -149,7 +156,6 @@ describe('Turn', () => {
           type: GeminiEventType.Thought,
           value: { subject: '', description: 'reasoning...' },
         },
-        { type: GeminiEventType.Content, value: 'final answer' },
       ]);
     });
 

packages/core/src/core/turn.ts

@@ -27,11 +27,7 @@ import {
   toFriendlyError,
 } from '../utils/errors.js';
 import type { GeminiChat } from './geminiChat.js';
-import {
-  getThoughtText,
-  parseThought,
-  type ThoughtSummary,
-} from '../utils/thoughtUtils.js';
+import { getThoughtText, type ThoughtSummary } from '../utils/thoughtUtils.js';
 
 // Define a structure for tools passed to the server
 export interface ServerTool {
@@ -270,12 +266,13 @@ export class Turn {
           this.currentResponseId = resp.responseId;
         }
 
-        const thoughtText = getThoughtText(resp);
-        if (thoughtText) {
+        const thoughtPart = getThoughtText(resp);
+        if (thoughtPart) {
           yield {
             type: GeminiEventType.Thought,
-            value: parseThought(thoughtText),
+            value: { subject: '', description: thoughtPart },
           };
+          continue;
         }
 
         const text = getResponseText(resp);

packages/core/src/fallback/handler.test.ts

@@ -4,10 +4,36 @@
  * SPDX-License-Identifier: Apache-2.0
  */
 
-import { describe, it, expect, vi, beforeEach } from 'vitest';
+import {
+  describe,
+  it,
+  expect,
+  vi,
+  beforeEach,
+  type Mock,
+  type MockInstance,
+  afterEach,
+} from 'vitest';
 import { handleFallback } from './handler.js';
 import type { Config } from '../config/config.js';
 import { AuthType } from '../core/contentGenerator.js';
+import {
+  DEFAULT_GEMINI_FLASH_MODEL,
+  DEFAULT_GEMINI_MODEL,
+} from '../config/models.js';
+import { logFlashFallback } from '../telemetry/index.js';
+import type { FallbackModelHandler } from './types.js';
+
+// Mock the telemetry logger and event class
+vi.mock('../telemetry/index.js', () => ({
+  logFlashFallback: vi.fn(),
+  FlashFallbackEvent: class {},
+}));
+
+const MOCK_PRO_MODEL = DEFAULT_GEMINI_MODEL;
+const FALLBACK_MODEL = DEFAULT_GEMINI_FLASH_MODEL;
+const AUTH_OAUTH = AuthType.LOGIN_WITH_GOOGLE;
+const AUTH_API_KEY = AuthType.USE_GEMINI;
 
 const createMockConfig = (overrides: Partial<Config> = {}): Config =>
   ({
@@ -19,28 +45,174 @@ const createMockConfig = (overrides: Partial<Config> = {}): Config =>
 
 describe('handleFallback', () => {
   let mockConfig: Config;
+  let mockHandler: Mock<FallbackModelHandler>;
+  let consoleErrorSpy: MockInstance;
 
   beforeEach(() => {
     vi.clearAllMocks();
-    mockConfig = createMockConfig();
+    mockHandler = vi.fn();
+    // Default setup: OAuth user, Pro model failed, handler injected
+    mockConfig = createMockConfig({
+      fallbackModelHandler: mockHandler,
+    });
+    consoleErrorSpy = vi.spyOn(console, 'error').mockImplementation(() => {});
   });
 
-  it('should return null for unknown auth types', async () => {
+  afterEach(() => {
+    consoleErrorSpy.mockRestore();
+  });
+
+  it('should return null immediately if authType is not OAuth', async () => {
     const result = await handleFallback(
       mockConfig,
-      'test-model',
-      'unknown-auth',
+      MOCK_PRO_MODEL,
+      AUTH_API_KEY,
+    );
+    expect(result).toBeNull();
+    expect(mockHandler).not.toHaveBeenCalled();
+    expect(mockConfig.setFallbackMode).not.toHaveBeenCalled();
+  });
+
+  it('should return null if the failed model is already the fallback model', async () => {
+    const result = await handleFallback(
+      mockConfig,
+      FALLBACK_MODEL, // Failed model is Flash
+      AUTH_OAUTH,
+    );
+    expect(result).toBeNull();
+    expect(mockHandler).not.toHaveBeenCalled();
+  });
+
+  it('should return null if no fallbackHandler is injected in config', async () => {
+    const configWithoutHandler = createMockConfig({
+      fallbackModelHandler: undefined,
+    });
+    const result = await handleFallback(
+      configWithoutHandler,
+      MOCK_PRO_MODEL,
+      AUTH_OAUTH,
     );
     expect(result).toBeNull();
   });
 
-  it('should handle Qwen OAuth error', async () => {
-    const result = await handleFallback(
-      mockConfig,
-      'test-model',
-      AuthType.QWEN_OAUTH,
-      new Error('unauthorized'),
+  describe('when handler returns "retry"', () => {
+    it('should activate fallback mode, log telemetry, and return true', async () => {
+      mockHandler.mockResolvedValue('retry');
+
+      const result = await handleFallback(
+        mockConfig,
+        MOCK_PRO_MODEL,
+        AUTH_OAUTH,
+      );
+
+      expect(result).toBe(true);
+      expect(mockConfig.setFallbackMode).toHaveBeenCalledWith(true);
+      expect(logFlashFallback).toHaveBeenCalled();
+    });
+  });
+
+  describe('when handler returns "stop"', () => {
+    it('should activate fallback mode, log telemetry, and return false', async () => {
+      mockHandler.mockResolvedValue('stop');
+
+      const result = await handleFallback(
+        mockConfig,
+        MOCK_PRO_MODEL,
+        AUTH_OAUTH,
+      );
+
+      expect(result).toBe(false);
+      expect(mockConfig.setFallbackMode).toHaveBeenCalledWith(true);
+      expect(logFlashFallback).toHaveBeenCalled();
+    });
+  });
+
+  describe('when handler returns "auth"', () => {
+    it('should NOT activate fallback mode and return false', async () => {
+      mockHandler.mockResolvedValue('auth');
+
+      const result = await handleFallback(
+        mockConfig,
+        MOCK_PRO_MODEL,
+        AUTH_OAUTH,
+      );
+
+      expect(result).toBe(false);
+      expect(mockConfig.setFallbackMode).not.toHaveBeenCalled();
+      expect(logFlashFallback).not.toHaveBeenCalled();
+    });
+  });
+
+  describe('when handler returns an unexpected value', () => {
+    it('should log an error and return null', async () => {
+      mockHandler.mockResolvedValue(null);
+
+      const result = await handleFallback(
+        mockConfig,
+        MOCK_PRO_MODEL,
+        AUTH_OAUTH,
+      );
+
+      expect(result).toBeNull();
+      expect(consoleErrorSpy).toHaveBeenCalledWith(
+        'Fallback UI handler failed:',
+        new Error(
+          'Unexpected fallback intent received from fallbackModelHandler: "null"',
+        ),
+      );
+      expect(mockConfig.setFallbackMode).not.toHaveBeenCalled();
+    });
+  });
+
+  it('should pass the correct context (failedModel, fallbackModel, error) to the handler', async () => {
+    const mockError = new Error('Quota Exceeded');
+    mockHandler.mockResolvedValue('retry');
+
+    await handleFallback(mockConfig, MOCK_PRO_MODEL, AUTH_OAUTH, mockError);
+
+    expect(mockHandler).toHaveBeenCalledWith(
+      MOCK_PRO_MODEL,
+      FALLBACK_MODEL,
+      mockError,
     );
+  });
+
+  it('should not call setFallbackMode or log telemetry if already in fallback mode', async () => {
+    // Setup config where fallback mode is already active
+    const activeFallbackConfig = createMockConfig({
+      fallbackModelHandler: mockHandler,
+      isInFallbackMode: vi.fn(() => true), // Already active
+      setFallbackMode: vi.fn(),
+    });
+
+    mockHandler.mockResolvedValue('retry');
+
+    const result = await handleFallback(
+      activeFallbackConfig,
+      MOCK_PRO_MODEL,
+      AUTH_OAUTH,
+    );
+
+    // Should still return true to allow the retry (which will use the active fallback mode)
+    expect(result).toBe(true);
+    // Should still consult the handler
+    expect(mockHandler).toHaveBeenCalled();
+    // But should not mutate state or log telemetry again
+    expect(activeFallbackConfig.setFallbackMode).not.toHaveBeenCalled();
+    expect(logFlashFallback).not.toHaveBeenCalled();
+  });
+
+  it('should catch errors from the handler, log an error, and return null', async () => {
+    const handlerError = new Error('UI interaction failed');
+    mockHandler.mockRejectedValue(handlerError);
+
+    const result = await handleFallback(mockConfig, MOCK_PRO_MODEL, AUTH_OAUTH);
+
     expect(result).toBeNull();
+    expect(consoleErrorSpy).toHaveBeenCalledWith(
+      'Fallback UI handler failed:',
+      handlerError,
+    );
+    expect(mockConfig.setFallbackMode).not.toHaveBeenCalled();
   });
 });

