chore(release): v0.1.3-preview.0

package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "@qwen-code/qwen-code",
-  "version": "0.1.2",
+  "version": "0.1.3-preview.0",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "@qwen-code/qwen-code",
-      "version": "0.1.2",
+      "version": "0.1.3-preview.0",
       "workspaces": [
         "packages/*"
       ],
@@ -16024,7 +16024,7 @@
     },
     "packages/cli": {
       "name": "@qwen-code/qwen-code",
-      "version": "0.1.2",
+      "version": "0.1.3-preview.0",
       "dependencies": {
         "@google/genai": "1.16.0",
         "@iarna/toml": "^2.2.5",
@@ -16139,7 +16139,7 @@
     },
     "packages/core": {
       "name": "@qwen-code/qwen-code-core",
-      "version": "0.1.2",
+      "version": "0.1.3-preview.0",
       "hasInstallScript": true,
       "dependencies": {
         "@google/genai": "1.16.0",
@@ -16278,7 +16278,7 @@
     },
     "packages/test-utils": {
       "name": "@qwen-code/qwen-code-test-utils",
-      "version": "0.1.2",
+      "version": "0.1.3-preview.0",
       "dev": true,
       "license": "Apache-2.0",
       "devDependencies": {
@@ -16290,7 +16290,7 @@
     },
     "packages/vscode-ide-companion": {
       "name": "qwen-code-vscode-ide-companion",
-      "version": "0.1.2",
+      "version": "0.1.3-preview.0",
       "license": "LICENSE",
       "dependencies": {
         "@modelcontextprotocol/sdk": "^1.15.1",

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@qwen-code/qwen-code",
-  "version": "0.1.2",
+  "version": "0.1.3-preview.0",
   "engines": {
     "node": ">=20.0.0"
   },
@@ -13,7 +13,7 @@
     "url": "git+https://github.com/QwenLM/qwen-code.git"
   },
   "config": {
-    "sandboxImageUri": "ghcr.io/qwenlm/qwen-code:0.1.2"
+    "sandboxImageUri": "ghcr.io/qwenlm/qwen-code:0.1.3-preview.0"
   },
   "scripts": {
     "start": "cross-env node scripts/start.js",

packages/cli/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@qwen-code/qwen-code",
-  "version": "0.1.2",
+  "version": "0.1.3-preview.0",
   "description": "Qwen Code",
   "repository": {
     "type": "git",
@@ -25,7 +25,7 @@
     "dist"
   ],
   "config": {
-    "sandboxImageUri": "ghcr.io/qwenlm/qwen-code:0.1.2"
+    "sandboxImageUri": "ghcr.io/qwenlm/qwen-code:0.1.3-preview.0"
   },
   "dependencies": {
     "@google/genai": "1.16.0",

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

@@ -66,6 +66,8 @@ import {
   loadEnvironment,
   migrateDeprecatedSettings,
   SettingScope,
+  SETTINGS_VERSION,
+  SETTINGS_VERSION_KEY,
 } from './settings.js';
 import { FatalConfigError, QWEN_DIR } from '@qwen-code/qwen-code-core';
 
@@ -94,6 +96,7 @@ vi.mock('fs', async (importOriginal) => {
     existsSync: vi.fn(),
     readFileSync: vi.fn(),
     writeFileSync: vi.fn(),
+    renameSync: vi.fn(),
     mkdirSync: vi.fn(),
     realpathSync: (p: string) => p,
   };
@@ -171,11 +174,15 @@ describe('Settings Loading and Merging', () => {
         getSystemSettingsPath(),
         'utf-8',
       );
-      expect(settings.system.settings).toEqual(systemSettingsContent);
+      expect(settings.system.settings).toEqual({
+        ...systemSettingsContent,
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
+      });
       expect(settings.user.settings).toEqual({});
       expect(settings.workspace.settings).toEqual({});
       expect(settings.merged).toEqual({
         ...systemSettingsContent,
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
       });
     });
 
@@ -207,10 +214,14 @@ describe('Settings Loading and Merging', () => {
         expectedUserSettingsPath,
         'utf-8',
       );
-      expect(settings.user.settings).toEqual(userSettingsContent);
+      expect(settings.user.settings).toEqual({
+        ...userSettingsContent,
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
+      });
       expect(settings.workspace.settings).toEqual({});
       expect(settings.merged).toEqual({
         ...userSettingsContent,
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
       });
     });
 
@@ -241,9 +252,13 @@ describe('Settings Loading and Merging', () => {
         'utf-8',
       );
       expect(settings.user.settings).toEqual({});
-      expect(settings.workspace.settings).toEqual(workspaceSettingsContent);
+      expect(settings.workspace.settings).toEqual({
+        ...workspaceSettingsContent,
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
+      });
       expect(settings.merged).toEqual({
         ...workspaceSettingsContent,
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
       });
     });
 
@@ -304,10 +319,20 @@ describe('Settings Loading and Merging', () => {
 
       const settings = loadSettings(MOCK_WORKSPACE_DIR);
 
-      expect(settings.system.settings).toEqual(systemSettingsContent);
-      expect(settings.user.settings).toEqual(userSettingsContent);
-      expect(settings.workspace.settings).toEqual(workspaceSettingsContent);
+      expect(settings.system.settings).toEqual({
+        ...systemSettingsContent,
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
+      });
+      expect(settings.user.settings).toEqual({
+        ...userSettingsContent,
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
+      });
+      expect(settings.workspace.settings).toEqual({
+        ...workspaceSettingsContent,
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
+      });
       expect(settings.merged).toEqual({
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
         ui: {
           theme: 'system-theme',
         },
@@ -361,6 +386,7 @@ describe('Settings Loading and Merging', () => {
       const settings = loadSettings(MOCK_WORKSPACE_DIR);
 
       expect(settings.merged).toEqual({
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
         ui: {
           theme: 'legacy-dark',
         },
@@ -413,6 +439,132 @@ describe('Settings Loading and Merging', () => {
       expect((settings.merged as TestSettings)['allowedTools']).toBeUndefined();
     });
 
+    it('should add version field to migrated settings file', () => {
+      (mockFsExistsSync as Mock).mockImplementation(
+        (p: fs.PathLike) => p === USER_SETTINGS_PATH,
+      );
+      const legacySettingsContent = {
+        theme: 'dark',
+        model: 'qwen-coder',
+      };
+      (fs.readFileSync as Mock).mockImplementation(
+        (p: fs.PathOrFileDescriptor) => {
+          if (p === USER_SETTINGS_PATH)
+            return JSON.stringify(legacySettingsContent);
+          return '{}';
+        },
+      );
+
+      loadSettings(MOCK_WORKSPACE_DIR);
+
+      // Verify that fs.writeFileSync was called with migrated settings including version
+      expect(fs.writeFileSync).toHaveBeenCalled();
+      const writeCall = (fs.writeFileSync as Mock).mock.calls[0];
+      const writtenContent = JSON.parse(writeCall[1] as string);
+      expect(writtenContent[SETTINGS_VERSION_KEY]).toBe(SETTINGS_VERSION);
+    });
+
+    it('should not re-migrate settings that have version field', () => {
+      (mockFsExistsSync as Mock).mockImplementation(
+        (p: fs.PathLike) => p === USER_SETTINGS_PATH,
+      );
+      const migratedSettingsContent = {
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
+        ui: {
+          theme: 'dark',
+        },
+        model: {
+          name: 'qwen-coder',
+        },
+      };
+      (fs.readFileSync as Mock).mockImplementation(
+        (p: fs.PathOrFileDescriptor) => {
+          if (p === USER_SETTINGS_PATH)
+            return JSON.stringify(migratedSettingsContent);
+          return '{}';
+        },
+      );
+
+      loadSettings(MOCK_WORKSPACE_DIR);
+
+      // Verify that fs.renameSync and fs.writeFileSync were NOT called
+      // (because no migration was needed)
+      expect(fs.renameSync).not.toHaveBeenCalled();
+      expect(fs.writeFileSync).not.toHaveBeenCalled();
+    });
+
+    it('should add version field to V2 settings without version and write to disk', () => {
+      (mockFsExistsSync as Mock).mockImplementation(
+        (p: fs.PathLike) => p === USER_SETTINGS_PATH,
+      );
+      // V2 format but no version field
+      const v2SettingsWithoutVersion = {
+        ui: {
+          theme: 'dark',
+        },
+        model: {
+          name: 'qwen-coder',
+        },
+      };
+      (fs.readFileSync as Mock).mockImplementation(
+        (p: fs.PathOrFileDescriptor) => {
+          if (p === USER_SETTINGS_PATH)
+            return JSON.stringify(v2SettingsWithoutVersion);
+          return '{}';
+        },
+      );
+
+      loadSettings(MOCK_WORKSPACE_DIR);
+
+      // Verify that fs.writeFileSync was called (to add version)
+      // but NOT fs.renameSync (no backup needed, just adding version)
+      expect(fs.renameSync).not.toHaveBeenCalled();
+      expect(fs.writeFileSync).toHaveBeenCalledTimes(1);
+
+      const writeCall = (fs.writeFileSync as Mock).mock.calls[0];
+      const writtenPath = writeCall[0];
+      const writtenContent = JSON.parse(writeCall[1] as string);
+
+      expect(writtenPath).toBe(USER_SETTINGS_PATH);
+      expect(writtenContent[SETTINGS_VERSION_KEY]).toBe(SETTINGS_VERSION);
+      expect(writtenContent.ui?.theme).toBe('dark');
+      expect(writtenContent.model?.name).toBe('qwen-coder');
+    });
+
+    it('should correctly handle partially migrated settings without version field', () => {
+      (mockFsExistsSync as Mock).mockImplementation(
+        (p: fs.PathLike) => p === USER_SETTINGS_PATH,
+      );
+      // Edge case: model already in V2 format (object), but autoAccept in V1 format
+      const partiallyMigratedContent = {
+        model: {
+          name: 'qwen-coder',
+        },
+        autoAccept: false, // V1 key
+      };
+      (fs.readFileSync as Mock).mockImplementation(
+        (p: fs.PathOrFileDescriptor) => {
+          if (p === USER_SETTINGS_PATH)
+            return JSON.stringify(partiallyMigratedContent);
+          return '{}';
+        },
+      );
+
+      loadSettings(MOCK_WORKSPACE_DIR);
+
+      // Verify that the migrated settings preserve the model object correctly
+      expect(fs.writeFileSync).toHaveBeenCalled();
+      const writeCall = (fs.writeFileSync as Mock).mock.calls[0];
+      const writtenContent = JSON.parse(writeCall[1] as string);
+
+      // Model should remain as an object, not double-nested
+      expect(writtenContent.model).toEqual({ name: 'qwen-coder' });
+      // autoAccept should be migrated to tools.autoAccept
+      expect(writtenContent.tools?.autoAccept).toBe(false);
+      // Version field should be added
+      expect(writtenContent[SETTINGS_VERSION_KEY]).toBe(SETTINGS_VERSION);
+    });
+
     it('should correctly merge and migrate legacy array properties from multiple scopes', () => {
       (mockFsExistsSync as Mock).mockReturnValue(true);
       const legacyUserSettings = {
@@ -515,11 +667,24 @@ describe('Settings Loading and Merging', () => {
 
