diff --git a/CHANGELOG.md b/CHANGELOG.md
index 72bfdbd415..cc9721773a 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,5 +1,35 @@
 # Zoo Code Changelog
 
+## [3.80.0]
+
+### Minor Changes
+
+- Add separate Read and Write allowlists so users can restrict file operations to approved paths (#1273 by @nh2, PR #1274 by @nh2)
+- Add official GLM 5.3 support to the Z.ai API (PR #1284 by @app/zoomote)
+- Add Gemini 3.5 Flash Lite and Gemini 3.1 Flash Lite to the Gemini API provider (#1251 by @MrMody0, PR #1334 by @easonLiangWorldedtech)
+- Prevent webview gray screens and out-of-memory failures by throttling extension-state pushes (#629 by @edelauna, PR #1078 by @JunyongParkDev)
+- Recover stale delegated child tasks after an extension restart (#1100 by @aniruddhabag3, PR #1210 by @edelauna)
+- Replace the shared task-history index with atomic per-task merges to prevent lost or corrupted history (#1231 by @martin-rueegg, PR #1261 by @edelauna)
+- Mark interrupted tool calls as errors in persisted task history so resumed conversations remain consistent (#1283 by @martin-rueegg, PR #1323 by @easonLiangWorldedtech)
+- Complete OpenAI Codex prompts over the streaming transport instead of waiting indefinitely (#1242 by @Rafael-Silva-Oliveira, PR #1243 by @Rafael-Silva-Oliveira)
+- Stream LM Studio `reasoning_content` so model reasoning is preserved in responses (#1075 by @marunguy, PR #1175 by @daewoongoh)
+- Fix NanoGPT Muse Spark tool-use requests (PR #1310 by @app/zoomote)
+- Make settings import await provider-state persistence before reporting success (PR #1340 by @app/zoomote)
+- Default telemetry to opt-out and require explicit consent through the telemetry UI (PR #1069 by @edelauna)
+- Introduce an SDK-agnostic request configuration builder for consistent abort-signal support (#1007 by @easonLiangWorldedtech, PR #1008 by @easonLiangWorldedtech)
+- Delegate VS Code shell-profile reads to Terminal methods for more reliable profile detection (#634 by @marty-a11y, PR #1033 by @edelauna)
+- Increase the Semble archive download limit for larger codebase-intelligence binaries (#1305 by @navedmerchant, PR #1306 by @navedmerchant)
+- Stop estimating Zoo Gateway UI costs from unrelated default-model prices when authoritative cost data is unavailable (PR #1339 by @JamesRobert20)
+- Improve Mermaid label contrast in light IDE themes (PR #1285 by @app/zoomote)
+- Keep chat controls legible in light IDE themes (PR #1298 by @app/zoomote)
+- Fix the remaining chat controls that faded into light IDE themes (PR #1312 by @app/zoomote)
+- Improve shared-control contrast across IDE themes (PR #1333 by @app/zoomote)
+- Canonicalize provider and model identifiers in webview messages (PR #1143 by @WebMad)
+- Canonicalize model-router calls across providers (PR #1165 by @WebMad)
+- Enforce canonical provider identifiers through provider lint rules (#1294 by @WebMad, PR #1297 by @WebMad)
+- Enforce safe asynchronous handling in core tools (PR #1255 by @app/zoomote)
+- Build downloadable VSIX artifacts for pull requests (PR #1314 by @app/zoomote)
+
 ## [3.78.0]
 
 ### Minor Changes
diff --git a/README.md b/README.md
index b0bd6a083c..fac6c536ee 100644
--- a/README.md
+++ b/README.md
@@ -46,13 +46,11 @@ Zoo Code builds on the foundation created by Roo Code and continues to expand it
 - **More dependable terminal and editing workflows** — fixes for premature terminal completion, task-state races, context management, diff editing, and provider-specific tool use.
 - **More control over your workspace** — rules management, per-mode MCP restrictions, multi-root path controls, model reasoning options, and completion change review actions.
 
-## What's New in v3.78.0
+## What's New in v3.80.0
 
-- **Three major new models have arrived** — use the brand-new Gemini 3.7 Flash, GLM 5.3, and Qwen3.8 Max models, plus updated DeepSeek V4 reasoning, pricing, and provider coverage.
-- **Connect to NanoGPT** — use dynamic model discovery, streaming and prompt completions, and routing preferences for speed, price, latency, throughput, tool support, and caching.
-- **More reliable providers and tasks** — fixes improve Azure OpenAI endpoint setup, Kimi Code output limits, task-history title preservation, and Zoo settings import/export.
-- Destructive Command Guard now supports Intel-based Macs.
-- Security updates address vulnerabilities in `undici` and Mermaid.
+- **Safer file access** — use separate Read and Write allowlists to control which paths Zoo Code can access.
+- **More model choice** — GLM 5.3 is now available through Z.ai, with Gemini 3.5 Flash Lite and Gemini 3.1 Flash Lite added to the Gemini provider.
+- **More reliable everyday workflows** — improvements strengthen task recovery and history, provider streaming and tool use, settings imports, and readability across IDE themes.
 
 <details>
   <summary>🌐 Available languages</summary>
diff --git a/locales/ca/README.md b/locales/ca/README.md
index 4cc35f4364..ba8f554efb 100644
--- a/locales/ca/README.md
+++ b/locales/ca/README.md
@@ -47,13 +47,11 @@ Zoo Code parteix de la base creada per Roo Code i continua ampliant-la amb:
 - **Fluxos de terminal i edició més fiables** — correccions per a la finalització prematura del terminal, les condicions de cursa en l'estat de les tasques, la gestió del context, l'edició de diff i l'ús d'eines específiques de cada proveïdor.
 - **Més control sobre el teu espai de treball** — gestió de regles, restriccions MCP per mode, controls de rutes multiarrel, opcions de raonament dels models i accions per revisar els canvis en completar una tasca.
 