packages/core/src/fallback/handler.ts

@@ -6,6 +6,8 @@
 
 import type { Config } from '../config/config.js';
 import { AuthType } from '../core/contentGenerator.js';
+import { DEFAULT_GEMINI_FLASH_MODEL } from '../config/models.js';
+import { logFlashFallback, FlashFallbackEvent } from '../telemetry/index.js';
 
 export async function handleFallback(
   config: Config,
@@ -18,7 +20,48 @@ export async function handleFallback(
     return handleQwenOAuthError(error);
   }
 
-  return null;
+  // Applicability Checks
+  if (authType !== AuthType.LOGIN_WITH_GOOGLE) return null;
+
+  const fallbackModel = DEFAULT_GEMINI_FLASH_MODEL;
+
+  if (failedModel === fallbackModel) return null;
+
+  // Consult UI Handler for Intent
+  const fallbackModelHandler = config.fallbackModelHandler;
+  if (typeof fallbackModelHandler !== 'function') return null;
+
+  try {
+    // Pass the specific failed model to the UI handler.
+    const intent = await fallbackModelHandler(
+      failedModel,
+      fallbackModel,
+      error,
+    );
+
+    // Process Intent and Update State
+    switch (intent) {
+      case 'retry':
+        // Activate fallback mode. The NEXT retry attempt will pick this up.
+        activateFallbackMode(config, authType);
+        return true; // Signal retryWithBackoff to continue.
+
+      case 'stop':
+        activateFallbackMode(config, authType);
+        return false;
+
+      case 'auth':
+        return false;
+
+      default:
+        throw new Error(
+          `Unexpected fallback intent received from fallbackModelHandler: "${intent}"`,
+        );
+    }
+  } catch (handlerError) {
+    console.error('Fallback UI handler failed:', handlerError);
+    return null;
+  }
 }
 
 /**
@@ -75,3 +118,12 @@ async function handleQwenOAuthError(error?: unknown): Promise<string | null> {
   // For other errors, don't handle them specially
   return null;
 }
+
+function activateFallbackMode(config: Config, authType: string | undefined) {
+  if (!config.isInFallbackMode()) {
+    config.setFallbackMode(true);
+    if (authType) {
+      logFlashFallback(config, new FlashFallbackEvent(authType));
+    }
+  }
+}

packages/core/src/index.ts

@@ -12,6 +12,7 @@ export * from './output/json-formatter.js';
 // Export Core Logic
 export * from './core/client.js';
 export * from './core/contentGenerator.js';
+export * from './core/loggingContentGenerator.js';
 export * from './core/geminiChat.js';
 export * from './core/logger.js';
 export * from './core/prompts.js';
@@ -23,7 +24,11 @@ export * from './core/nonInteractiveToolExecutor.js';
 
 export * from './fallback/types.js';
 
+export * from './code_assist/codeAssist.js';
+export * from './code_assist/oauth2.js';
 export * from './qwen/qwenOAuth2.js';
+export * from './code_assist/server.js';
+export * from './code_assist/types.js';
 
 // Export utilities
 export * from './utils/paths.js';

packages/core/src/qwen/qwenOAuth2.ts

@@ -907,5 +907,3 @@ export async function clearQwenCredentials(): Promise<void> {
 function getQwenCachedCredentialPath(): string {
   return path.join(os.homedir(), QWEN_DIR, QWEN_CREDENTIAL_FILENAME);
 }
-
-export const clearCachedCredentialFile = clearQwenCredentials;

packages/core/src/telemetry/clearcut-logger/clearcut-logger.test.ts

@@ -30,6 +30,7 @@ import {
   ToolCallEvent,
 } from '../types.js';
 import { GIT_COMMIT_INFO, CLI_VERSION } from '../../generated/git-commit.js';
+import { UserAccountManager } from '../../utils/userAccountManager.js';
 import { InstallationManager } from '../../utils/installationManager.js';
 import { safeJsonStringify } from '../../utils/safeJsonStringify.js';
 
@@ -89,8 +90,10 @@ expect.extend({
   },
 });
 
+vi.mock('../../utils/userAccountManager.js');
 vi.mock('../../utils/installationManager.js');
 
+const mockUserAccount = vi.mocked(UserAccountManager.prototype);
 const mockInstallMgr = vi.mocked(InstallationManager.prototype);
 
 // TODO(richieforeman): Consider moving this to test setup globally.
@@ -125,7 +128,11 @@ describe('ClearcutLogger', () => {
     vi.unstubAllEnvs();
   });
 
-  function setup({ config = {} as Partial<ConfigParameters> } = {}) {
+  function setup({
+    config = {} as Partial<ConfigParameters>,
+    lifetimeGoogleAccounts = 1,
+    cachedGoogleAccount = 'test@google.com',
+  } = {}) {
     server.resetHandlers(
       http.post(CLEARCUT_URL, () => HttpResponse.text(EXAMPLE_RESPONSE)),
     );
@@ -139,6 +146,10 @@ describe('ClearcutLogger', () => {
     });
     ClearcutLogger.clearInstance();
 
+    mockUserAccount.getCachedGoogleAccount.mockReturnValue(cachedGoogleAccount);
+    mockUserAccount.getLifetimeGoogleAccounts.mockReturnValue(
+      lifetimeGoogleAccounts,
+    );
     mockInstallMgr.getInstallationId = vi
       .fn()
       .mockReturnValue('test-installation-id');
@@ -184,6 +195,19 @@ describe('ClearcutLogger', () => {
   });
 
   describe('createLogEvent', () => {
+    it('logs the total number of google accounts', () => {
+      const { logger } = setup({
+        lifetimeGoogleAccounts: 9001,
+      });
+
+      const event = logger?.createLogEvent(EventNames.API_ERROR, []);
+
+      expect(event?.event_metadata[0]).toContainEqual({
+        gemini_cli_key: EventMetadataKey.GEMINI_CLI_GOOGLE_ACCOUNTS_COUNT,
+        value: '9001',
+      });
+    });
+
     it('logs the current surface from a github action', () => {
       const { logger } = setup({});
 