       const settings = loadSettings(MOCK_WORKSPACE_DIR);
 
-      expect(settings.systemDefaults.settings).toEqual(systemDefaultsContent);
-      expect(settings.system.settings).toEqual(systemSettingsContent);
-      expect(settings.user.settings).toEqual(userSettingsContent);
-      expect(settings.workspace.settings).toEqual(workspaceSettingsContent);
+      expect(settings.systemDefaults.settings).toEqual({
+        ...systemDefaultsContent,
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
+      });
+      expect(settings.system.settings).toEqual({
+        ...systemSettingsContent,
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
+      });
+      expect(settings.user.settings).toEqual({
+        ...userSettingsContent,
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
+      });
+      expect(settings.workspace.settings).toEqual({
+        ...workspaceSettingsContent,
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
+      });
       expect(settings.merged).toEqual({
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
         context: {
           fileName: 'WORKSPACE_CONTEXT.md',
           includeDirectories: [
@@ -866,8 +1031,14 @@ describe('Settings Loading and Merging', () => {
 
       const settings = loadSettings(MOCK_WORKSPACE_DIR);
 
-      expect(settings.user.settings).toEqual(userSettingsContent);
-      expect(settings.workspace.settings).toEqual(workspaceSettingsContent);
+      expect(settings.user.settings).toEqual({
+        ...userSettingsContent,
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
+      });
+      expect(settings.workspace.settings).toEqual({
+        ...workspaceSettingsContent,
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
+      });
       expect(settings.merged.mcpServers).toEqual({
         'user-server': {
           command: 'user-command',
@@ -1696,9 +1867,13 @@ describe('Settings Loading and Merging', () => {
           'utf-8',
         );
         expect(settings.system.path).toBe(MOCK_ENV_SYSTEM_SETTINGS_PATH);
-        expect(settings.system.settings).toEqual(systemSettingsContent);
+        expect(settings.system.settings).toEqual({
+          ...systemSettingsContent,
+          [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
+        });
         expect(settings.merged).toEqual({
           ...systemSettingsContent,
+          [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
         });
       });
     });
@@ -2248,6 +2423,44 @@ describe('Settings Loading and Merging', () => {
         customWittyPhrases: ['test phrase'],
       });
     });
+
+    it('should remove version field when migrating to V1', () => {
+      const v2Settings = {
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
+        ui: {
+          theme: 'dark',
+        },
+        model: {
+          name: 'qwen-coder',
+        },
+      };
+      const v1Settings = migrateSettingsToV1(v2Settings);
+
+      // Version field should not be present in V1 settings
+      expect(v1Settings[SETTINGS_VERSION_KEY]).toBeUndefined();
+      // Other fields should be properly migrated
+      expect(v1Settings).toEqual({
+        theme: 'dark',
+        model: 'qwen-coder',
+      });
+    });
+
+    it('should handle version field in unrecognized properties', () => {
+      const v2Settings = {
+        [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
+        general: {
+          vimMode: true,
+        },
+        someUnrecognizedKey: 'value',
+      };
+      const v1Settings = migrateSettingsToV1(v2Settings);
+
+      // Version field should be filtered out
+      expect(v1Settings[SETTINGS_VERSION_KEY]).toBeUndefined();
+      // Unrecognized keys should be preserved
+      expect(v1Settings['someUnrecognizedKey']).toBe('value');
+      expect(v1Settings['vimMode']).toBe(true);
+    });
   });
 
   describe('loadEnvironment', () => {
@@ -2368,6 +2581,73 @@ describe('Settings Loading and Merging', () => {
       };
       expect(needsMigration(settings)).toBe(false);
     });
+
+    describe('with version field', () => {
+      it('should return false when version field indicates current or newer version', () => {
+        const settingsWithVersion = {
+          [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
+          theme: 'dark', // Even though this is a V1 key, version field takes precedence
+        };
+        expect(needsMigration(settingsWithVersion)).toBe(false);
+      });
+
+      it('should return false when version field indicates a newer version', () => {
+        const settingsWithNewerVersion = {
+          [SETTINGS_VERSION_KEY]: SETTINGS_VERSION + 1,
+          theme: 'dark',
+        };
+        expect(needsMigration(settingsWithNewerVersion)).toBe(false);
+      });
+
+      it('should return true when version field indicates an older version', () => {
+        const settingsWithOldVersion = {
+          [SETTINGS_VERSION_KEY]: SETTINGS_VERSION - 1,
+          theme: 'dark',
+        };
+        expect(needsMigration(settingsWithOldVersion)).toBe(true);
+      });
+
+      it('should use fallback logic when version field is not a number', () => {
+        const settingsWithInvalidVersion = {
+          [SETTINGS_VERSION_KEY]: 'not-a-number',
+          theme: 'dark',
+        };
+        expect(needsMigration(settingsWithInvalidVersion)).toBe(true);
+      });
+
+      it('should use fallback logic when version field is missing', () => {
+        const settingsWithoutVersion = {
+          theme: 'dark',
+        };
+        expect(needsMigration(settingsWithoutVersion)).toBe(true);
+      });
+    });
+
+    describe('edge case: partially migrated settings', () => {
+      it('should return true for partially migrated settings without version field', () => {
+        // This simulates the dangerous edge case: model already in V2 format,
+        // but other fields in V1 format
+        const partiallyMigrated = {
+          model: {
+            name: 'qwen-coder',
+          },
+          autoAccept: false, // V1 key
+        };
+        expect(needsMigration(partiallyMigrated)).toBe(true);
+      });
+
+      it('should return false for partially migrated settings WITH version field', () => {
+        // With version field, we trust that it's been properly migrated
+        const partiallyMigratedWithVersion = {
+          [SETTINGS_VERSION_KEY]: SETTINGS_VERSION,
+          model: {
+            name: 'qwen-coder',
+          },
+          autoAccept: false, // This would look like V1 but version says it's V2
+        };
+        expect(needsMigration(partiallyMigratedWithVersion)).toBe(false);
+      });
+    });
   });
 
   describe('migrateDeprecatedSettings', () => {

packages/cli/src/config/settings.ts

@@ -56,6 +56,10 @@ export const DEFAULT_EXCLUDED_ENV_VARS = ['DEBUG', 'DEBUG_MODE'];
 
 const MIGRATE_V2_OVERWRITE = true;
 