-## Novetats a la v3.78.0
+## Novetats a la v3.80.0
 
-- **Han arribat tres grans models nous** — utilitza els nous Gemini 3.7 Flash, GLM 5.3 i Qwen3.8 Max, a més de millores en el raonament, els preus i la cobertura de proveïdors de DeepSeek V4.
-- **Connecta't a NanoGPT** — utilitza el descobriment dinàmic de models, streaming i completions de prompts, i preferències d'encaminament per velocitat, preu, latència, rendiment, compatibilitat amb eines i memòria cau.
-- **Proveïdors i tasques més fiables** — les correccions milloren la configuració dels endpoints d'Azure OpenAI, els límits de sortida de Kimi Code, la conservació dels títols de l'historial de tasques i la importació/exportació de configuració de Zoo.
-- Destructive Command Guard ara és compatible amb els Mac basats en Intel.
-- Les actualitzacions de seguretat solucionen vulnerabilitats a `undici` i Mermaid.
+- **Accés als fitxers més segur** — utilitza llistes de permisos de lectura i escriptura separades per controlar a quines rutes pot accedir Zoo Code.
+- **Més models per triar** — GLM 5.3 ja està disponible mitjançant Z.ai, i Gemini 3.5 Flash Lite i Gemini 3.1 Flash Lite s'han afegit al proveïdor Gemini.
+- **Fluxos de treball quotidians més fiables** — les millores reforcen la recuperació i l'historial de tasques, l'streaming i l'ús d'eines dels proveïdors, la importació de configuració i la llegibilitat en diferents temes de l'IDE.
 
 ## Què pot fer Zoo Code per TU?
 
diff --git a/locales/de/README.md b/locales/de/README.md
index 62626dd353..57bd315e24 100644
--- a/locales/de/README.md
+++ b/locales/de/README.md
@@ -47,13 +47,11 @@ Zoo Code baut auf dem von Roo Code geschaffenen Fundament auf und erweitert es f
 - **Zuverlässigere Terminal- und Bearbeitungsabläufe** — Korrekturen für vorzeitige Terminalabschlüsse, Race Conditions beim Aufgabenstatus, Kontextverwaltung, diff-Bearbeitung und anbieterspezifische Tool-Nutzung.
 - **Mehr Kontrolle über deinen Workspace** — Regelverwaltung, MCP-Beschränkungen pro Modus, Pfadsteuerung für Multi-Root-Workspaces, Reasoning-Optionen für Modelle und Aktionen zur Prüfung von Änderungen nach Abschluss.
 
-## Neu in v3.78.0
+## Neu in v3.80.0
 
-- **Drei bedeutende neue Modelle sind da** — nutze die brandneuen Modelle Gemini 3.7 Flash, GLM 5.3 und Qwen3.8 Max sowie aktualisiertes Reasoning, Preise und Anbieterabdeckung für DeepSeek V4.
-- **Verbinde dich mit NanoGPT** — nutze dynamische Modellerkennung, Streaming und Prompt-Vervollständigung sowie Routing-Einstellungen für Geschwindigkeit, Preis, Latenz, Durchsatz, Tool-Unterstützung und Caching.
-- **Zuverlässigere Anbieter und Aufgaben** — Korrekturen verbessern die Einrichtung von Azure-OpenAI-Endpunkten, Kimi-Code-Ausgabelimits, die Beibehaltung von Titeln im Aufgabenverlauf sowie den Import und Export von Zoo-Einstellungen.
-- Destructive Command Guard unterstützt jetzt Intel-basierte Macs.
-- Sicherheitsupdates beheben Schwachstellen in `undici` und Mermaid.
+- **Sicherer Dateizugriff** — nutze getrennte Lese- und Schreib-Positivlisten, um zu steuern, auf welche Pfade Zoo Code zugreifen darf.
+- **Mehr Modellauswahl** — GLM 5.3 ist jetzt über Z.ai verfügbar; außerdem wurden Gemini 3.5 Flash Lite und Gemini 3.1 Flash Lite zum Gemini-Anbieter hinzugefügt.
+- **Zuverlässigere tägliche Workflows** — Verbesserungen stärken die Wiederherstellung und den Verlauf von Aufgaben, Streaming und Tool-Nutzung bei Anbietern, Einstellungsimporte und die Lesbarkeit in verschiedenen IDE-Themes.
 
 ## Was kann Zoo Code für DICH tun?
 
diff --git a/locales/es/README.md b/locales/es/README.md
index a7eabc790d..4c6740f2bf 100644
--- a/locales/es/README.md
+++ b/locales/es/README.md
@@ -47,13 +47,11 @@ Zoo Code parte de los cimientos creados por Roo Code y continúa ampliándolos c
 - **Flujos de terminal y edición más fiables** — correcciones para la finalización prematura del terminal, las condiciones de carrera del estado de las tareas, la gestión del contexto, la edición de diff y el uso de herramientas específicas de cada proveedor.
 - **Más control sobre tu espacio de trabajo** — gestión de reglas, restricciones de MCP por modo, controles de rutas multirraíz, opciones de razonamiento de modelos y acciones para revisar los cambios al completar una tarea.
 