@@ -227,6 +251,7 @@ describe('ClearcutLogger', () => {
       // Define expected values
       const session_id = 'test-session-id';
       const auth_type = AuthType.USE_GEMINI;
+      const google_accounts = 123;
       const surface = 'ide-1234';
       const cli_version = CLI_VERSION;
       const git_commit_hash = GIT_COMMIT_INFO;
@@ -235,6 +260,7 @@ describe('ClearcutLogger', () => {
 
       // Setup logger with expected values
       const { logger, loggerConfig } = setup({
+        lifetimeGoogleAccounts: google_accounts,
         config: {},
       });
       vi.spyOn(loggerConfig, 'getContentGeneratorConfig').mockReturnValue({
@@ -257,6 +283,10 @@ describe('ClearcutLogger', () => {
             gemini_cli_key: EventMetadataKey.GEMINI_CLI_AUTH_TYPE,
             value: JSON.stringify(auth_type),
           },
+          {
+            gemini_cli_key: EventMetadataKey.GEMINI_CLI_GOOGLE_ACCOUNTS_COUNT,
+            value: `${google_accounts}`,
+          },
           {
             gemini_cli_key: EventMetadataKey.GEMINI_CLI_SURFACE,
             value: surface,
@@ -374,14 +404,10 @@ describe('ClearcutLogger', () => {
           vi.stubEnv(key, value);
         }
         const event = logger?.createLogEvent(EventNames.API_ERROR, []);
-        expect(event?.event_metadata[0]).toEqual(
-          expect.arrayContaining([
-            {
-              gemini_cli_key: EventMetadataKey.GEMINI_CLI_SURFACE,
-              value: expectedValue,
-            },
-          ]),
-        );
+        expect(event?.event_metadata[0][3]).toEqual({
+          gemini_cli_key: EventMetadataKey.GEMINI_CLI_SURFACE,
+          value: expectedValue,
+        });
       },
     );
   });

packages/core/src/telemetry/clearcut-logger/clearcut-logger.ts

@@ -34,6 +34,7 @@ import type {
 import { EventMetadataKey } from './event-metadata-key.js';
 import type { Config } from '../../config/config.js';
 import { InstallationManager } from '../../utils/installationManager.js';
+import { UserAccountManager } from '../../utils/userAccountManager.js';
 import { safeJsonStringify } from '../../utils/safeJsonStringify.js';
 import { FixedDeque } from 'mnemonist';
 import { GIT_COMMIT_INFO, CLI_VERSION } from '../../generated/git-commit.js';
@@ -156,6 +157,7 @@ export class ClearcutLogger {
   private sessionData: EventValue[] = [];
   private promptId: string = '';
   private readonly installationManager: InstallationManager;
+  private readonly userAccountManager: UserAccountManager;
 
   /**
    * Queue of pending events that need to be flushed to the server.  New events
@@ -184,6 +186,7 @@ export class ClearcutLogger {
     this.events = new FixedDeque<LogEventEntry[]>(Array, MAX_EVENTS);
     this.promptId = config?.getSessionId() ?? '';
     this.installationManager = new InstallationManager();
+    this.userAccountManager = new UserAccountManager();
   }
 
   static getInstance(config?: Config): ClearcutLogger | undefined {
@@ -230,11 +233,14 @@ export class ClearcutLogger {
   }
 
   createLogEvent(eventName: EventNames, data: EventValue[] = []): LogEvent {
+    const email = this.userAccountManager.getCachedGoogleAccount();
+
     if (eventName !== EventNames.START_SESSION) {
       data.push(...this.sessionData);
     }
+    const totalAccounts = this.userAccountManager.getLifetimeGoogleAccounts();
 
-    data = this.addDefaultFields(data);
+    data = this.addDefaultFields(data, totalAccounts);
 
     const logEvent: LogEvent = {
       console_type: 'GEMINI_CLI',
@@ -243,7 +249,12 @@ export class ClearcutLogger {
       event_metadata: [data],
     };
 
-    logEvent.client_install_id = this.installationManager.getInstallationId();
+    // Should log either email or install ID, not both. See go/cloudmill-1p-oss-instrumentation#define-sessionable-id
+    if (email) {
+      logEvent.client_email = email;
+    } else {
+      logEvent.client_install_id = this.installationManager.getInstallationId();
+    }
 
     return logEvent;
   }
@@ -1007,7 +1018,7 @@ export class ClearcutLogger {
    * Adds default fields to data, and returns a new data array.  This fields
    * should exist on all log events.
    */
-  addDefaultFields(data: EventValue[]): EventValue[] {
+  addDefaultFields(data: EventValue[], totalAccounts: number): EventValue[] {
     const surface = determineSurface();
 
     const defaultLogMetadata: EventValue[] = [
@@ -1021,6 +1032,10 @@ export class ClearcutLogger {
           this.config?.getContentGeneratorConfig()?.authType,
         ),
       },
+      {
+        gemini_cli_key: EventMetadataKey.GEMINI_CLI_GOOGLE_ACCOUNTS_COUNT,
+        value: `${totalAccounts}`,
+      },
       {
         gemini_cli_key: EventMetadataKey.GEMINI_CLI_SURFACE,
         value: surface,

packages/core/src/telemetry/loggers.test.ts

@@ -83,6 +83,7 @@ import type {
 } from '@google/genai';
 import { DiscoveredMCPTool } from '../tools/mcp-tool.js';
 import * as uiTelemetry from './uiTelemetry.js';
+import { UserAccountManager } from '../utils/userAccountManager.js';
 import { makeFakeConfig } from '../test-utils/config.js';
 
 describe('loggers', () => {
@@ -100,6 +101,10 @@ describe('loggers', () => {
     vi.spyOn(uiTelemetry.uiTelemetryService, 'addEvent').mockImplementation(
       mockUiEvent.addEvent,
     );
+    vi.spyOn(
+      UserAccountManager.prototype,
+      'getCachedGoogleAccount',
+    ).mockReturnValue('test-user@example.com');
     vi.useFakeTimers();
     vi.setSystemTime(new Date('2025-01-01T00:00:00.000Z'));
   });
@@ -183,6 +188,7 @@ describe('loggers', () => {
         body: 'CLI configuration loaded.',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_CLI_CONFIG,
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           model: 'test-model',
@@ -227,6 +233,7 @@ describe('loggers', () => {
         body: 'User prompt. Length: 11.',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_USER_PROMPT,
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           prompt_length: 11,
@@ -248,7 +255,7 @@ describe('loggers', () => {
       const event = new UserPromptEvent(
         11,
         'prompt-id-9',
-        AuthType.USE_GEMINI,
+        AuthType.CLOUD_SHELL,
         'test-prompt',
       );
 
@@ -258,11 +265,12 @@ describe('loggers', () => {
         body: 'User prompt. Length: 11.',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_USER_PROMPT,
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           prompt_length: 11,
           prompt_id: 'prompt-id-9',
-          auth_type: 'gemini-api-key',
+          auth_type: 'cloud-shell',
         },
       });
     });
@@ -305,7 +313,7 @@ describe('loggers', () => {
         'test-model',
         100,
         'prompt-id-1',
-        AuthType.USE_GEMINI,
+        AuthType.LOGIN_WITH_GOOGLE,
         usageData,
         'test-response',
       );
@@ -316,6 +324,7 @@ describe('loggers', () => {
         body: 'API response from test-model. Status: 200. Duration: 100ms.',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_API_RESPONSE,
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           [SemanticAttributes.HTTP_STATUS_CODE]: 200,
@@ -331,7 +340,7 @@ describe('loggers', () => {
           total_token_count: 0,
           response_text: 'test-response',
           prompt_id: 'prompt-id-1',
-          auth_type: 'gemini-api-key',
+          auth_type: 'oauth-personal',
         },
       });
 