+// Settings version to track migration state
+export const SETTINGS_VERSION = 2;
+export const SETTINGS_VERSION_KEY = '$version';
+
 const MIGRATION_MAP: Record<string, string> = {
   accessibility: 'ui.accessibility',
   allowedTools: 'tools.allowed',
@@ -216,8 +220,16 @@ function setNestedProperty(
 }
 
 export function needsMigration(settings: Record<string, unknown>): boolean {
-  // A file needs migration if it contains any top-level key that is moved to a
-  // nested location in V2.
+  // Check version field first - if present and matches current version, no migration needed
+  if (SETTINGS_VERSION_KEY in settings) {
+    const version = settings[SETTINGS_VERSION_KEY];
+    if (typeof version === 'number' && version >= SETTINGS_VERSION) {
+      return false;
+    }
+  }
+
+  // Fallback to legacy detection: A file needs migration if it contains any
+  // top-level key that is moved to a nested location in V2.
   const hasV1Keys = Object.entries(MIGRATION_MAP).some(([v1Key, v2Path]) => {
     if (v1Key === v2Path || !(v1Key in settings)) {
       return false;
@@ -250,6 +262,21 @@ function migrateSettingsToV2(
 
   for (const [oldKey, newPath] of Object.entries(MIGRATION_MAP)) {
     if (flatKeys.has(oldKey)) {
+      // Safety check: If this key is a V2 container (like 'model') and it's
+      // already an object, it's likely already in V2 format. Skip migration
+      // to prevent double-nesting (e.g., model.name.name).
+      if (
+        KNOWN_V2_CONTAINERS.has(oldKey) &&
+        typeof flatSettings[oldKey] === 'object' &&
+        flatSettings[oldKey] !== null &&
+        !Array.isArray(flatSettings[oldKey])
+      ) {
+        // This is already a V2 container, carry it over as-is
+        v2Settings[oldKey] = flatSettings[oldKey];
+        flatKeys.delete(oldKey);
+        continue;
+      }
+
       setNestedProperty(v2Settings, newPath, flatSettings[oldKey]);
       flatKeys.delete(oldKey);
     }
@@ -287,6 +314,9 @@ function migrateSettingsToV2(
     }
   }
 
+  // Set version field to indicate this is a V2 settings file
+  v2Settings[SETTINGS_VERSION_KEY] = SETTINGS_VERSION;
+
   return v2Settings;
 }
 
@@ -336,6 +366,11 @@ export function migrateSettingsToV1(
 
   // Carry over any unrecognized keys
   for (const remainingKey of v2Keys) {
+    // Skip the version field - it's only for V2 format
+    if (remainingKey === SETTINGS_VERSION_KEY) {
+      continue;
+    }
+
     const value = v2Settings[remainingKey];
     if (value === undefined) {
       continue;
@@ -621,6 +656,22 @@ export function loadSettings(
             }
             settingsObject = migratedSettings;
           }
+        } else if (!(SETTINGS_VERSION_KEY in settingsObject)) {
+          // No migration needed, but version field is missing - add it for future optimizations
+          settingsObject[SETTINGS_VERSION_KEY] = SETTINGS_VERSION;
+          if (MIGRATE_V2_OVERWRITE) {
+            try {
+              fs.writeFileSync(
+                filePath,
+                JSON.stringify(settingsObject, null, 2),
+                'utf-8',
+              );
+            } catch (e) {
+              console.error(
+                `Error adding version to settings file: ${getErrorMessage(e)}`,
+              );
+            }
+          }
         }
         return { settings: settingsObject as Settings, rawJson: content };
       }

packages/cli/src/ui/AppContainer.tsx

@@ -916,17 +916,9 @@ export const AppContainer = (props: AppContainerProps) => {
     (result: IdeIntegrationNudgeResult) => {
       if (result.userSelection === 'yes') {
         handleSlashCommand('/ide install');
-        settings.setValue(
-          SettingScope.User,
-          'hasSeenIdeIntegrationNudge',
-          true,
-        );
+        settings.setValue(SettingScope.User, 'ide.hasSeenNudge', true);
       } else if (result.userSelection === 'dismiss') {
-        settings.setValue(
-          SettingScope.User,
-          'hasSeenIdeIntegrationNudge',
-          true,
-        );
+        settings.setValue(SettingScope.User, 'ide.hasSeenNudge', true);
       }
       setIdePromptAnswered(true);
     },

packages/core/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@qwen-code/qwen-code-core",
-  "version": "0.1.2",
+  "version": "0.1.3-preview.0",
   "description": "Qwen Code Core",
   "repository": {
     "type": "git",

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

@@ -16,11 +16,11 @@ import {
 
 import type { Content, GenerateContentResponse, Part } from '@google/genai';
 import {
-  findCompressSplitPoint,
   isThinkingDefault,
   isThinkingSupported,
   GeminiClient,
 } from './client.js';
+import { findCompressSplitPoint } from '../services/chatCompressionService.js';
 import {
   AuthType,
   type ContentGenerator,
@@ -42,7 +42,6 @@ import { setSimulate429 } from '../utils/testUtils.js';
 import { tokenLimit } from './tokenLimits.js';
 import { ideContextStore } from '../ide/ideContext.js';
 import { uiTelemetryService } from '../telemetry/uiTelemetry.js';
-import { QwenLogger } from '../telemetry/index.js';
 
 // Mock fs module to prevent actual file system operations during tests
 const mockFileSystem = new Map<string, string>();
@@ -101,6 +100,22 @@ vi.mock('../utils/errorReporting', () => ({ reportError: vi.fn() }));
 vi.mock('../utils/nextSpeakerChecker', () => ({
   checkNextSpeaker: vi.fn().mockResolvedValue(null),
 }));
+vi.mock('../utils/environmentContext', () => ({
+  getEnvironmentContext: vi
+    .fn()
+    .mockResolvedValue([{ text: 'Mocked env context' }]),
+  getInitialChatHistory: vi.fn(async (_config, extraHistory) => [
+    {
+      role: 'user',
+      parts: [{ text: 'Mocked env context' }],
+    },
+    {
+      role: 'model',
+      parts: [{ text: 'Got it. Thanks for the context!' }],
+    },
+    ...(extraHistory ?? []),
+  ]),
+}));
 vi.mock('../utils/generateContentResponseUtilities', () => ({
   getResponseText: (result: GenerateContentResponse) =>
     result.candidates?.[0]?.content?.parts?.map((part) => part.text).join('') ||
@@ -136,6 +151,10 @@ vi.mock('../ide/ideContext.js');
 vi.mock('../telemetry/uiTelemetry.js', () => ({
   uiTelemetryService: mockUiTelemetryService,
 }));
+vi.mock('../telemetry/loggers.js', () => ({
+  logChatCompression: vi.fn(),
+  logNextSpeakerCheck: vi.fn(),
+}));
 
 /**
  * Array.fromAsync ponyfill, which will be available in es 2024.
@@ -619,7 +638,8 @@ describe('Gemini Client (client.ts)', () => {
     });
 
     it('logs a telemetry event when compressing', async () => {
-      vi.spyOn(QwenLogger.prototype, 'logChatCompressionEvent');
+      const { logChatCompression } = await import('../telemetry/loggers.js');
+      vi.mocked(logChatCompression).mockClear();
 
       const MOCKED_TOKEN_LIMIT = 1000;
       const MOCKED_CONTEXT_PERCENTAGE_THRESHOLD = 0.5;
@@ -627,19 +647,37 @@ describe('Gemini Client (client.ts)', () => {
       vi.spyOn(client['config'], 'getChatCompression').mockReturnValue({
         contextPercentageThreshold: MOCKED_CONTEXT_PERCENTAGE_THRESHOLD,
       });
-      const history = [{ role: 'user', parts: [{ text: '...history...' }] }];
+      // Need multiple history items so there's something to compress
+      const history = [
+        { role: 'user', parts: [{ text: '...history 1...' }] },
+        { role: 'model', parts: [{ text: '...history 2...' }] },
+        { role: 'user', parts: [{ text: '...history 3...' }] },
+        { role: 'model', parts: [{ text: '...history 4...' }] },
+      ];
       mockGetHistory.mockReturnValue(history);
 
+      // Token count needs to be ABOVE the threshold to trigger compression
       const originalTokenCount =
-        MOCKED_TOKEN_LIMIT * MOCKED_CONTEXT_PERCENTAGE_THRESHOLD;
+        MOCKED_TOKEN_LIMIT * MOCKED_CONTEXT_PERCENTAGE_THRESHOLD + 1;
 
       vi.mocked(uiTelemetryService.getLastPromptTokenCount).mockReturnValue(
         originalTokenCount,
       );
 
-      // We need to control the estimated new token count.
-      // We mock startChat to return a chat with a known history.
+      // Mock the summary response from the chat
       const summaryText = 'This is a summary.';
+      mockGenerateContentFn.mockResolvedValue({
+        candidates: [
+          {
+            content: {
+              role: 'model',
+              parts: [{ text: summaryText }],
+            },
+          },
+        ],
+      } as unknown as GenerateContentResponse);
+
+      // Mock startChat to complete the compression flow
       const splitPoint = findCompressSplitPoint(history, 0.7);
       const historyToKeep = history.slice(splitPoint);
       const newCompressedHistory: Content[] = [
@@ -659,52 +697,36 @@ describe('Gemini Client (client.ts)', () => {
         .fn()
         .mockResolvedValue(mockNewChat as GeminiChat);
 