-## Novedades de la v3.78.0
+## Novedades de la v3.80.0
 
-- **Han llegado tres importantes modelos nuevos** — usa los flamantes Gemini 3.7 Flash, GLM 5.3 y Qwen3.8 Max, además de mejoras en el razonamiento, los precios y la cobertura de proveedores de DeepSeek V4.
-- **Conéctate a NanoGPT** — usa descubrimiento dinámico de modelos, streaming y completado de prompts, y preferencias de enrutamiento por velocidad, precio, latencia, rendimiento, compatibilidad con herramientas y caché.
-- **Proveedores y tareas más fiables** — las correcciones mejoran la configuración de endpoints de Azure OpenAI, los límites de salida de Kimi Code, la conservación de títulos del historial de tareas y la importación/exportación de ajustes de Zoo.
-- Destructive Command Guard ahora es compatible con Macs basados en Intel.
-- Las actualizaciones de seguridad corrigen vulnerabilidades en `undici` y Mermaid.
+- **Acceso a archivos más seguro** — usa listas de permitidos separadas para lectura y escritura a fin de controlar a qué rutas puede acceder Zoo Code.
+- **Más modelos para elegir** — GLM 5.3 ya está disponible mediante Z.ai, y Gemini 3.5 Flash Lite y Gemini 3.1 Flash Lite se han añadido al proveedor de Gemini.
+- **Flujos de trabajo diarios más fiables** — las mejoras refuerzan la recuperación y el historial de tareas, el streaming y el uso de herramientas de los proveedores, la importación de ajustes y la legibilidad en distintos temas del IDE.
 
 ## ¿Qué puede hacer Zoo Code por TI?
 
diff --git a/locales/fr/README.md b/locales/fr/README.md
index b2bddceb55..e06c06c678 100644
--- a/locales/fr/README.md
+++ b/locales/fr/README.md
@@ -47,13 +47,11 @@ Zoo Code s'appuie sur les fondations créées par Roo Code et continue de les en
 - **Des workflows de terminal et d'édition plus fiables** — correctifs pour les fins prématurées de commandes dans le terminal, les conditions de concurrence liées à l'état des tâches, la gestion du contexte, l'édition de diff et l'utilisation d'outils propres aux providers.
 - **Davantage de contrôle sur ton espace de travail** — gestion des règles, restrictions MCP par mode, contrôle des chemins multi-root, options de raisonnement des modèles et actions de vérification des modifications à la fin d'une tâche.
 
-## Nouveautés de la v3.78.0
+## Nouveautés de la v3.80.0
 
-- **Trois nouveaux modèles majeurs sont arrivés** — utilise les tout nouveaux Gemini 3.7 Flash, GLM 5.3 et Qwen3.8 Max, ainsi que les améliorations du raisonnement, des tarifs et de la couverture des providers de DeepSeek V4.
-- **Connecte-toi à NanoGPT** — profite de la découverte dynamique des modèles, du streaming et de la complétion des prompts, avec des préférences de routage pour la vitesse, le prix, la latence, le débit, la prise en charge des outils et le cache.
-- **Providers et tâches plus fiables** — les correctifs améliorent la configuration des endpoints Azure OpenAI, les limites de sortie de Kimi Code, la conservation des titres dans l'historique des tâches et l'import/export des paramètres Zoo.
-- Destructive Command Guard prend désormais en charge les Mac équipés d'un processeur Intel.
-- Des mises à jour de sécurité corrigent des vulnérabilités dans `undici` et Mermaid.
+- **Accès aux fichiers plus sûr** — utilise des listes d'autorisation distinctes pour la lecture et l'écriture afin de contrôler les chemins auxquels Zoo Code peut accéder.
+- **Plus de modèles au choix** — GLM 5.3 est désormais disponible via Z.ai, et Gemini 3.5 Flash Lite ainsi que Gemini 3.1 Flash Lite ont été ajoutés au provider Gemini.
+- **Des workflows quotidiens plus fiables** — les améliorations renforcent la récupération et l'historique des tâches, le streaming et l'utilisation des outils par les providers, l'import des paramètres et la lisibilité dans les différents thèmes d'IDE.
 
 ## Que peut faire Zoo Code pour VOUS ?
 
diff --git a/locales/hi/README.md b/locales/hi/README.md
index 388ba7c525..d0762c6096 100644
--- a/locales/hi/README.md
+++ b/locales/hi/README.md
@@ -46,13 +46,11 @@ Zoo Code, Roo Code की बनाई नींव पर आगे बढ़
 - **ज़्यादा भरोसेमंद terminal और editing workflows** — terminal के समय से पहले पूरा होने, task-state race conditions, context management, diff editing और provider-specific tool use से जुड़ी समस्याओं के fixes।
 - **अपने workspace पर ज़्यादा control** — rules management, हर mode के लिए MCP restrictions, multi-root path controls, model reasoning options और completion changes की review actions।
 