@@ -377,6 +386,7 @@ describe('loggers', () => {
         body: 'API request to test-model.',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_API_REQUEST,
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           model: 'test-model',
@@ -395,6 +405,7 @@ describe('loggers', () => {
         body: 'API request to test-model.',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_API_REQUEST,
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           model: 'test-model',
@@ -419,6 +430,7 @@ describe('loggers', () => {
         body: 'Switching to flash as Fallback.',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_FLASH_FALLBACK,
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           auth_type: 'vertex-ai',
@@ -453,6 +465,7 @@ describe('loggers', () => {
       expect(emittedEvent.attributes).toEqual(
         expect.objectContaining({
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_RIPGREP_FALLBACK,
           error: 'ripgrep is not available',
         }),
@@ -471,6 +484,7 @@ describe('loggers', () => {
       expect(emittedEvent.attributes).toEqual(
         expect.objectContaining({
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_RIPGREP_FALLBACK,
           error: 'rg not found',
         }),
@@ -584,6 +598,7 @@ describe('loggers', () => {
         body: 'Tool call: test-function. Decision: accept. Success: true. Duration: 100ms.',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_TOOL_CALL,
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           function_name: 'test-function',
@@ -667,6 +682,7 @@ describe('loggers', () => {
         body: 'Tool call: test-function. Decision: reject. Success: false. Duration: 100ms.',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_TOOL_CALL,
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           function_name: 'test-function',
@@ -743,6 +759,7 @@ describe('loggers', () => {
         body: 'Tool call: test-function. Decision: modify. Success: true. Duration: 100ms.',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_TOOL_CALL,
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           function_name: 'test-function',
@@ -818,6 +835,7 @@ describe('loggers', () => {
         body: 'Tool call: test-function. Success: true. Duration: 100ms.',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_TOOL_CALL,
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           function_name: 'test-function',
@@ -892,6 +910,7 @@ describe('loggers', () => {
         body: 'Tool call: test-function. Success: false. Duration: 100ms.',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_TOOL_CALL,
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           function_name: 'test-function',
@@ -980,6 +999,7 @@ describe('loggers', () => {
         body: 'Tool call: mock_mcp_tool. Success: true. Duration: 100ms.',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_TOOL_CALL,
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           function_name: 'mock_mcp_tool',
@@ -1027,6 +1047,7 @@ describe('loggers', () => {
         body: 'Malformed JSON response from test-model.',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_MALFORMED_JSON_RESPONSE,
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           model: 'test-model',
@@ -1070,6 +1091,7 @@ describe('loggers', () => {
         body: 'File operation: read. Lines: 10.',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_FILE_OPERATION,
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           tool_name: 'test-tool',
@@ -1115,6 +1137,7 @@ describe('loggers', () => {
         body: 'Tool output truncated for test-tool.',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': 'tool_output_truncated',
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           eventName: 'tool_output_truncated',
@@ -1161,6 +1184,7 @@ describe('loggers', () => {
         body: 'Installed extension vscode',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_EXTENSION_INSTALL,
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           extension_name: 'vscode',
@@ -1199,6 +1223,7 @@ describe('loggers', () => {
         body: 'Uninstalled extension vscode',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_EXTENSION_UNINSTALL,
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           extension_name: 'vscode',
@@ -1235,6 +1260,7 @@ describe('loggers', () => {
         body: 'Enabled extension vscode',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_EXTENSION_ENABLE,
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           extension_name: 'vscode',
@@ -1271,6 +1297,7 @@ describe('loggers', () => {
         body: 'Disabled extension vscode',
         attributes: {
           'session.id': 'test-session-id',
+          'user.email': 'test-user@example.com',
           'event.name': EVENT_EXTENSION_DISABLE,
           'event.timestamp': '2025-01-01T00:00:00.000Z',
           extension_name: 'vscode',

packages/core/src/telemetry/loggers.ts

@@ -9,6 +9,7 @@ import { logs } from '@opentelemetry/api-logs';
 import { SemanticAttributes } from '@opentelemetry/semantic-conventions';
 import type { Config } from '../config/config.js';
 import { safeJsonStringify } from '../utils/safeJsonStringify.js';
+import { UserAccountManager } from '../utils/userAccountManager.js';
 import {
   EVENT_API_ERROR,
   EVENT_API_CANCEL,
@@ -92,8 +93,11 @@ const shouldLogUserPrompts = (config: Config): boolean =>
   config.getTelemetryLogPromptsEnabled();
 
 function getCommonAttributes(config: Config): LogAttributes {
+  const userAccountManager = new UserAccountManager();
+  const email = userAccountManager.getCachedGoogleAccount();
   return {
     'session.id': config.getSessionId(),
+    ...(email && { 'user.email': email }),
   };
 }
 

packages/core/src/tools/mcp-client.test.ts

@@ -217,9 +217,9 @@ describe('mcp-client', () => {
           false,
         );
 
-        expect(transport).toBeInstanceOf(StreamableHTTPClientTransport);
-        // eslint-disable-next-line @typescript-eslint/no-explicit-any
-        expect((transport as any)._url).toEqual(new URL('http://test-server'));
+        expect(transport).toEqual(
+          new StreamableHTTPClientTransport(new URL('http://test-server'), {}),
+        );
       });
 
       it('with headers', async () => {
@@ -232,13 +232,13 @@ describe('mcp-client', () => {
           false,
         );
 
-        expect(transport).toBeInstanceOf(StreamableHTTPClientTransport);
-        // eslint-disable-next-line @typescript-eslint/no-explicit-any
-        expect((transport as any)._url).toEqual(new URL('http://test-server'));
-        // eslint-disable-next-line @typescript-eslint/no-explicit-any
-        expect((transport as any)._requestInit?.headers).toEqual({
-          Authorization: 'derp',
-        });
+        expect(transport).toEqual(
+          new StreamableHTTPClientTransport(new URL('http://test-server'), {
+            requestInit: {
+              headers: { Authorization: 'derp' },
+            },
+          }),
+        );
       });
     });
 
@@ -251,9 +251,9 @@ describe('mcp-client', () => {
           },
           false,
         );
-        expect(transport).toBeInstanceOf(SSEClientTransport);
-        // eslint-disable-next-line @typescript-eslint/no-explicit-any
-        expect((transport as any)._url).toEqual(new URL('http://test-server'));
+        expect(transport).toEqual(
+          new SSEClientTransport(new URL('http://test-server'), {}),
+        );
       });
 
       it('with headers', async () => {
@@ -266,13 +266,13 @@ describe('mcp-client', () => {
           false,
         );
 
-        expect(transport).toBeInstanceOf(SSEClientTransport);
-        // eslint-disable-next-line @typescript-eslint/no-explicit-any
-        expect((transport as any)._url).toEqual(new URL('http://test-server'));
-        // eslint-disable-next-line @typescript-eslint/no-explicit-any
-        expect((transport as any)._requestInit?.headers).toEqual({
-          Authorization: 'derp',
-        });
+        expect(transport).toEqual(
+          new SSEClientTransport(new URL('http://test-server'), {
+            requestInit: {
+              headers: { Authorization: 'derp' },
+            },
+          }),
+        );
       });
     });
 

packages/core/src/utils/errorParsing.test.ts

@@ -6,6 +6,9 @@
 
 import { describe, it, expect } from 'vitest';
 import { parseAndFormatApiError } from './errorParsing.js';
+import { isProQuotaExceededError } from './quotaErrorDetection.js';
+import { DEFAULT_GEMINI_FLASH_MODEL } from '../config/models.js';
+import { UserTierId } from '../code_assist/types.js';
 import { AuthType } from '../core/contentGenerator.js';
 import type { StructuredError } from '../core/turn.js';
 