-      const totalChars = newCompressedHistory.reduce(
-        (total, content) => total + JSON.stringify(content).length,
-        0,
-      );
-      const newTokenCount = Math.floor(totalChars / 4);
-
-      // Mock the summary response from the chat
-      mockGenerateContentFn.mockResolvedValue({
-        candidates: [
-          {
-            content: {
-              role: 'model',
-              parts: [{ text: summaryText }],
-            },
-          },
-        ],
-      } as unknown as GenerateContentResponse);
-
       await client.tryCompressChat('prompt-id-3', false);
 
-      expect(QwenLogger.prototype.logChatCompressionEvent).toHaveBeenCalledWith(
+      expect(logChatCompression).toHaveBeenCalledWith(
+        expect.anything(),
         expect.objectContaining({
           tokens_before: originalTokenCount,
-          tokens_after: newTokenCount,
         }),
       );
-      expect(uiTelemetryService.setLastPromptTokenCount).toHaveBeenCalledWith(
-        newTokenCount,
-      );
-      expect(uiTelemetryService.setLastPromptTokenCount).toHaveBeenCalledTimes(
-        1,
-      );
+      expect(uiTelemetryService.setLastPromptTokenCount).toHaveBeenCalled();
     });
 
-    it('should trigger summarization if token count is at threshold with contextPercentageThreshold setting', async () => {
+    it('should trigger summarization if token count is above threshold with contextPercentageThreshold setting', async () => {
       const MOCKED_TOKEN_LIMIT = 1000;
       const MOCKED_CONTEXT_PERCENTAGE_THRESHOLD = 0.5;
       vi.mocked(tokenLimit).mockReturnValue(MOCKED_TOKEN_LIMIT);
       vi.spyOn(client['config'], 'getChatCompression').mockReturnValue({
         contextPercentageThreshold: MOCKED_CONTEXT_PERCENTAGE_THRESHOLD,
       });
-      const history = [{ role: 'user', parts: [{ text: '...history...' }] }];
+      // Need multiple history items so there's something to compress
+      const history = [
+        { role: 'user', parts: [{ text: '...history 1...' }] },
+        { role: 'model', parts: [{ text: '...history 2...' }] },
+        { role: 'user', parts: [{ text: '...history 3...' }] },
+        { role: 'model', parts: [{ text: '...history 4...' }] },
+      ];
       mockGetHistory.mockReturnValue(history);
 
+      // Token count needs to be ABOVE the threshold to trigger compression
       const originalTokenCount =
-        MOCKED_TOKEN_LIMIT * MOCKED_CONTEXT_PERCENTAGE_THRESHOLD;
+        MOCKED_TOKEN_LIMIT * MOCKED_CONTEXT_PERCENTAGE_THRESHOLD + 1;
 
       vi.mocked(uiTelemetryService.getLastPromptTokenCount).mockReturnValue(
         originalTokenCount,
@@ -864,7 +886,13 @@ describe('Gemini Client (client.ts)', () => {
     });
 
     it('should always trigger summarization when force is true, regardless of token count', async () => {
-      const history = [{ role: 'user', parts: [{ text: '...history...' }] }];
+      // Need multiple history items so there's something to compress
+      const history = [
+        { role: 'user', parts: [{ text: '...history 1...' }] },
+        { role: 'model', parts: [{ text: '...history 2...' }] },
+        { role: 'user', parts: [{ text: '...history 3...' }] },
+        { role: 'model', parts: [{ text: '...history 4...' }] },
+      ];
       mockGetHistory.mockReturnValue(history);
 
       const originalTokenCount = 100; // Well below threshold, but > estimated new count

packages/core/src/core/client.ts

@@ -25,13 +25,11 @@ import {
 import type { ContentGenerator } from './contentGenerator.js';
 import { GeminiChat } from './geminiChat.js';
 import {
-  getCompressionPrompt,
   getCoreSystemPrompt,
   getCustomSystemPrompt,
   getPlanModeSystemReminder,
   getSubagentSystemReminder,
 } from './prompts.js';
-import { tokenLimit } from './tokenLimits.js';
 import {
   CompressionStatus,
   GeminiEventType,
@@ -42,6 +40,11 @@ import {
 
 // Services
 import { type ChatRecordingService } from '../services/chatRecordingService.js';
+import {
+  ChatCompressionService,
+  COMPRESSION_PRESERVE_THRESHOLD,
+  COMPRESSION_TOKEN_THRESHOLD,
+} from '../services/chatCompressionService.js';
 import { LoopDetectionService } from '../services/loopDetectionService.js';
 
 // Tools
@@ -50,21 +53,18 @@ import { TaskTool } from '../tools/task.js';
 // Telemetry
 import {
   NextSpeakerCheckEvent,
-  logChatCompression,
   logNextSpeakerCheck,
-  makeChatCompressionEvent,
-  uiTelemetryService,
 } from '../telemetry/index.js';
 
 // Utilities
 import {
   getDirectoryContextString,
-  getEnvironmentContext,
+  getInitialChatHistory,
 } from '../utils/environmentContext.js';
 import { reportError } from '../utils/errorReporting.js';
 import { getErrorMessage } from '../utils/errors.js';
 import { checkNextSpeaker } from '../utils/nextSpeakerChecker.js';
-import { flatMapTextParts, getResponseText } from '../utils/partUtils.js';
+import { flatMapTextParts } from '../utils/partUtils.js';
 import { retryWithBackoff } from '../utils/retry.js';
 
 // IDE integration
@@ -85,68 +85,8 @@ export function isThinkingDefault(model: string) {
   return model.startsWith('gemini-2.5') || model === DEFAULT_GEMINI_MODEL_AUTO;
 }
 
-/**
- * Returns the index of the oldest item to keep when compressing. May return
- * contents.length which indicates that everything should be compressed.
- *
- * Exported for testing purposes.
- */
-export function findCompressSplitPoint(
-  contents: Content[],
-  fraction: number,
-): number {
-  if (fraction <= 0 || fraction >= 1) {
-    throw new Error('Fraction must be between 0 and 1');
-  }
-
-  const charCounts = contents.map((content) => JSON.stringify(content).length);
-  const totalCharCount = charCounts.reduce((a, b) => a + b, 0);
-  const targetCharCount = totalCharCount * fraction;
-
-  let lastSplitPoint = 0; // 0 is always valid (compress nothing)
-  let cumulativeCharCount = 0;
-  for (let i = 0; i < contents.length; i++) {
-    const content = contents[i];
-    if (
-      content.role === 'user' &&
-      !content.parts?.some((part) => !!part.functionResponse)
-    ) {
-      if (cumulativeCharCount >= targetCharCount) {
-        return i;
-      }
-      lastSplitPoint = i;
-    }
-    cumulativeCharCount += charCounts[i];
-  }
-
-  // We found no split points after targetCharCount.
-  // Check if it's safe to compress everything.
-  const lastContent = contents[contents.length - 1];
-  if (
-    lastContent?.role === 'model' &&
-    !lastContent?.parts?.some((part) => part.functionCall)
-  ) {
-    return contents.length;
-  }
-
-  // Can't compress everything so just compress at last splitpoint.
-  return lastSplitPoint;
-}
-
 const MAX_TURNS = 100;
 
-/**
- * Threshold for compression token count as a fraction of the model's token limit.
- * If the chat history exceeds this threshold, it will be compressed.
- */
-const COMPRESSION_TOKEN_THRESHOLD = 0.7;
-
-/**
- * The fraction of the latest chat history to keep. A value of 0.3
- * means that only the last 30% of the chat history will be kept after compression.
- */
-const COMPRESSION_PRESERVE_THRESHOLD = 0.3;
-
 export class GeminiClient {
   private chat?: GeminiChat;
   private readonly generateContentConfig: GenerateContentConfig = {
@@ -243,23 +183,13 @@ export class GeminiClient {
   async startChat(extraHistory?: Content[]): Promise<GeminiChat> {
     this.forceFullIdeContext = true;
     this.hasFailedCompressionAttempt = false;
-    const envParts = await getEnvironmentContext(this.config);
 
     const toolRegistry = this.config.getToolRegistry();
     const toolDeclarations = toolRegistry.getFunctionDeclarations();
     const tools: Tool[] = [{ functionDeclarations: toolDeclarations }];
 
-    const history: Content[] = [
-      {
-        role: 'user',
-        parts: envParts,
-      },
-      {
-        role: 'model',
-        parts: [{ text: 'Got it. Thanks for the context!' }],
-      },
-      ...(extraHistory ?? []),
-    ];
+    const history = await getInitialChatHistory(this.config, extraHistory);
+
     try {
       const userMemory = this.config.getUserMemory();
       const model = this.config.getModel();
@@ -503,14 +433,15 @@ export class GeminiClient {
         userMemory,
         this.config.getModel(),
       );
-      const environment = await getEnvironmentContext(this.config);
+      const initialHistory = await getInitialChatHistory(this.config);
 