-## v3.78.0 में नया क्या है
+## v3.80.0 में नया क्या है
 
-- **तीन बड़े नए models आ गए हैं** — बिल्कुल नए Gemini 3.7 Flash, GLM 5.3 और Qwen3.8 Max models का उपयोग करो, साथ ही DeepSeek V4 के updated reasoning, pricing और provider coverage का लाभ लो।
-- **NanoGPT से connect करो** — dynamic model discovery, streaming और prompt completions के साथ speed, price, latency, throughput, tool support और caching के लिए routing preferences का उपयोग करो।
-- **अधिक reliable providers और tasks** — fixes Azure OpenAI endpoint setup, Kimi Code output limits, task-history title preservation और Zoo settings import/export को बेहतर बनाते हैं।
-- Destructive Command Guard अब Intel-आधारित Macs को support करता है।
-- Security updates `undici` और Mermaid की vulnerabilities को ठीक करते हैं।
+- **ज़्यादा सुरक्षित file access** — Zoo Code किन paths को access कर सकता है, इसे नियंत्रित करने के लिए अलग-अलग Read और Write allowlists का उपयोग करें।
+- **Models के ज़्यादा विकल्प** — GLM 5.3 अब Z.ai के माध्यम से उपलब्ध है, और Gemini provider में Gemini 3.5 Flash Lite और Gemini 3.1 Flash Lite जोड़े गए हैं।
+- **रोज़मर्रा के ज़्यादा reliable workflows** — सुधार task recovery और history, provider streaming और tool use, settings imports तथा IDE themes में readability को बेहतर बनाते हैं।
 
 ## Zoo Code आपके लिए क्या कर सकता है?
 
diff --git a/locales/id/README.md b/locales/id/README.md
index 794dc46dd1..1b30e87546 100644
--- a/locales/id/README.md
+++ b/locales/id/README.md
@@ -46,13 +46,11 @@ Zoo Code dikembangkan di atas fondasi yang dibuat oleh Roo Code dan terus memper
 - **Workflow terminal dan pengeditan yang lebih andal** — perbaikan untuk terminal yang selesai terlalu dini, race condition status task, pengelolaan konteks, pengeditan diff, dan penggunaan tool khusus provider.
 - **Kontrol lebih besar atas workspace kamu** — pengelolaan rules, pembatasan MCP per mode, kontrol path multi-root, opsi reasoning model, dan tindakan untuk meninjau perubahan saat selesai.
 
-## Yang Baru di v3.78.0
+## Yang Baru di v3.80.0
 
-- **Tiga model baru utama telah hadir** — gunakan model terbaru Gemini 3.7 Flash, GLM 5.3, dan Qwen3.8 Max, ditambah pembaruan reasoning, harga, dan cakupan provider DeepSeek V4.
-- **Hubungkan ke NanoGPT** — gunakan penemuan model dinamis, streaming dan penyelesaian prompt, serta preferensi routing untuk kecepatan, harga, latensi, throughput, dukungan tool, dan caching.
-- **Provider dan task yang lebih andal** — perbaikan meningkatkan pengaturan endpoint Azure OpenAI, batas output Kimi Code, penyimpanan judul riwayat task, serta impor/ekspor pengaturan Zoo.
-- Destructive Command Guard kini mendukung Mac berbasis Intel.
-- Pembaruan keamanan mengatasi kerentanan di `undici` dan Mermaid.
+- **Akses file yang lebih aman** — gunakan allowlist Baca dan Tulis yang terpisah untuk mengontrol path yang dapat diakses Zoo Code.
+- **Lebih banyak pilihan model** — GLM 5.3 kini tersedia melalui Z.ai, serta Gemini 3.5 Flash Lite dan Gemini 3.1 Flash Lite telah ditambahkan ke provider Gemini.
+- **Workflow sehari-hari yang lebih andal** — peningkatan memperkuat pemulihan dan riwayat task, streaming dan penggunaan tool provider, impor pengaturan, serta keterbacaan di berbagai tema IDE.
 
 ## Apa yang Bisa Zoo Code Lakukan Untuk ANDA?
 
diff --git a/locales/it/README.md b/locales/it/README.md
index 7091aaca0c..ee0ba44d97 100644
--- a/locales/it/README.md
+++ b/locales/it/README.md
@@ -47,13 +47,11 @@ Zoo Code parte dalle fondamenta create da Roo Code e continua ad ampliarle con:
 - **Workflow di terminale e modifica più affidabili** — correzioni per il completamento prematuro del terminale, le race condition dello stato delle attività, la gestione del contesto, la modifica dei diff e l'uso di strumenti specifici dei provider.
 - **Più controllo sul tuo workspace** — gestione delle regole, restrizioni MCP per modalità, controlli dei percorsi multi-root, opzioni di reasoning dei modelli e azioni per esaminare le modifiche al completamento.
 
-## Novità in v3.78.0
+## Novità in v3.80.0
 
-- **Sono arrivati tre importanti nuovi modelli** — usa i nuovissimi Gemini 3.7 Flash, GLM 5.3 e Qwen3.8 Max, oltre agli aggiornamenti di reasoning, prezzi e copertura dei provider per DeepSeek V4.
-- **Connettiti a NanoGPT** — usa la scoperta dinamica dei modelli, streaming e completamenti dei prompt, con preferenze di routing per velocità, prezzo, latenza, throughput, supporto agli strumenti e caching.
-- **Provider e task più affidabili** — le correzioni migliorano la configurazione degli endpoint Azure OpenAI, i limiti di output di Kimi Code, la conservazione dei titoli nella cronologia dei task e l'importazione/esportazione delle impostazioni Zoo.
-- Destructive Command Guard ora supporta i Mac basati su Intel.
-- Gli aggiornamenti di sicurezza risolvono vulnerabilità in `undici` e Mermaid.
+- **Accesso ai file più sicuro** — usa allowlist separate per lettura e scrittura per controllare i percorsi a cui Zoo Code può accedere.
+- **Più modelli tra cui scegliere** — GLM 5.3 è ora disponibile tramite Z.ai, mentre Gemini 3.5 Flash Lite e Gemini 3.1 Flash Lite sono stati aggiunti al provider Gemini.
+- **Workflow quotidiani più affidabili** — i miglioramenti rafforzano il recupero e la cronologia dei task, lo streaming e l'uso degli strumenti dei provider, l'importazione delle impostazioni e la leggibilità nei vari temi dell'IDE.
 