@@ -24,10 +27,32 @@ describe('parseAndFormatApiError', () => {
   it('should format a 429 API error with the default message', () => {
     const errorMessage =
       'got status: 429 Too Many Requests. {"error":{"code":429,"message":"Rate limit exceeded","status":"RESOURCE_EXHAUSTED"}}';
-    const result = parseAndFormatApiError(errorMessage, undefined);
+    const result = parseAndFormatApiError(
+      errorMessage,
+      undefined,
+      undefined,
+      'gemini-2.5-pro',
+      DEFAULT_GEMINI_FLASH_MODEL,
+    );
     expect(result).toContain('[API Error: Rate limit exceeded');
     expect(result).toContain(
-      'Possible quota limitations in place or slow response times detected. Please wait and try again later.',
+      'Possible quota limitations in place or slow response times detected. Switching to the gemini-2.5-flash model',
+    );
+  });
+
+  it('should format a 429 API error with the personal message', () => {
+    const errorMessage =
+      'got status: 429 Too Many Requests. {"error":{"code":429,"message":"Rate limit exceeded","status":"RESOURCE_EXHAUSTED"}}';
+    const result = parseAndFormatApiError(
+      errorMessage,
+      AuthType.LOGIN_WITH_GOOGLE,
+      undefined,
+      'gemini-2.5-pro',
+      DEFAULT_GEMINI_FLASH_MODEL,
+    );
+    expect(result).toContain('[API Error: Rate limit exceeded');
+    expect(result).toContain(
+      'Possible quota limitations in place or slow response times detected. Switching to the gemini-2.5-flash model',
     );
   });
 
@@ -107,4 +132,230 @@ describe('parseAndFormatApiError', () => {
     const expected = '[API Error: An unknown error occurred.]';
     expect(parseAndFormatApiError(error)).toBe(expected);
   });
+
+  it('should format a 429 API error with Pro quota exceeded message for Google auth (Free tier)', () => {
+    const errorMessage =
+      'got status: 429 Too Many Requests. {"error":{"code":429,"message":"Quota exceeded for quota metric \'Gemini 2.5 Pro Requests\' and limit \'RequestsPerDay\' of service \'generativelanguage.googleapis.com\' for consumer \'project_number:123456789\'.","status":"RESOURCE_EXHAUSTED"}}';
+    const result = parseAndFormatApiError(
+      errorMessage,
+      AuthType.LOGIN_WITH_GOOGLE,
+      undefined,
+      'gemini-2.5-pro',
+      DEFAULT_GEMINI_FLASH_MODEL,
+    );
+    expect(result).toContain(
+      "[API Error: Quota exceeded for quota metric 'Gemini 2.5 Pro Requests'",
+    );
+    expect(result).toContain(
+      'You have reached your daily gemini-2.5-pro quota limit',
+    );
+    expect(result).toContain('upgrade to get higher limits');
+  });
+
+  it('should format a regular 429 API error with standard message for Google auth', () => {
+    const errorMessage =
+      'got status: 429 Too Many Requests. {"error":{"code":429,"message":"Rate limit exceeded","status":"RESOURCE_EXHAUSTED"}}';
+    const result = parseAndFormatApiError(
+      errorMessage,
+      AuthType.LOGIN_WITH_GOOGLE,
+      undefined,
+      'gemini-2.5-pro',
+      DEFAULT_GEMINI_FLASH_MODEL,
+    );
+    expect(result).toContain('[API Error: Rate limit exceeded');
+    expect(result).toContain(
+      'Possible quota limitations in place or slow response times detected. Switching to the gemini-2.5-flash model',
+    );
+    expect(result).not.toContain(
+      'You have reached your daily gemini-2.5-pro quota limit',
+    );
+  });
+
+  it('should format a 429 API error with generic quota exceeded message for Google auth', () => {
+    const errorMessage =
+      'got status: 429 Too Many Requests. {"error":{"code":429,"message":"Quota exceeded for quota metric \'GenerationRequests\' and limit \'RequestsPerDay\' of service \'generativelanguage.googleapis.com\' for consumer \'project_number:123456789\'.","status":"RESOURCE_EXHAUSTED"}}';
+    const result = parseAndFormatApiError(
+      errorMessage,
+      AuthType.LOGIN_WITH_GOOGLE,
+      undefined,
+      'gemini-2.5-pro',
+      DEFAULT_GEMINI_FLASH_MODEL,
+    );
+    expect(result).toContain(
+      "[API Error: Quota exceeded for quota metric 'GenerationRequests'",
+    );
+    expect(result).toContain('You have reached your daily quota limit');
+    expect(result).not.toContain(
+      'You have reached your daily Gemini 2.5 Pro quota limit',
+    );
+  });
+
+  it('should prioritize Pro quota message over generic quota message for Google auth', () => {
+    const errorMessage =
+      'got status: 429 Too Many Requests. {"error":{"code":429,"message":"Quota exceeded for quota metric \'Gemini 2.5 Pro Requests\' and limit \'RequestsPerDay\' of service \'generativelanguage.googleapis.com\' for consumer \'project_number:123456789\'.","status":"RESOURCE_EXHAUSTED"}}';
+    const result = parseAndFormatApiError(
+      errorMessage,
+      AuthType.LOGIN_WITH_GOOGLE,
+      undefined,
+      'gemini-2.5-pro',
+      DEFAULT_GEMINI_FLASH_MODEL,
+    );
+    expect(result).toContain(
+      "[API Error: Quota exceeded for quota metric 'Gemini 2.5 Pro Requests'",
+    );
+    expect(result).toContain(
+      'You have reached your daily gemini-2.5-pro quota limit',
+    );
+    expect(result).not.toContain('You have reached your daily quota limit');
+  });
+
+  it('should format a 429 API error with Pro quota exceeded message for Google auth (Standard tier)', () => {
+    const errorMessage =
+      'got status: 429 Too Many Requests. {"error":{"code":429,"message":"Quota exceeded for quota metric \'Gemini 2.5 Pro Requests\' and limit \'RequestsPerDay\' of service \'generativelanguage.googleapis.com\' for consumer \'project_number:123456789\'.","status":"RESOURCE_EXHAUSTED"}}';
+    const result = parseAndFormatApiError(
+      errorMessage,
+      AuthType.LOGIN_WITH_GOOGLE,
+      UserTierId.STANDARD,
+      'gemini-2.5-pro',
+      DEFAULT_GEMINI_FLASH_MODEL,
+    );