       // Create a mock request content to count total tokens
       const mockRequestContent = [
         {
           role: 'system' as const,
-          parts: [{ text: systemPrompt }, ...environment],
+          parts: [{ text: systemPrompt }],
         },
+        ...initialHistory,
         ...currentHistory,
       ];
 
@@ -732,127 +663,37 @@ export class GeminiClient {
     prompt_id: string,
     force: boolean = false,
   ): Promise<ChatCompressionInfo> {
-    const model = this.config.getModel();
+    const compressionService = new ChatCompressionService();
 
-    const curatedHistory = this.getChat().getHistory(true);
+    const { newHistory, info } = await compressionService.compress(
+      this.getChat(),
+      prompt_id,
+      force,
+      this.config.getModel(),
+      this.config,
+      this.hasFailedCompressionAttempt,
+    );
 
-    // Regardless of `force`, don't do anything if the history is empty.
-    if (
-      curatedHistory.length === 0 ||
-      (this.hasFailedCompressionAttempt && !force)
+    // Handle compression result
+    if (info.compressionStatus === CompressionStatus.COMPRESSED) {
+      // Success: update chat with new compressed history
+      if (newHistory) {
+        this.chat = await this.startChat(newHistory);
+        this.forceFullIdeContext = true;
+      }
+    } else if (
+      info.compressionStatus ===
+        CompressionStatus.COMPRESSION_FAILED_INFLATED_TOKEN_COUNT ||
+      info.compressionStatus ===
+        CompressionStatus.COMPRESSION_FAILED_EMPTY_SUMMARY
     ) {
-      return {
-        originalTokenCount: 0,
-        newTokenCount: 0,
-        compressionStatus: CompressionStatus.NOOP,
-      };
-    }
-
-    const originalTokenCount = uiTelemetryService.getLastPromptTokenCount();
-
-    const contextPercentageThreshold =
-      this.config.getChatCompression()?.contextPercentageThreshold;
-
-    // Don't compress if not forced and we are under the limit.
-    if (!force) {
-      const threshold =
-        contextPercentageThreshold ?? COMPRESSION_TOKEN_THRESHOLD;
-      if (originalTokenCount < threshold * tokenLimit(model)) {
-        return {
-          originalTokenCount,
-          newTokenCount: originalTokenCount,
-          compressionStatus: CompressionStatus.NOOP,
-        };
+      // Track failed attempts (only mark as failed if not forced)
+      if (!force) {
+        this.hasFailedCompressionAttempt = true;
       }
     }
 
-    const splitPoint = findCompressSplitPoint(
-      curatedHistory,
-      1 - COMPRESSION_PRESERVE_THRESHOLD,
-    );
-
-    const historyToCompress = curatedHistory.slice(0, splitPoint);
-    const historyToKeep = curatedHistory.slice(splitPoint);
-
-    const summaryResponse = await this.config
-      .getContentGenerator()
-      .generateContent(
-        {
-          model,
-          contents: [
-            ...historyToCompress,
-            {
-              role: 'user',
-              parts: [
-                {
-                  text: 'First, reason in your scratchpad. Then, generate the <state_snapshot>.',
-                },
-              ],
-            },
-          ],
-          config: {
-            systemInstruction: { text: getCompressionPrompt() },
-          },
-        },
-        prompt_id,
-      );
-    const summary = getResponseText(summaryResponse) ?? '';
-
-    const chat = await this.startChat([
-      {
-        role: 'user',
-        parts: [{ text: summary }],
-      },
-      {
-        role: 'model',
-        parts: [{ text: 'Got it. Thanks for the additional context!' }],
-      },
-      ...historyToKeep,
-    ]);
-    this.forceFullIdeContext = true;
-
-    // Estimate token count 1 token ≈ 4 characters
-    const newTokenCount = Math.floor(
-      chat
-        .getHistory()
-        .reduce((total, content) => total + JSON.stringify(content).length, 0) /
-        4,
-    );
-
-    logChatCompression(
-      this.config,
-      makeChatCompressionEvent({
-        tokens_before: originalTokenCount,
-        tokens_after: newTokenCount,
-      }),
-    );
-
-    if (newTokenCount > originalTokenCount) {
-      this.hasFailedCompressionAttempt = !force && true;
-      return {
-        originalTokenCount,
-        newTokenCount,
-        compressionStatus:
-          CompressionStatus.COMPRESSION_FAILED_INFLATED_TOKEN_COUNT,
-      };
-    } else {
-      this.chat = chat; // Chat compression successful, set new state.
-      uiTelemetryService.setLastPromptTokenCount(newTokenCount);
-    }
-
-    logChatCompression(
-      this.config,
-      makeChatCompressionEvent({
-        tokens_before: originalTokenCount,
-        tokens_after: newTokenCount,
-      }),
-    );
-
-    return {
-      originalTokenCount,
-      newTokenCount,
-      compressionStatus: CompressionStatus.COMPRESSED,
-    };
+    return info;
   }
 }
 

packages/core/src/core/turn.ts

@@ -153,6 +153,9 @@ export enum CompressionStatus {
   /** The compression failed due to an error counting tokens */
   COMPRESSION_FAILED_TOKEN_COUNT_ERROR,
 
+  /** The compression failed due to receiving an empty or null summary */
+  COMPRESSION_FAILED_EMPTY_SUMMARY,
+
   /** The compression was not necessary and no action was taken */
   NOOP,
 }

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

@@ -113,7 +113,7 @@ describe('IdeClient', () => {
         'utf8',
       );
       expect(StreamableHTTPClientTransport).toHaveBeenCalledWith(
-        new URL('http://localhost:8080/mcp'),
+        new URL('http://127.0.0.1:8080/mcp'),
         expect.any(Object),
       );
       expect(mockClient.connect).toHaveBeenCalledWith(mockHttpTransport);
@@ -181,7 +181,7 @@ describe('IdeClient', () => {
       await ideClient.connect();
 
       expect(StreamableHTTPClientTransport).toHaveBeenCalledWith(
-        new URL('http://localhost:9090/mcp'),
+        new URL('http://127.0.0.1:9090/mcp'),
         expect.any(Object),
       );
       expect(mockClient.connect).toHaveBeenCalledWith(mockHttpTransport);
@@ -230,7 +230,7 @@ describe('IdeClient', () => {
       await ideClient.connect();
 
       expect(StreamableHTTPClientTransport).toHaveBeenCalledWith(
-        new URL('http://localhost:8080/mcp'),
+        new URL('http://127.0.0.1:8080/mcp'),
         expect.any(Object),
       );
       expect(ideClient.getConnectionStatus().status).toBe(
@@ -665,7 +665,7 @@ describe('IdeClient', () => {
       await ideClient.connect();
 
       expect(StreamableHTTPClientTransport).toHaveBeenCalledWith(
-        new URL('http://localhost:8080/mcp'),
+        new URL('http://127.0.0.1:8080/mcp'),
         expect.objectContaining({
           requestInit: {
             headers: {

packages/core/src/ide/ide-client.ts

@@ -667,10 +667,10 @@ export class IdeClient {
   }
 
   private createProxyAwareFetch() {
-    // ignore proxy for 'localhost' by deafult to allow connecting to the ide mcp server
+    // ignore proxy for '127.0.0.1' by deafult to allow connecting to the ide mcp server
     const existingNoProxy = process.env['NO_PROXY'] || '';
     const agent = new EnvHttpProxyAgent({
-      noProxy: [existingNoProxy, 'localhost'].filter(Boolean).join(','),
+      noProxy: [existingNoProxy, '127.0.0.1'].filter(Boolean).join(','),
     });
     const undiciPromise = import('undici');
     return async (url: string | URL, init?: RequestInit): Promise<Response> => {
@@ -851,5 +851,5 @@ export class IdeClient {
 function getIdeServerHost() {
   const isInContainer =
     fs.existsSync('/.dockerenv') || fs.existsSync('/run/.containerenv');
-  return isInContainer ? 'host.docker.internal' : 'localhost';
+  return isInContainer ? 'host.docker.internal' : '127.0.0.1';
 }

packages/core/src/ide/ide-installer.test.ts

@@ -112,14 +112,19 @@ describe('ide-installer', () => {
           platform: 'linux',
         });
         await installer.install();
+
+        // Note: The implementation uses process.platform, not the mocked platform
+        const isActuallyWindows = process.platform === 'win32';
+        const expectedCommand = isActuallyWindows ? '"code"' : 'code';
+
         expect(child_process.spawnSync).toHaveBeenCalledWith(
-          'code',
+          expectedCommand,
           [
             '--install-extension',
             'qwenlm.qwen-code-vscode-ide-companion',
             '--force',
           ],
-          { stdio: 'pipe' },
+          { stdio: 'pipe', shell: isActuallyWindows },
         );
       });
 

packages/core/src/ide/ide-installer.ts

@@ -117,15 +117,16 @@ class VsCodeInstaller implements IdeInstaller {
       };
     }
 