 ## Cosa può fare Zoo Code per TE?
 
diff --git a/locales/ja/README.md b/locales/ja/README.md
index bda6e078c2..cff363b907 100644
--- a/locales/ja/README.md
+++ b/locales/ja/README.md
@@ -46,13 +46,11 @@ Zoo Code は Roo Code が築いた基盤を引き継ぎ、次の機能で拡張
 - **より信頼性の高いターミナルと編集ワークフロー** — ターミナルの早期完了、タスク状態の競合、コンテキスト管理、diff 編集、プロバイダー固有のツール利用に関する問題を修正。
 - **ワークスペースをより細かく制御** — ルール管理、モードごとの MCP 制限、マルチルートのパス制御、モデルの reasoning オプション、完了時の変更レビュー操作を追加。
 
-## v3.78.0 の新機能
+## v3.80.0 の新機能
 
-- **注目の新モデルが3つ登場** — 最新の Gemini 3.7 Flash、GLM 5.3、Qwen3.8 Max に加え、DeepSeek V4 の reasoning、価格、プロバイダー対応も更新されました。
-- **NanoGPT に接続** — 動的なモデル検出、ストリーミング、Prompt 補完に加え、速度、価格、レイテンシ、スループット、ツール対応、キャッシュに基づくルーティング設定を利用できます。
-- **プロバイダーとタスクの信頼性を向上** — Azure OpenAI エンドポイント設定、Kimi Code の出力上限、タスク履歴タイトルの保持、Zoo 設定のインポート/エクスポートを改善しました。
-- Destructive Command Guard が Intel 搭載 Mac に対応しました。
-- セキュリティアップデートにより `undici` と Mermaid の脆弱性を修正しました。
+- **より安全なファイルアクセス** — 読み取り用と書き込み用の許可リストを分けて、Zoo Code がアクセスできるパスを制御できます。
+- **モデルの選択肢がさらに充実** — GLM 5.3 が Z.ai で利用可能になり、Gemini プロバイダーに Gemini 3.5 Flash Lite と Gemini 3.1 Flash Lite が追加されました。
+- **日常のワークフローをより確実に** — タスクの復元と履歴、プロバイダーのストリーミングとツール使用、設定のインポート、IDE テーマ全体での読みやすさを改善しました。
 
 ## Zoo Codeがあなたのためにできること
 
diff --git a/locales/ko/README.md b/locales/ko/README.md
index 31d30f02c5..f9955cf889 100644
--- a/locales/ko/README.md
+++ b/locales/ko/README.md
@@ -45,13 +45,11 @@ Zoo Code는 Roo Code가 만든 기반 위에서 다음 기능을 더하며 계
 - **더 안정적인 터미널 및 편집 워크플로우** — 터미널 조기 완료, 작업 상태 경합, 컨텍스트 관리, diff 편집, 프로바이더별 도구 사용 문제를 수정.
 - **워크스페이스를 더 세밀하게 제어** — 규칙 관리, 모드별 MCP 제한, 멀티 루트 경로 제어, 모델 reasoning 옵션, 완료 시 변경 사항 검토 작업을 제공.
 
-## v3.78.0의 새로운 기능
+## v3.80.0의 새로운 기능
 
-- **주요 신규 모델 3종이 출시되었습니다** — 완전히 새로운 Gemini 3.7 Flash, GLM 5.3, Qwen3.8 Max 모델과 업데이트된 DeepSeek V4 reasoning, 가격 및 프로바이더 지원을 사용해 보세요.
-- **NanoGPT에 연결하세요** — 동적 모델 검색, 스트리밍 및 prompt completion과 함께 속도, 가격, 지연 시간, 처리량, 도구 지원, 캐싱을 기준으로 한 라우팅 설정을 사용할 수 있습니다.
-- **더 안정적인 프로바이더와 작업** — Azure OpenAI endpoint 설정, Kimi Code 출력 제한, 작업 기록 제목 보존, Zoo 설정 가져오기/내보내기가 개선되었습니다.
-- Destructive Command Guard가 이제 Intel 기반 Mac을 지원합니다.
-- 보안 업데이트로 `undici`와 Mermaid의 취약점을 해결했습니다.
+- **더 안전한 파일 접근** — 읽기와 쓰기 허용 목록을 분리해 Zoo Code가 접근할 수 있는 경로를 제어하세요.
+- **더 다양한 모델 선택** — 이제 Z.ai에서 GLM 5.3을 사용할 수 있으며 Gemini provider에 Gemini 3.5 Flash Lite와 Gemini 3.1 Flash Lite가 추가되었습니다.
+- **더 안정적인 일상 workflow** — task 복구와 기록, provider 스트리밍과 tool 사용, 설정 가져오기, 다양한 IDE 테마에서의 가독성이 개선되었습니다.
 
 ## Zoo Code가 당신을 위해 무엇을 할 수 있을까요?
 
diff --git a/locales/nl/README.md b/locales/nl/README.md
index 8af72a0103..b2d695a668 100644
--- a/locales/nl/README.md
+++ b/locales/nl/README.md
@@ -47,13 +47,11 @@ Zoo Code bouwt voort op het fundament van Roo Code en breidt dit verder uit met:
 - **Betrouwbaardere terminal- en bewerkingsworkflows** — oplossingen voor voortijdige terminalvoltooiing, race conditions in taakstatussen, contextbeheer, diff-bewerking en providerspecifiek toolgebruik.
 - **Meer controle over je workspace** — regelbeheer, MCP-beperkingen per modus, padbeheer voor multi-root-workspaces, reasoning-opties voor modellen en acties om wijzigingen bij voltooiing te beoordelen.
 