+    expect(result).toContain(
+      "[API Error: Quota exceeded for quota metric 'Gemini 2.5 Pro Requests'",
+    );
+    expect(result).toContain(
+      'You have reached your daily gemini-2.5-pro quota limit',
+    );
+    expect(result).toContain(
+      'We appreciate you for choosing Gemini Code Assist and the Gemini CLI',
+    );
+    expect(result).not.toContain('upgrade to get higher limits');
+  });
+
+  it('should format a 429 API error with Pro quota exceeded message for Google auth (Legacy tier)', () => {
+    const errorMessage =
+      'got status: 429 Too Many Requests. {"error":{"code":429,"message":"Quota exceeded for quota metric \'Gemini 2.5 Pro Requests\' and limit \'RequestsPerDay\' of service \'generativelanguage.googleapis.com\' for consumer \'project_number:123456789\'.","status":"RESOURCE_EXHAUSTED"}}';
+    const result = parseAndFormatApiError(
+      errorMessage,
+      AuthType.LOGIN_WITH_GOOGLE,
+      UserTierId.LEGACY,
+      'gemini-2.5-pro',
+      DEFAULT_GEMINI_FLASH_MODEL,
+    );
+    expect(result).toContain(
+      "[API Error: Quota exceeded for quota metric 'Gemini 2.5 Pro Requests'",
+    );
+    expect(result).toContain(
+      'You have reached your daily gemini-2.5-pro quota limit',
+    );
+    expect(result).toContain(
+      'We appreciate you for choosing Gemini Code Assist and the Gemini CLI',
+    );
+    expect(result).not.toContain('upgrade to get higher limits');
+  });
+
+  it('should handle different Gemini 2.5 version strings in Pro quota exceeded errors', () => {
+    const errorMessage25 =
+      'got status: 429 Too Many Requests. {"error":{"code":429,"message":"Quota exceeded for quota metric \'Gemini 2.5 Pro Requests\' and limit \'RequestsPerDay\' of service \'generativelanguage.googleapis.com\' for consumer \'project_number:123456789\'.","status":"RESOURCE_EXHAUSTED"}}';
+    const errorMessagePreview =
+      'got status: 429 Too Many Requests. {"error":{"code":429,"message":"Quota exceeded for quota metric \'Gemini 2.5-preview Pro Requests\' and limit \'RequestsPerDay\' of service \'generativelanguage.googleapis.com\' for consumer \'project_number:123456789\'.","status":"RESOURCE_EXHAUSTED"}}';
+
+    const result25 = parseAndFormatApiError(
+      errorMessage25,
+      AuthType.LOGIN_WITH_GOOGLE,
+      undefined,
+      'gemini-2.5-pro',
+      DEFAULT_GEMINI_FLASH_MODEL,
+    );
+    const resultPreview = parseAndFormatApiError(
+      errorMessagePreview,
+      AuthType.LOGIN_WITH_GOOGLE,
+      undefined,
+      'gemini-2.5-preview-pro',
+      DEFAULT_GEMINI_FLASH_MODEL,
+    );
+
+    expect(result25).toContain(
+      'You have reached your daily gemini-2.5-pro quota limit',
+    );
+    expect(resultPreview).toContain(
+      'You have reached your daily gemini-2.5-preview-pro quota limit',
+    );
+    expect(result25).toContain('upgrade to get higher limits');
+    expect(resultPreview).toContain('upgrade to get higher limits');
+  });
+
+  it('should not match non-Pro models with similar version strings', () => {
+    // Test that Flash models with similar version strings don't match
+    expect(
+      isProQuotaExceededError(
+        "Quota exceeded for quota metric 'Gemini 2.5 Flash Requests' and limit",
+      ),
+    ).toBe(false);
+    expect(
+      isProQuotaExceededError(
+        "Quota exceeded for quota metric 'Gemini 2.5-preview Flash Requests' and limit",
+      ),
+    ).toBe(false);
+
+    // Test other model types
+    expect(
+      isProQuotaExceededError(
+        "Quota exceeded for quota metric 'Gemini 2.5 Ultra Requests' and limit",
+      ),
+    ).toBe(false);
+    expect(
+      isProQuotaExceededError(
+        "Quota exceeded for quota metric 'Gemini 2.5 Standard Requests' and limit",
+      ),
+    ).toBe(false);
+
+    // Test generic quota messages
+    expect(
+      isProQuotaExceededError(
+        "Quota exceeded for quota metric 'GenerationRequests' and limit",
+      ),
+    ).toBe(false);
+    expect(
+      isProQuotaExceededError(
+        "Quota exceeded for quota metric 'EmbeddingRequests' and limit",
+      ),
+    ).toBe(false);
+  });
+
+  it('should format a generic quota exceeded message for Google auth (Standard tier)', () => {
+    const errorMessage =
+      'got status: 429 Too Many Requests. {"error":{"code":429,"message":"Quota exceeded for quota metric \'GenerationRequests\' and limit \'RequestsPerDay\' of service \'generativelanguage.googleapis.com\' for consumer \'project_number:123456789\'.","status":"RESOURCE_EXHAUSTED"}}';
+    const result = parseAndFormatApiError(
+      errorMessage,
+      AuthType.LOGIN_WITH_GOOGLE,
+      UserTierId.STANDARD,
+      'gemini-2.5-pro',
+      DEFAULT_GEMINI_FLASH_MODEL,
+    );
+    expect(result).toContain(
+      "[API Error: Quota exceeded for quota metric 'GenerationRequests'",
+    );
+    expect(result).toContain('You have reached your daily quota limit');
+    expect(result).toContain(
+      'We appreciate you for choosing Gemini Code Assist and the Gemini CLI',
+    );
+    expect(result).not.toContain('upgrade to get higher limits');
+  });
+
+  it('should format a regular 429 API error with standard message for Google auth (Standard tier)', () => {
+    const errorMessage =
+      'got status: 429 Too Many Requests. {"error":{"code":429,"message":"Rate limit exceeded","status":"RESOURCE_EXHAUSTED"}}';
+    const result = parseAndFormatApiError(
+      errorMessage,
+      AuthType.LOGIN_WITH_GOOGLE,
+      UserTierId.STANDARD,
+      'gemini-2.5-pro',
+      DEFAULT_GEMINI_FLASH_MODEL,
+    );
+    expect(result).toContain('[API Error: Rate limit exceeded');
+    expect(result).toContain(
+      'We appreciate you for choosing Gemini Code Assist and the Gemini CLI',
+    );
+    expect(result).not.toContain('upgrade to get higher limits');
+  });
 });