+    const isWindows = process.platform === 'win32';
     try {
       const result = child_process.spawnSync(
-        commandPath,
+        isWindows ? `"${commandPath}"` : commandPath,
         [
           '--install-extension',
           'qwenlm.qwen-code-vscode-ide-companion',
           '--force',
         ],
-        { stdio: 'pipe' },
+        { stdio: 'pipe', shell: isWindows },
       );
 
       if (result.status !== 0) {

packages/core/src/services/chatCompressionService.test.ts

@@ -0,0 +1,372 @@
+/**
+ * @license
+ * Copyright 2025 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import { describe, it, expect, vi, beforeEach, afterEach } from 'vitest';
+import {
+  ChatCompressionService,
+  findCompressSplitPoint,
+} from './chatCompressionService.js';
+import type { Content, GenerateContentResponse } from '@google/genai';
+import { CompressionStatus } from '../core/turn.js';
+import { uiTelemetryService } from '../telemetry/uiTelemetry.js';
+import { tokenLimit } from '../core/tokenLimits.js';
+import type { GeminiChat } from '../core/geminiChat.js';
+import type { Config } from '../config/config.js';
+import { getInitialChatHistory } from '../utils/environmentContext.js';
+import type { ContentGenerator } from '../core/contentGenerator.js';
+
+vi.mock('../telemetry/uiTelemetry.js');
+vi.mock('../core/tokenLimits.js');
+vi.mock('../telemetry/loggers.js');
+vi.mock('../utils/environmentContext.js');
+
+describe('findCompressSplitPoint', () => {
+  it('should throw an error for non-positive numbers', () => {
+    expect(() => findCompressSplitPoint([], 0)).toThrow(
+      'Fraction must be between 0 and 1',
+    );
+  });
+
+  it('should throw an error for a fraction greater than or equal to 1', () => {
+    expect(() => findCompressSplitPoint([], 1)).toThrow(
+      'Fraction must be between 0 and 1',
+    );
+  });
+
+  it('should handle an empty history', () => {
+    expect(findCompressSplitPoint([], 0.5)).toBe(0);
+  });
+
+  it('should handle a fraction in the middle', () => {
+    const history: Content[] = [
+      { role: 'user', parts: [{ text: 'This is the first message.' }] }, // JSON length: 66 (19%)
+      { role: 'model', parts: [{ text: 'This is the second message.' }] }, // JSON length: 68 (40%)
+      { role: 'user', parts: [{ text: 'This is the third message.' }] }, // JSON length: 66 (60%)
+      { role: 'model', parts: [{ text: 'This is the fourth message.' }] }, // JSON length: 68 (80%)
+      { role: 'user', parts: [{ text: 'This is the fifth message.' }] }, // JSON length: 65 (100%)
+    ];
+    expect(findCompressSplitPoint(history, 0.5)).toBe(4);
+  });
+
+  it('should handle a fraction of last index', () => {
+    const history: Content[] = [
+      { role: 'user', parts: [{ text: 'This is the first message.' }] }, // JSON length: 66 (19%)
+      { role: 'model', parts: [{ text: 'This is the second message.' }] }, // JSON length: 68 (40%)
+      { role: 'user', parts: [{ text: 'This is the third message.' }] }, // JSON length: 66 (60%)
+      { role: 'model', parts: [{ text: 'This is the fourth message.' }] }, // JSON length: 68 (80%)
+      { role: 'user', parts: [{ text: 'This is the fifth message.' }] }, // JSON length: 65 (100%)
+    ];
+    expect(findCompressSplitPoint(history, 0.9)).toBe(4);
+  });
+
+  it('should handle a fraction of after last index', () => {
+    const history: Content[] = [
+      { role: 'user', parts: [{ text: 'This is the first message.' }] }, // JSON length: 66 (24%)
+      { role: 'model', parts: [{ text: 'This is the second message.' }] }, // JSON length: 68 (50%)
+      { role: 'user', parts: [{ text: 'This is the third message.' }] }, // JSON length: 66 (74%)
+      { role: 'model', parts: [{ text: 'This is the fourth message.' }] }, // JSON length: 68 (100%)
+    ];
+    expect(findCompressSplitPoint(history, 0.8)).toBe(4);
+  });
+
+  it('should return earlier splitpoint if no valid ones are after threshhold', () => {
+    const history: Content[] = [
+      { role: 'user', parts: [{ text: 'This is the first message.' }] },
+      { role: 'model', parts: [{ text: 'This is the second message.' }] },
+      { role: 'user', parts: [{ text: 'This is the third message.' }] },
+      { role: 'model', parts: [{ functionCall: { name: 'foo', args: {} } }] },
+    ];
+    // Can't return 4 because the previous item has a function call.
+    expect(findCompressSplitPoint(history, 0.99)).toBe(2);
+  });
+
+  it('should handle a history with only one item', () => {
+    const historyWithEmptyParts: Content[] = [
+      { role: 'user', parts: [{ text: 'Message 1' }] },
+    ];
+    expect(findCompressSplitPoint(historyWithEmptyParts, 0.5)).toBe(0);
+  });
+
+  it('should handle history with weird parts', () => {
+    const historyWithEmptyParts: Content[] = [
+      { role: 'user', parts: [{ text: 'Message 1' }] },
+      {
+        role: 'model',
+        parts: [{ fileData: { fileUri: 'derp', mimeType: 'text/plain' } }],
+      },
+      { role: 'user', parts: [{ text: 'Message 2' }] },
+    ];
+    expect(findCompressSplitPoint(historyWithEmptyParts, 0.5)).toBe(2);
+  });
+});
+
+describe('ChatCompressionService', () => {
+  let service: ChatCompressionService;
+  let mockChat: GeminiChat;
+  let mockConfig: Config;
+  const mockModel = 'gemini-pro';
+  const mockPromptId = 'test-prompt-id';
+
+  beforeEach(() => {
+    service = new ChatCompressionService();
+    mockChat = {
+      getHistory: vi.fn(),
+    } as unknown as GeminiChat;
+    mockConfig = {
+      getChatCompression: vi.fn(),
+      getContentGenerator: vi.fn(),
+    } as unknown as Config;
+
+    vi.mocked(tokenLimit).mockReturnValue(1000);
+    vi.mocked(uiTelemetryService.getLastPromptTokenCount).mockReturnValue(500);
+    vi.mocked(getInitialChatHistory).mockImplementation(
+      async (_config, extraHistory) => extraHistory || [],
+    );
+  });
+
+  afterEach(() => {
+    vi.restoreAllMocks();
+  });
+
+  it('should return NOOP if history is empty', async () => {
+    vi.mocked(mockChat.getHistory).mockReturnValue([]);
+    const result = await service.compress(
+      mockChat,
+      mockPromptId,
+      false,
+      mockModel,
+      mockConfig,
+      false,
+    );
+    expect(result.info.compressionStatus).toBe(CompressionStatus.NOOP);
+    expect(result.newHistory).toBeNull();
+  });
+
+  it('should return NOOP if previously failed and not forced', async () => {
+    vi.mocked(mockChat.getHistory).mockReturnValue([
+      { role: 'user', parts: [{ text: 'hi' }] },
+    ]);
+    const result = await service.compress(
+      mockChat,
+      mockPromptId,
+      false,
+      mockModel,
+      mockConfig,
+      true,
+    );
+    expect(result.info.compressionStatus).toBe(CompressionStatus.NOOP);
+    expect(result.newHistory).toBeNull();