-## Nieuw in v3.78.0
+## Nieuw in v3.80.0
 
-- **Drie belangrijke nieuwe modellen zijn gearriveerd** — gebruik de gloednieuwe Gemini 3.7 Flash-, GLM 5.3- en Qwen3.8 Max-modellen, plus bijgewerkte reasoning, prijzen en providerondersteuning voor DeepSeek V4.
-- **Maak verbinding met NanoGPT** — gebruik dynamische modeldetectie, streaming en promptaanvulling, plus routeringsvoorkeuren voor snelheid, prijs, latentie, doorvoer, toolondersteuning en caching.
-- **Betrouwbaardere providers en taken** — oplossingen verbeteren de instelling van Azure OpenAI-endpoints, Kimi Code-uitvoerlimieten, het behouden van titels in de taakgeschiedenis en het importeren/exporteren van Zoo-instellingen.
-- Destructive Command Guard ondersteunt nu Intel-gebaseerde Macs.
-- Beveiligingsupdates verhelpen kwetsbaarheden in `undici` en Mermaid.
+- **Veiligere bestandstoegang** — gebruik afzonderlijke toelatingslijsten voor lezen en schrijven om te bepalen tot welke paden Zoo Code toegang heeft.
+- **Meer modelkeuze** — GLM 5.3 is nu beschikbaar via Z.ai en Gemini 3.5 Flash Lite en Gemini 3.1 Flash Lite zijn toegevoegd aan de Gemini-provider.
+- **Betrouwbaardere dagelijkse workflows** — verbeteringen versterken taakherstel en -geschiedenis, streaming en toolgebruik van providers, het importeren van instellingen en de leesbaarheid in verschillende IDE-thema's.
 
 ## Wat kan Zoo Code voor JOU doen?
 
diff --git a/locales/pl/README.md b/locales/pl/README.md
index 7a6133d4e9..4814894699 100644
--- a/locales/pl/README.md
+++ b/locales/pl/README.md
@@ -45,13 +45,11 @@ Zoo Code rozwija fundament stworzony przez Roo Code i stale rozszerza go o:
 - **Bardziej niezawodne workflow terminala i edycji** — poprawki przedwczesnego kończenia poleceń terminala, race condition stanu zadań, zarządzania kontekstem, edycji diff i użycia narzędzi właściwych dla providerów.
 - **Większą kontrolę nad workspace** — zarządzanie regułami, ograniczenia MCP dla poszczególnych trybów, kontrolę ścieżek multi-root, opcje reasoning modeli i akcje przeglądu zmian po ukończeniu.
 
-## Nowości w v3.78.0
+## Nowości w v3.80.0
 
-- **Pojawiły się trzy ważne nowe modele** — korzystaj z zupełnie nowych Gemini 3.7 Flash, GLM 5.3 i Qwen3.8 Max oraz zaktualizowanego reasoning, cen i obsługi providerów dla DeepSeek V4.
-- **Połącz się z NanoGPT** — korzystaj z dynamicznego wykrywania modeli, streamingu i uzupełniania promptów oraz preferencji routingu według szybkości, ceny, opóźnienia, przepustowości, obsługi narzędzi i cache.
-- **Bardziej niezawodni providerzy i zadania** — poprawki usprawniają konfigurację endpointów Azure OpenAI, limity wyjścia Kimi Code, zachowywanie tytułów historii zadań oraz import/eksport ustawień Zoo.
-- Destructive Command Guard obsługuje teraz komputery Mac z procesorami Intel.
-- Aktualizacje zabezpieczeń usuwają luki w `undici` i Mermaid.
+- **Bezpieczniejszy dostęp do plików** — używaj oddzielnych list dozwolonych ścieżek odczytu i zapisu, aby kontrolować, do których ścieżek Zoo Code ma dostęp.
+- **Większy wybór modeli** — GLM 5.3 jest teraz dostępny przez Z.ai, a Gemini 3.5 Flash Lite i Gemini 3.1 Flash Lite dodano do providera Gemini.
+- **Bardziej niezawodne codzienne workflow** — ulepszenia wzmacniają odzyskiwanie zadań i historię, streaming i użycie narzędzi przez providerów, import ustawień oraz czytelność w różnych motywach IDE.
 
 ## Co Zoo Code może zrobić dla CIEBIE?
 
diff --git a/locales/pt-BR/README.md b/locales/pt-BR/README.md
index dfdd18c4e8..75f66222e1 100644
--- a/locales/pt-BR/README.md
+++ b/locales/pt-BR/README.md
@@ -47,13 +47,11 @@ O Zoo Code aproveita a base criada pelo Roo Code e continua ampliando-a com:
 - **Workflows de terminal e edição mais confiáveis** — correções para encerramento prematuro do terminal, race conditions no estado das tarefas, gerenciamento de contexto, edição de diff e uso de ferramentas específicas de cada provider.
 - **Mais controle sobre seu workspace** — gerenciamento de regras, restrições de MCP por modo, controles de caminhos multi-root, opções de reasoning dos modelos e ações para revisar alterações ao concluir uma tarefa.
 