packages/core/src/utils/errorParsing.ts

@@ -4,36 +4,120 @@
  * SPDX-License-Identifier: Apache-2.0
  */
 
-import { isApiError, isStructuredError } from './quotaErrorDetection.js';
+import {
+  isProQuotaExceededError,
+  isGenericQuotaExceededError,
+  isApiError,
+  isStructuredError,
+} from './quotaErrorDetection.js';
+import {
+  DEFAULT_GEMINI_MODEL,
+  DEFAULT_GEMINI_FLASH_MODEL,
+} from '../config/models.js';
+import { UserTierId } from '../code_assist/types.js';
 import { AuthType } from '../core/contentGenerator.js';
 
 // Free Tier message functions
+const getRateLimitErrorMessageGoogleFree = (
+  fallbackModel: string = DEFAULT_GEMINI_FLASH_MODEL,
+) =>
+  `\nPossible quota limitations in place or slow response times detected. Switching to the ${fallbackModel} model for the rest of this session.`;
+
+const getRateLimitErrorMessageGoogleProQuotaFree = (
+  currentModel: string = DEFAULT_GEMINI_MODEL,
+  fallbackModel: string = DEFAULT_GEMINI_FLASH_MODEL,
+) =>
+  `\nYou have reached your daily ${currentModel} quota limit. You will be switched to the ${fallbackModel} model for the rest of this session. To increase your limits, upgrade to get higher limits at https://goo.gle/set-up-gemini-code-assist, or use /auth to switch to using a paid API key from AI Studio at https://aistudio.google.com/apikey`;
+
+const getRateLimitErrorMessageGoogleGenericQuotaFree = () =>
+  `\nYou have reached your daily quota limit. To increase your limits, upgrade to get higher limits at https://goo.gle/set-up-gemini-code-assist, or use /auth to switch to using a paid API key from AI Studio at https://aistudio.google.com/apikey`;
+
+// Legacy/Standard Tier message functions
+const getRateLimitErrorMessageGooglePaid = (
+  fallbackModel: string = DEFAULT_GEMINI_FLASH_MODEL,
+) =>
+  `\nPossible quota limitations in place or slow response times detected. Switching to the ${fallbackModel} model for the rest of this session. We appreciate you for choosing Gemini Code Assist and the Gemini CLI.`;
+
+const getRateLimitErrorMessageGoogleProQuotaPaid = (
+  currentModel: string = DEFAULT_GEMINI_MODEL,
+  fallbackModel: string = DEFAULT_GEMINI_FLASH_MODEL,
+) =>
+  `\nYou have reached your daily ${currentModel} quota limit. You will be switched to the ${fallbackModel} model for the rest of this session. We appreciate you for choosing Gemini Code Assist and the Gemini CLI. To continue accessing the ${currentModel} model today, consider using /auth to switch to using a paid API key from AI Studio at https://aistudio.google.com/apikey`;
+
+const getRateLimitErrorMessageGoogleGenericQuotaPaid = (
+  currentModel: string = DEFAULT_GEMINI_MODEL,
+) =>
+  `\nYou have reached your daily quota limit. We appreciate you for choosing Gemini Code Assist and the Gemini CLI. To continue accessing the ${currentModel} model today, consider using /auth to switch to using a paid API key from AI Studio at https://aistudio.google.com/apikey`;
 const RATE_LIMIT_ERROR_MESSAGE_USE_GEMINI =
   '\nPlease wait and try again later. To increase your limits, request a quota increase through AI Studio, or switch to another /auth method';
 const RATE_LIMIT_ERROR_MESSAGE_VERTEX =
   '\nPlease wait and try again later. To increase your limits, request a quota increase through Vertex, or switch to another /auth method';
-const RATE_LIMIT_ERROR_MESSAGE_DEFAULT =
-  '\nPossible quota limitations in place or slow response times detected. Please wait and try again later.';
+const getRateLimitErrorMessageDefault = (
+  fallbackModel: string = DEFAULT_GEMINI_FLASH_MODEL,
+) =>
+  `\nPossible quota limitations in place or slow response times detected. Switching to the ${fallbackModel} model for the rest of this session.`;
 
-function getRateLimitMessage(authType?: AuthType): string {
+function getRateLimitMessage(
+  authType?: AuthType,
+  error?: unknown,
+  userTier?: UserTierId,
+  currentModel?: string,
+  fallbackModel?: string,
+): string {
   switch (authType) {
+    case AuthType.LOGIN_WITH_GOOGLE: {
+      // Determine if user is on a paid tier (Legacy or Standard) - default to FREE if not specified
+      const isPaidTier =
+        userTier === UserTierId.LEGACY || userTier === UserTierId.STANDARD;
+
+      if (isProQuotaExceededError(error)) {
+        return isPaidTier
+          ? getRateLimitErrorMessageGoogleProQuotaPaid(
+              currentModel || DEFAULT_GEMINI_MODEL,
+              fallbackModel,
+            )
+          : getRateLimitErrorMessageGoogleProQuotaFree(
+              currentModel || DEFAULT_GEMINI_MODEL,
+              fallbackModel,
+            );
+      } else if (isGenericQuotaExceededError(error)) {
+        return isPaidTier
+          ? getRateLimitErrorMessageGoogleGenericQuotaPaid(
+              currentModel || DEFAULT_GEMINI_MODEL,
+            )
+          : getRateLimitErrorMessageGoogleGenericQuotaFree();
+      } else {
+        return isPaidTier
+          ? getRateLimitErrorMessageGooglePaid(fallbackModel)
+          : getRateLimitErrorMessageGoogleFree(fallbackModel);
+      }
+    }
     case AuthType.USE_GEMINI:
       return RATE_LIMIT_ERROR_MESSAGE_USE_GEMINI;
     case AuthType.USE_VERTEX_AI:
       return RATE_LIMIT_ERROR_MESSAGE_VERTEX;
     default:
-      return RATE_LIMIT_ERROR_MESSAGE_DEFAULT;
+      return getRateLimitErrorMessageDefault(fallbackModel);
   }
 }
 
 export function parseAndFormatApiError(
   error: unknown,
   authType?: AuthType,
+  userTier?: UserTierId,
+  currentModel?: string,
+  fallbackModel?: string,
 ): string {
   if (isStructuredError(error)) {
     let text = `[API Error: ${error.message}]`;
     if (error.status === 429) {
-      text += getRateLimitMessage(authType);
+      text += getRateLimitMessage(
+        authType,
+        error,
+        userTier,
+        currentModel,
+        fallbackModel,
+      );
     }
     return text;
   }
@@ -62,7 +146,13 @@ export function parseAndFormatApiError(
         }
         let text = `[API Error: ${finalMessage} (Status: ${parsedError.error.status})]`;
         if (parsedError.error.code === 429) {
-          text += getRateLimitMessage(authType);
+          text += getRateLimitMessage(
+            authType,
+            parsedError,
+            userTier,
+            currentModel,
+            fallbackModel,
+          );
         }
         return text;
       }

packages/core/src/utils/flashFallback.test.ts

@@ -11,9 +11,12 @@ import {
   setSimulate429,
   disableSimulationAfterFallback,
   shouldSimulate429,
+  createSimulated429Error,
   resetRequestCounter,
 } from './testUtils.js';
 import { DEFAULT_GEMINI_FLASH_MODEL } from '../config/models.js';
+import { retryWithBackoff } from './retry.js';
+import { AuthType } from '../core/contentGenerator.js';
 // Import the new types (Assuming this test file is in packages/core/src/utils/)
 import type { FallbackModelHandler } from '../fallback/types.js';
 
@@ -58,6 +61,84 @@ describe('Retry Utility Fallback Integration', () => {
     expect(result).toBe('retry');
   });
 
+  // This test validates the retry utility's logic for triggering the callback.
+  it('should trigger onPersistent429 after 2 consecutive 429 errors for OAuth users', async () => {
+    let fallbackCalled = false;
+    // Removed fallbackModel variable as it's no longer relevant here.
+
+    // Mock function that simulates exactly 2 429 errors, then succeeds after fallback
+    const mockApiCall = vi
+      .fn()
+      .mockRejectedValueOnce(createSimulated429Error())
+      .mockRejectedValueOnce(createSimulated429Error())
+      .mockResolvedValueOnce('success after fallback');
+
+    // Mock the onPersistent429 callback (this is what client.ts/geminiChat.ts provides)
+    const mockPersistent429Callback = vi.fn(async (_authType?: string) => {
+      fallbackCalled = true;
+      // Return true to signal retryWithBackoff to reset attempts and continue.
+      return true;
+    });
+
+    // Test with OAuth personal auth type, with maxAttempts = 2 to ensure fallback triggers
+    const result = await retryWithBackoff(mockApiCall, {
+      maxAttempts: 2,
+      initialDelayMs: 1,
+      maxDelayMs: 10,
+      shouldRetryOnError: (error: Error) => {
+        const status = (error as Error & { status?: number }).status;
+        return status === 429;
+      },
+      onPersistent429: mockPersistent429Callback,
+      authType: AuthType.LOGIN_WITH_GOOGLE,
+    });
+
+    // Verify fallback mechanism was triggered
+    expect(fallbackCalled).toBe(true);
+    expect(mockPersistent429Callback).toHaveBeenCalledWith(
+      AuthType.LOGIN_WITH_GOOGLE,
+      expect.any(Error),
+    );
+    expect(result).toBe('success after fallback');
+    // Should have: 2 failures, then fallback triggered, then 1 success after retry reset
+    expect(mockApiCall).toHaveBeenCalledTimes(3);
+  });
+
+  it('should not trigger onPersistent429 for API key users', async () => {
+    let fallbackCalled = false;
+
+    // Mock function that simulates 429 errors
+    const mockApiCall = vi.fn().mockRejectedValue(createSimulated429Error());
+
+    // Mock the callback
+    const mockPersistent429Callback = vi.fn(async () => {
+      fallbackCalled = true;
+      return true;
+    });
+
+    // Test with API key auth type - should not trigger fallback
+    try {
+      await retryWithBackoff(mockApiCall, {
+        maxAttempts: 5,
+        initialDelayMs: 10,
+        maxDelayMs: 100,
+        shouldRetryOnError: (error: Error) => {
+          const status = (error as Error & { status?: number }).status;
+          return status === 429;
+        },
+        onPersistent429: mockPersistent429Callback,
+        authType: AuthType.USE_GEMINI, // API key auth type
+      });
+    } catch (error) {
+      // Expected to throw after max attempts
+      expect((error as Error).message).toContain('Rate limit exceeded');
+    }
+
+    // Verify fallback was NOT triggered for API key users
+    expect(fallbackCalled).toBe(false);
+    expect(mockPersistent429Callback).not.toHaveBeenCalled();
+  });
+
   // This test validates the test utilities themselves.
   it('should properly disable simulation state after fallback (Test Utility)', () => {
     // Enable simulation