+  });
+
+  it('should return NOOP if under token threshold and not forced', async () => {
+    vi.mocked(mockChat.getHistory).mockReturnValue([
+      { role: 'user', parts: [{ text: 'hi' }] },
+    ]);
+    vi.mocked(uiTelemetryService.getLastPromptTokenCount).mockReturnValue(600);
+    vi.mocked(tokenLimit).mockReturnValue(1000);
+    // Threshold is 0.7 * 1000 = 700. 600 < 700, so NOOP.
+
+    const result = await service.compress(
+      mockChat,
+      mockPromptId,
+      false,
+      mockModel,
+      mockConfig,
+      false,
+    );
+    expect(result.info.compressionStatus).toBe(CompressionStatus.NOOP);
+    expect(result.newHistory).toBeNull();
+  });
+
+  it('should compress if over token threshold', async () => {
+    const history: Content[] = [
+      { role: 'user', parts: [{ text: 'msg1' }] },
+      { role: 'model', parts: [{ text: 'msg2' }] },
+      { role: 'user', parts: [{ text: 'msg3' }] },
+      { role: 'model', parts: [{ text: 'msg4' }] },
+    ];
+    vi.mocked(mockChat.getHistory).mockReturnValue(history);
+    vi.mocked(uiTelemetryService.getLastPromptTokenCount).mockReturnValue(800);
+    vi.mocked(tokenLimit).mockReturnValue(1000);
+    const mockGenerateContent = vi.fn().mockResolvedValue({
+      candidates: [
+        {
+          content: {
+            parts: [{ text: 'Summary' }],
+          },
+        },
+      ],
+    } as unknown as GenerateContentResponse);
+    vi.mocked(mockConfig.getContentGenerator).mockReturnValue({
+      generateContent: mockGenerateContent,
+    } as unknown as ContentGenerator);
+
+    const result = await service.compress(
+      mockChat,
+      mockPromptId,
+      false,
+      mockModel,
+      mockConfig,
+      false,
+    );
+
+    expect(result.info.compressionStatus).toBe(CompressionStatus.COMPRESSED);
+    expect(result.newHistory).not.toBeNull();
+    expect(result.newHistory![0].parts![0].text).toBe('Summary');
+    expect(mockGenerateContent).toHaveBeenCalled();
+  });
+
+  it('should force compress even if under threshold', async () => {
+    const history: Content[] = [
+      { role: 'user', parts: [{ text: 'msg1' }] },
+      { role: 'model', parts: [{ text: 'msg2' }] },
+      { role: 'user', parts: [{ text: 'msg3' }] },
+      { role: 'model', parts: [{ text: 'msg4' }] },
+    ];
+    vi.mocked(mockChat.getHistory).mockReturnValue(history);
+    vi.mocked(uiTelemetryService.getLastPromptTokenCount).mockReturnValue(100);
+    vi.mocked(tokenLimit).mockReturnValue(1000);
+
+    const mockGenerateContent = vi.fn().mockResolvedValue({
+      candidates: [
+        {
+          content: {
+            parts: [{ text: 'Summary' }],
+          },
+        },
+      ],
+    } as unknown as GenerateContentResponse);
+    vi.mocked(mockConfig.getContentGenerator).mockReturnValue({
+      generateContent: mockGenerateContent,
+    } as unknown as ContentGenerator);
+
+    const result = await service.compress(
+      mockChat,
+      mockPromptId,
+      true, // forced
+      mockModel,
+      mockConfig,
+      false,
+    );
+
+    expect(result.info.compressionStatus).toBe(CompressionStatus.COMPRESSED);
+    expect(result.newHistory).not.toBeNull();
+  });
+
+  it('should return FAILED if new token count is inflated', async () => {
+    const history: Content[] = [
+      { role: 'user', parts: [{ text: 'msg1' }] },
+      { role: 'model', parts: [{ text: 'msg2' }] },
+    ];
+    vi.mocked(mockChat.getHistory).mockReturnValue(history);
+    vi.mocked(uiTelemetryService.getLastPromptTokenCount).mockReturnValue(10);
+    vi.mocked(tokenLimit).mockReturnValue(1000);
+
+    const longSummary = 'a'.repeat(1000); // Long summary to inflate token count
+    const mockGenerateContent = vi.fn().mockResolvedValue({
+      candidates: [
+        {
+          content: {
+            parts: [{ text: longSummary }],
+          },
+        },
+      ],
+    } as unknown as GenerateContentResponse);
+    vi.mocked(mockConfig.getContentGenerator).mockReturnValue({
+      generateContent: mockGenerateContent,
+    } as unknown as ContentGenerator);
+
+    const result = await service.compress(
+      mockChat,
+      mockPromptId,
+      true,
+      mockModel,
+      mockConfig,
+      false,
+    );
+
+    expect(result.info.compressionStatus).toBe(
+      CompressionStatus.COMPRESSION_FAILED_INFLATED_TOKEN_COUNT,
+    );
+    expect(result.newHistory).toBeNull();
+  });
+
+  it('should return FAILED if summary is empty string', async () => {
+    const history: Content[] = [
+      { role: 'user', parts: [{ text: 'msg1' }] },
+      { role: 'model', parts: [{ text: 'msg2' }] },
+    ];
+    vi.mocked(mockChat.getHistory).mockReturnValue(history);
+    vi.mocked(uiTelemetryService.getLastPromptTokenCount).mockReturnValue(100);
+    vi.mocked(tokenLimit).mockReturnValue(1000);
+
+    const mockGenerateContent = vi.fn().mockResolvedValue({
+      candidates: [
+        {
+          content: {
+            parts: [{ text: '' }], // Empty summary
+          },
+        },
+      ],
+    } as unknown as GenerateContentResponse);
+    vi.mocked(mockConfig.getContentGenerator).mockReturnValue({
+      generateContent: mockGenerateContent,
+    } as unknown as ContentGenerator);
+
+    const result = await service.compress(
+      mockChat,
+      mockPromptId,
+      true,
+      mockModel,
+      mockConfig,
+      false,
+    );
+
+    expect(result.info.compressionStatus).toBe(
+      CompressionStatus.COMPRESSION_FAILED_EMPTY_SUMMARY,
+    );
+    expect(result.newHistory).toBeNull();
+    expect(result.info.originalTokenCount).toBe(100);
+    expect(result.info.newTokenCount).toBe(100);
+  });
+
+  it('should return FAILED if summary is only whitespace', async () => {
+    const history: Content[] = [
+      { role: 'user', parts: [{ text: 'msg1' }] },
+      { role: 'model', parts: [{ text: 'msg2' }] },
+    ];
+    vi.mocked(mockChat.getHistory).mockReturnValue(history);
+    vi.mocked(uiTelemetryService.getLastPromptTokenCount).mockReturnValue(100);
+    vi.mocked(tokenLimit).mockReturnValue(1000);
+
+    const mockGenerateContent = vi.fn().mockResolvedValue({
+      candidates: [
+        {
+          content: {
+            parts: [{ text: '   \n\t  ' }], // Only whitespace
+          },
+        },
+      ],
+    } as unknown as GenerateContentResponse);
+    vi.mocked(mockConfig.getContentGenerator).mockReturnValue({
+      generateContent: mockGenerateContent,
+    } as unknown as ContentGenerator);
+
+    const result = await service.compress(
+      mockChat,
+      mockPromptId,
+      true,
+      mockModel,
+      mockConfig,
+      false,
+    );
+
+    expect(result.info.compressionStatus).toBe(
+      CompressionStatus.COMPRESSION_FAILED_EMPTY_SUMMARY,
+    );
+    expect(result.newHistory).toBeNull();
+  });
+});