-## Novidades na v3.78.0
+## Novidades na v3.80.0
 
-- **Três novos modelos importantes chegaram** — use os novíssimos Gemini 3.7 Flash, GLM 5.3 e Qwen3.8 Max, além das atualizações de reasoning, preços e cobertura de providers do DeepSeek V4.
-- **Conecte-se ao NanoGPT** — use descoberta dinâmica de modelos, streaming e conclusão de prompts, com preferências de roteamento por velocidade, preço, latência, throughput, suporte a ferramentas e cache.
-- **Providers e tarefas mais confiáveis** — as correções melhoram a configuração de endpoints do Azure OpenAI, os limites de saída do Kimi Code, a preservação de títulos no histórico de tarefas e a importação/exportação das configurações do Zoo.
-- O Destructive Command Guard agora oferece suporte a Macs com processadores Intel.
-- Atualizações de segurança corrigem vulnerabilidades no `undici` e no Mermaid.
+- **Acesso a arquivos mais seguro** — use listas de permissões separadas para leitura e gravação a fim de controlar quais caminhos o Zoo Code pode acessar.
+- **Mais opções de modelos** — o GLM 5.3 agora está disponível pela Z.ai, e o Gemini 3.5 Flash Lite e o Gemini 3.1 Flash Lite foram adicionados ao provider Gemini.
+- **Workflows cotidianos mais confiáveis** — as melhorias reforçam a recuperação e o histórico de tarefas, o streaming e o uso de ferramentas dos providers, a importação de configurações e a legibilidade em diferentes temas da IDE.
 
 ## O que o Zoo Code pode fazer por VOCÊ?
 
diff --git a/locales/ru/README.md b/locales/ru/README.md
index e5e6d66198..8a42d21dc2 100644
--- a/locales/ru/README.md
+++ b/locales/ru/README.md
@@ -47,13 +47,11 @@ Zoo Code развивает основу, созданную Roo Code, и про
 - **Более надёжные рабочие процессы в терминале и редакторе** — исправления преждевременного завершения команд, гонок состояния задач, управления контекстом, редактирования diff и использования инструментов отдельных провайдеров.
 - **Больше контроля над рабочей областью** — управление правилами, ограничения MCP для отдельных режимов, управление путями в многоуровневых рабочих областях, параметры reasoning моделей и действия для проверки изменений после завершения.
 
-## Что нового в v3.78.0
+## Что нового в v3.80.0
 
-- **Появились три важных новых модели** — используй совершенно новые Gemini 3.7 Flash, GLM 5.3 и Qwen3.8 Max, а также обновлённые reasoning, цены и поддержку провайдеров для DeepSeek V4.
-- **Подключайся к NanoGPT** — используй динамическое обнаружение моделей, streaming и дополнение Prompt, а также настройки маршрутизации по скорости, цене, задержке, пропускной способности, поддержке инструментов и кешированию.
-- **Более надёжные провайдеры и задачи** — исправления улучшают настройку endpoint Azure OpenAI, лимиты вывода Kimi Code, сохранение заголовков истории задач и импорт/экспорт настроек Zoo.
-- Destructive Command Guard теперь поддерживает Mac на базе Intel.
-- Обновления безопасности устраняют уязвимости в `undici` и Mermaid.
+- **Более безопасный доступ к файлам** — используйте отдельные списки разрешённых путей для чтения и записи, чтобы контролировать, к каким путям может обращаться Zoo Code.
+- **Больше моделей на выбор** — GLM 5.3 теперь доступна через Z.ai, а Gemini 3.5 Flash Lite и Gemini 3.1 Flash Lite добавлены в провайдер Gemini.
+- **Более надёжные повседневные процессы** — улучшены восстановление и история задач, streaming и использование инструментов провайдерами, импорт настроек и читаемость в разных темах IDE.
 
 ## Что Zoo Code может сделать для ВАС?
 
diff --git a/locales/tr/README.md b/locales/tr/README.md
index 5f90729546..ce76b1a9d7 100644
--- a/locales/tr/README.md
+++ b/locales/tr/README.md
@@ -47,13 +47,11 @@ Zoo Code, Roo Code'un oluşturduğu temel üzerine inşa ediliyor ve bu temeli 
 - **Daha güvenilir terminal ve düzenleme iş akışları** — terminalin erken tamamlanması, görev durumu yarış koşulları, bağlam yönetimi, diff düzenleme ve sağlayıcıya özel araç kullanımı için düzeltmeler.
 - **Çalışma alanın üzerinde daha fazla kontrol** — kural yönetimi, mod başına MCP kısıtlamaları, çok köklü yol denetimleri, model reasoning seçenekleri ve tamamlanan değişiklikleri inceleme eylemleri.
 
-## v3.78.0'daki Yenilikler
+## v3.80.0'daki Yenilikler
 
-- **Üç önemli yeni model geldi** — yepyeni Gemini 3.7 Flash, GLM 5.3 ve Qwen3.8 Max modellerini, ayrıca güncellenmiş DeepSeek V4 reasoning, fiyatlandırma ve sağlayıcı kapsamını kullan.
-- **NanoGPT'ye bağlan** — dinamik model keşfi, streaming ve Prompt tamamlama ile hız, fiyat, gecikme, throughput, araç desteği ve caching için yönlendirme tercihlerini kullan.
-- **Daha güvenilir sağlayıcılar ve görevler** — düzeltmeler Azure OpenAI endpoint kurulumunu, Kimi Code çıktı sınırlarını, görev geçmişi başlıklarının korunmasını ve Zoo ayarlarının içe/dışa aktarımını iyileştiriyor.
-- Destructive Command Guard artık Intel tabanlı Mac'leri destekliyor.
-- Güvenlik güncellemeleri `undici` ve Mermaid'deki güvenlik açıklarını gideriyor.
+- **Daha güvenli dosya erişimi** — Zoo Code'un hangi yollara erişebileceğini kontrol etmek için ayrı Okuma ve Yazma izin listeleri kullan.
+- **Daha fazla model seçeneği** — GLM 5.3 artık Z.ai üzerinden kullanılabilir; Gemini 3.5 Flash Lite ve Gemini 3.1 Flash Lite da Gemini provider'ına eklendi.
+- **Daha güvenilir günlük iş akışları** — iyileştirmeler görev kurtarma ve geçmişini, provider streaming ve araç kullanımını, ayar içe aktarmayı ve IDE temalarındaki okunabilirliği güçlendiriyor.
 