packages/core/src/utils/nextSpeakerChecker.test.ts

@@ -61,7 +61,6 @@ describe('checkNextSpeaker', () => {
         generateContentStream: vi.fn(),
         countTokens: vi.fn(),
         embedContent: vi.fn(),
-        useSummarizedThinking: vi.fn().mockReturnValue(false),
       } as ContentGenerator,
       {} as Config,
     );

packages/core/src/utils/partUtils.ts

@@ -81,7 +81,7 @@ export function getResponseText(
       candidate.content.parts.length > 0
     ) {
       return candidate.content.parts
-        .filter((part) => part.text && !part.thought)
+        .filter((part) => part.text)
         .map((part) => part.text)
         .join('');
     }

packages/core/src/utils/retry.test.ts

@@ -285,6 +285,173 @@ describe('retryWithBackoff', () => {
     });
   });
 
+  describe('Flash model fallback for OAuth users', () => {
+    it('should trigger fallback for OAuth personal users after persistent 429 errors', async () => {
+      const fallbackCallback = vi.fn().mockResolvedValue('gemini-2.5-flash');
+
+      let fallbackOccurred = false;
+      const mockFn = vi.fn().mockImplementation(async () => {
+        if (!fallbackOccurred) {
+          const error: HttpError = new Error('Rate limit exceeded');
+          error.status = 429;
+          throw error;
+        }
+        return 'success';
+      });
+
+      const promise = retryWithBackoff(mockFn, {
+        maxAttempts: 3,
+        initialDelayMs: 100,
+        onPersistent429: async (authType?: string) => {
+          fallbackOccurred = true;
+          return await fallbackCallback(authType);
+        },
+        authType: 'oauth-personal',
+      });
+
+      // Advance all timers to complete retries
+      await vi.runAllTimersAsync();
+
+      // Should succeed after fallback
+      await expect(promise).resolves.toBe('success');
+
+      // Verify callback was called with correct auth type
+      expect(fallbackCallback).toHaveBeenCalledWith('oauth-personal');
+
+      // Should retry again after fallback
+      expect(mockFn).toHaveBeenCalledTimes(3); // 2 initial attempts + 1 after fallback
+    });
+
+    it('should NOT trigger fallback for API key users', async () => {
+      const fallbackCallback = vi.fn();
+
+      const mockFn = vi.fn(async () => {
+        const error: HttpError = new Error('Rate limit exceeded');
+        error.status = 429;
+        throw error;
+      });
+
+      const promise = retryWithBackoff(mockFn, {
+        maxAttempts: 3,
+        initialDelayMs: 100,
+        onPersistent429: fallbackCallback,
+        authType: 'gemini-api-key',
+      });
+
+      // Handle the promise properly to avoid unhandled rejections
+      const resultPromise = promise.catch((error) => error);
+      await vi.runAllTimersAsync();
+      const result = await resultPromise;
+
+      // Should fail after all retries without fallback
+      expect(result).toBeInstanceOf(Error);
+      expect(result.message).toBe('Rate limit exceeded');
+
+      // Callback should not be called for API key users
+      expect(fallbackCallback).not.toHaveBeenCalled();
+    });
+
+    it('should reset attempt counter and continue after successful fallback', async () => {
+      let fallbackCalled = false;
+      const fallbackCallback = vi.fn().mockImplementation(async () => {
+        fallbackCalled = true;
+        return 'gemini-2.5-flash';
+      });
+
+      const mockFn = vi.fn().mockImplementation(async () => {
+        if (!fallbackCalled) {
+          const error: HttpError = new Error('Rate limit exceeded');
+          error.status = 429;
+          throw error;
+        }
+        return 'success';
+      });
+
+      const promise = retryWithBackoff(mockFn, {
+        maxAttempts: 3,
+        initialDelayMs: 100,
+        onPersistent429: fallbackCallback,
+        authType: 'oauth-personal',
+      });
+
+      await vi.runAllTimersAsync();
+
+      await expect(promise).resolves.toBe('success');
+      expect(fallbackCallback).toHaveBeenCalledOnce();
+    });
+
+    it('should continue with original error if fallback is rejected', async () => {
+      const fallbackCallback = vi.fn().mockResolvedValue(null); // User rejected fallback
+
+      const mockFn = vi.fn(async () => {
+        const error: HttpError = new Error('Rate limit exceeded');
+        error.status = 429;
+        throw error;
+      });
+
+      const promise = retryWithBackoff(mockFn, {
+        maxAttempts: 3,
+        initialDelayMs: 100,
+        onPersistent429: fallbackCallback,
+        authType: 'oauth-personal',
+      });
+
+      // Handle the promise properly to avoid unhandled rejections
+      const resultPromise = promise.catch((error) => error);
+      await vi.runAllTimersAsync();
+      const result = await resultPromise;
+
+      // Should fail with original error when fallback is rejected
+      expect(result).toBeInstanceOf(Error);
+      expect(result.message).toBe('Rate limit exceeded');
+      expect(fallbackCallback).toHaveBeenCalledWith(
+        'oauth-personal',
+        expect.any(Error),
+      );
+    });
+
+    it('should handle mixed error types (only count consecutive 429s)', async () => {
+      const fallbackCallback = vi.fn().mockResolvedValue('gemini-2.5-flash');
+      let attempts = 0;
+      let fallbackOccurred = false;
+
+      const mockFn = vi.fn().mockImplementation(async () => {
+        attempts++;
+        if (fallbackOccurred) {
+          return 'success';
+        }
+        if (attempts === 1) {
+          // First attempt: 500 error (resets consecutive count)
+          const error: HttpError = new Error('Server error');
+          error.status = 500;
+          throw error;
+        } else {
+          // Remaining attempts: 429 errors
+          const error: HttpError = new Error('Rate limit exceeded');
+          error.status = 429;
+          throw error;
+        }
+      });
+
+      const promise = retryWithBackoff(mockFn, {
+        maxAttempts: 5,
+        initialDelayMs: 100,
+        onPersistent429: async (authType?: string) => {
+          fallbackOccurred = true;
+          return await fallbackCallback(authType);
+        },
+        authType: 'oauth-personal',
+      });
+
+      await vi.runAllTimersAsync();
+
+      await expect(promise).resolves.toBe('success');
+
+      // Should trigger fallback after 2 consecutive 429s (attempts 2-3)
+      expect(fallbackCallback).toHaveBeenCalledWith('oauth-personal');
+    });
+  });
+
   describe('Qwen OAuth 429 error handling', () => {
     it('should retry for Qwen OAuth 429 errors that are throttling-related', async () => {
       const errorWith429: HttpError = new Error('Rate limit exceeded');