packages/core/src/services/chatCompressionService.ts

@@ -0,0 +1,235 @@
+/**
+ * @license
+ * Copyright 2025 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import type { Content } from '@google/genai';
+import type { Config } from '../config/config.js';
+import type { GeminiChat } from '../core/geminiChat.js';
+import { type ChatCompressionInfo, CompressionStatus } from '../core/turn.js';
+import { uiTelemetryService } from '../telemetry/uiTelemetry.js';
+import { tokenLimit } from '../core/tokenLimits.js';
+import { getCompressionPrompt } from '../core/prompts.js';
+import { getResponseText } from '../utils/partUtils.js';
+import { logChatCompression } from '../telemetry/loggers.js';
+import { makeChatCompressionEvent } from '../telemetry/types.js';
+import { getInitialChatHistory } from '../utils/environmentContext.js';
+
+/**
+ * Threshold for compression token count as a fraction of the model's token limit.
+ * If the chat history exceeds this threshold, it will be compressed.
+ */
+export const COMPRESSION_TOKEN_THRESHOLD = 0.7;
+
+/**
+ * The fraction of the latest chat history to keep. A value of 0.3
+ * means that only the last 30% of the chat history will be kept after compression.
+ */
+export const COMPRESSION_PRESERVE_THRESHOLD = 0.3;
+
+/**
+ * Returns the index of the oldest item to keep when compressing. May return
+ * contents.length which indicates that everything should be compressed.
+ *
+ * Exported for testing purposes.
+ */
+export function findCompressSplitPoint(
+  contents: Content[],
+  fraction: number,
+): number {
+  if (fraction <= 0 || fraction >= 1) {
+    throw new Error('Fraction must be between 0 and 1');
+  }
+
+  const charCounts = contents.map((content) => JSON.stringify(content).length);
+  const totalCharCount = charCounts.reduce((a, b) => a + b, 0);
+  const targetCharCount = totalCharCount * fraction;
+
+  let lastSplitPoint = 0; // 0 is always valid (compress nothing)
+  let cumulativeCharCount = 0;
+  for (let i = 0; i < contents.length; i++) {
+    const content = contents[i];
+    if (
+      content.role === 'user' &&
+      !content.parts?.some((part) => !!part.functionResponse)
+    ) {
+      if (cumulativeCharCount >= targetCharCount) {
+        return i;
+      }
+      lastSplitPoint = i;
+    }
+    cumulativeCharCount += charCounts[i];
+  }
+
+  // We found no split points after targetCharCount.
+  // Check if it's safe to compress everything.
+  const lastContent = contents[contents.length - 1];
+  if (
+    lastContent?.role === 'model' &&
+    !lastContent?.parts?.some((part) => part.functionCall)
+  ) {
+    return contents.length;
+  }
+
+  // Can't compress everything so just compress at last splitpoint.
+  return lastSplitPoint;
+}
+
+export class ChatCompressionService {
+  async compress(
+    chat: GeminiChat,
+    promptId: string,
+    force: boolean,
+    model: string,
+    config: Config,
+    hasFailedCompressionAttempt: boolean,
+  ): Promise<{ newHistory: Content[] | null; info: ChatCompressionInfo }> {
+    const curatedHistory = chat.getHistory(true);
+
+    // Regardless of `force`, don't do anything if the history is empty.
+    if (
+      curatedHistory.length === 0 ||
+      (hasFailedCompressionAttempt && !force)
+    ) {
+      return {
+        newHistory: null,
+        info: {
+          originalTokenCount: 0,
+          newTokenCount: 0,
+          compressionStatus: CompressionStatus.NOOP,
+        },
+      };
+    }
+
+    const originalTokenCount = uiTelemetryService.getLastPromptTokenCount();
+
+    const contextPercentageThreshold =
+      config.getChatCompression()?.contextPercentageThreshold;
+
+    // Don't compress if not forced and we are under the limit.
+    if (!force) {
+      const threshold =
+        contextPercentageThreshold ?? COMPRESSION_TOKEN_THRESHOLD;
+      if (originalTokenCount < threshold * tokenLimit(model)) {
+        return {
+          newHistory: null,
+          info: {
+            originalTokenCount,
+            newTokenCount: originalTokenCount,
+            compressionStatus: CompressionStatus.NOOP,
+          },
+        };
+      }
+    }
+
+    const splitPoint = findCompressSplitPoint(
+      curatedHistory,
+      1 - COMPRESSION_PRESERVE_THRESHOLD,
+    );
+
+    const historyToCompress = curatedHistory.slice(0, splitPoint);
+    const historyToKeep = curatedHistory.slice(splitPoint);
+
+    if (historyToCompress.length === 0) {
+      return {
+        newHistory: null,
+        info: {
+          originalTokenCount,
+          newTokenCount: originalTokenCount,
+          compressionStatus: CompressionStatus.NOOP,
+        },
+      };
+    }
+
+    const summaryResponse = await config.getContentGenerator().generateContent(
+      {
+        model,
+        contents: [
+          ...historyToCompress,
+          {
+            role: 'user',
+            parts: [
+              {
+                text: 'First, reason in your scratchpad. Then, generate the <state_snapshot>.',
+              },
+            ],
+          },
+        ],
+        config: {
+          systemInstruction: getCompressionPrompt(),
+        },
+      },
+      promptId,
+    );
+    const summary = getResponseText(summaryResponse) ?? '';
+    const isSummaryEmpty = !summary || summary.trim().length === 0;
+
+    let newTokenCount = originalTokenCount;
+    let extraHistory: Content[] = [];
+
+    if (!isSummaryEmpty) {
+      extraHistory = [
+        {
+          role: 'user',
+          parts: [{ text: summary }],
+        },
+        {
+          role: 'model',
+          parts: [{ text: 'Got it. Thanks for the additional context!' }],
+        },
+        ...historyToKeep,
+      ];
+
+      // Use a shared utility to construct the initial history for an accurate token count.
+      const fullNewHistory = await getInitialChatHistory(config, extraHistory);
+
+      // Estimate token count 1 token ≈ 4 characters
+      newTokenCount = Math.floor(
+        fullNewHistory.reduce(
+          (total, content) => total + JSON.stringify(content).length,
+          0,
+        ) / 4,
+      );
+    }
+
+    logChatCompression(
+      config,
+      makeChatCompressionEvent({
+        tokens_before: originalTokenCount,
+        tokens_after: newTokenCount,
+      }),
+    );
+
+    if (isSummaryEmpty) {
+      return {
+        newHistory: null,
+        info: {
+          originalTokenCount,
+          newTokenCount: originalTokenCount,
+          compressionStatus: CompressionStatus.COMPRESSION_FAILED_EMPTY_SUMMARY,
+        },
+      };
+    } else if (newTokenCount > originalTokenCount) {
+      return {
+        newHistory: null,
+        info: {
+          originalTokenCount,
+          newTokenCount,
+          compressionStatus:
+            CompressionStatus.COMPRESSION_FAILED_INFLATED_TOKEN_COUNT,
+        },
+      };
+    } else {
+      uiTelemetryService.setLastPromptTokenCount(newTokenCount);
+      return {
+        newHistory: extraHistory,
+        info: {
+          originalTokenCount,
+          newTokenCount,
+          compressionStatus: CompressionStatus.COMPRESSED,
+        },
+      };
+    }
+  }
+}

packages/core/src/subagents/subagent.test.ts

@@ -32,7 +32,6 @@ import { GeminiChat } from '../core/geminiChat.js';
 import { executeToolCall } from '../core/nonInteractiveToolExecutor.js';
 import type { ToolRegistry } from '../tools/tool-registry.js';
 import { type AnyDeclarativeTool } from '../tools/tools.js';
-import { getEnvironmentContext } from '../utils/environmentContext.js';
 import { ContextState, SubAgentScope } from './subagent.js';
 import type {
   ModelConfig,
@@ -44,7 +43,20 @@ import { SubagentTerminateMode } from './types.js';
 
 vi.mock('../core/geminiChat.js');
 vi.mock('../core/contentGenerator.js');
-vi.mock('../utils/environmentContext.js');
+vi.mock('../utils/environmentContext.js', () => ({
+  getEnvironmentContext: vi.fn().mockResolvedValue([{ text: 'Env Context' }]),
+  getInitialChatHistory: vi.fn(async (_config, extraHistory) => [
+    {
+      role: 'user',
+      parts: [{ text: 'Env Context' }],
+    },
+    {
+      role: 'model',
+      parts: [{ text: 'Got it. Thanks for the context!' }],
+    },
+    ...(extraHistory ?? []),
+  ]),
+}));
 vi.mock('../core/nonInteractiveToolExecutor.js');
 vi.mock('../ide/ide-client.js');
 vi.mock('../core/client.js');
@@ -174,9 +186,6 @@ describe('subagent.ts', () => {
     beforeEach(async () => {
       vi.clearAllMocks();
 
-      vi.mocked(getEnvironmentContext).mockResolvedValue([
-        { text: 'Env Context' },
-      ]);
       vi.mocked(createContentGenerator).mockResolvedValue({
         getGenerativeModel: vi.fn(),
         // eslint-disable-next-line @typescript-eslint/no-explicit-any

packages/core/src/subagents/subagent.ts

@@ -16,7 +16,7 @@ import type {
   ToolConfirmationOutcome,
   ToolCallConfirmationDetails,
 } from '../tools/tools.js';
-import { getEnvironmentContext } from '../utils/environmentContext.js';
+import { getInitialChatHistory } from '../utils/environmentContext.js';
 import type {
   Content,
   Part,
@@ -807,11 +807,7 @@ export class SubAgentScope {
       );
     }
 
-    const envParts = await getEnvironmentContext(this.runtimeContext);
-    const envHistory: Content[] = [
-      { role: 'user', parts: envParts },
-      { role: 'model', parts: [{ text: 'Got it. Thanks for the context!' }] },
-    ];
+    const envHistory = await getInitialChatHistory(this.runtimeContext);
 
     const start_history = [
       ...envHistory,

packages/core/src/utils/environmentContext.ts

@@ -4,7 +4,7 @@
  * SPDX-License-Identifier: Apache-2.0
  */
 
-import type { Part } from '@google/genai';
+import type { Content, Part } from '@google/genai';
 import type { Config } from '../config/config.js';
 import { getFolderStructure } from './getFolderStructure.js';
 
@@ -107,3 +107,23 @@ ${directoryContext}
 
   return initialParts;
 }
+
+export async function getInitialChatHistory(
+  config: Config,
+  extraHistory?: Content[],
+): Promise<Content[]> {
+  const envParts = await getEnvironmentContext(config);
+  const envContextString = envParts.map((part) => part.text || '').join('\n\n');
+
+  return [
+    {
+      role: 'user',
+      parts: [{ text: envContextString }],
+    },
+    {
+      role: 'model',
+      parts: [{ text: 'Got it. Thanks for the context!' }],
+    },
+    ...(extraHistory ?? []),
+  ];
+}

packages/test-utils/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@qwen-code/qwen-code-test-utils",
-  "version": "0.1.2",
+  "version": "0.1.3-preview.0",
   "private": true,
   "main": "src/index.ts",
   "license": "Apache-2.0",

packages/vscode-ide-companion/package.json

@@ -2,7 +2,7 @@
   "name": "qwen-code-vscode-ide-companion",
   "displayName": "Qwen Code Companion",
   "description": "Enable Qwen Code with direct access to your VS Code workspace.",
-  "version": "0.1.2",
+  "version": "0.1.3-preview.0",
   "publisher": "qwenlm",
   "icon": "assets/icon.png",
   "repository": {

scripts/prepare-package.js

@@ -85,7 +85,7 @@ const distPackageJson = {
   bin: {
     qwen: 'cli.js',
   },
-  files: ['cli.js', 'vendor', 'README.md', 'LICENSE'],
+  files: ['cli.js', 'vendor', '*.sb', 'README.md', 'LICENSE'],
   config: rootPackageJson.config,
   dependencies: runtimeDependencies,
   optionalDependencies: {