 ## Zoo Code SİZİN İçin Ne Yapabilir?
 
diff --git a/locales/vi/README.md b/locales/vi/README.md
index f2a95db23d..d44edd0ce2 100644
--- a/locales/vi/README.md
+++ b/locales/vi/README.md
@@ -46,13 +46,11 @@ Zoo Code phát triển trên nền tảng do Roo Code tạo ra và tiếp tục
 - **Workflow terminal và chỉnh sửa đáng tin cậy hơn** — sửa lỗi terminal hoàn tất quá sớm, xung đột trạng thái tác vụ, quản lý ngữ cảnh, chỉnh sửa diff và sử dụng công cụ riêng của từng provider.
 - **Kiểm soát workspace tốt hơn** — quản lý quy tắc, giới hạn MCP theo từng chế độ, kiểm soát đường dẫn multi-root, tùy chọn reasoning của model và thao tác xem lại thay đổi khi hoàn tất.
 
-## Điểm mới trong v3.78.0
+## Điểm mới trong v3.80.0
 
-- **Ba model mới quan trọng đã xuất hiện** — sử dụng các model hoàn toàn mới Gemini 3.7 Flash, GLM 5.3 và Qwen3.8 Max, cùng reasoning, giá và phạm vi provider được cập nhật cho DeepSeek V4.
-- **Kết nối với NanoGPT** — sử dụng khám phá model động, streaming và hoàn thành Prompt, cùng tùy chọn định tuyến theo tốc độ, giá, độ trễ, throughput, hỗ trợ tool và caching.
-- **Provider và task đáng tin cậy hơn** — các bản sửa lỗi cải thiện thiết lập endpoint Azure OpenAI, giới hạn đầu ra Kimi Code, giữ nguyên tiêu đề lịch sử task và nhập/xuất cài đặt Zoo.
-- Destructive Command Guard hiện hỗ trợ máy Mac dùng chip Intel.
-- Các bản cập nhật bảo mật khắc phục lỗ hổng trong `undici` và Mermaid.
+- **Truy cập file an toàn hơn** — sử dụng allowlist Đọc và Ghi riêng biệt để kiểm soát các path mà Zoo Code có thể truy cập.
+- **Nhiều lựa chọn model hơn** — GLM 5.3 hiện có qua Z.ai, đồng thời Gemini 3.5 Flash Lite và Gemini 3.1 Flash Lite đã được thêm vào provider Gemini.
+- **Workflow hằng ngày đáng tin cậy hơn** — các cải tiến tăng cường khôi phục và lịch sử task, streaming và sử dụng tool của provider, nhập cài đặt và khả năng đọc trên các theme IDE.
 
 ## Zoo Code có thể làm gì cho BẠN?
 
diff --git a/locales/zh-CN/README.md b/locales/zh-CN/README.md
index 4fba2e03de..5a1118a3f5 100644
--- a/locales/zh-CN/README.md
+++ b/locales/zh-CN/README.md
@@ -43,13 +43,11 @@ Zoo Code 基于 Roo Code 打下的基础持续扩展，新增了：
 - **更可靠的终端和编辑工作流** — 修复终端过早完成、任务状态竞争、上下文管理、差异更新编辑和提供商专用工具调用等问题。
 - **更全面的工作区控制** — 支持规则管理、按模式限制 MCP、多根工作区路径控制、模型推理选项和完成后的变更审查操作。
 
-## v3.78.0 新增内容
+## v3.80.0 新增内容
 
-- **三款重磅新模型现已推出** — 使用全新的 Gemini 3.7 Flash、GLM 5.3 和 Qwen3.8 Max 模型，以及更新后的 DeepSeek V4 推理、定价和提供商覆盖。
-- **连接 NanoGPT** — 使用动态模型发现、流式传输和 Prompt 补全，并按速度、价格、延迟、吞吐量、工具支持和缓存设置路由偏好。
-- **更可靠的提供商和任务** — 修复改进了 Azure OpenAI endpoint 设置、Kimi Code 输出限制、任务历史标题保留以及 Zoo 设置的导入/导出。
-- Destructive Command Guard 现在支持基于 Intel 的 Mac。
-- 安全更新修复了 `undici` 和 Mermaid 中的漏洞。
+- **更安全的文件访问** — 使用独立的读取和写入允许列表，控制 Zoo Code 可以访问的路径。
+- **更多模型选择** — GLM 5.3 现已通过 Z.ai 提供，Gemini provider 也新增了 Gemini 3.5 Flash Lite 和 Gemini 3.1 Flash Lite。
+- **更可靠的日常工作流** — 改进增强了任务恢复与历史记录、provider 流式传输与工具使用、设置导入，以及不同 IDE 主题下的可读性。
 
 ## Zoo Code 能为您做什么？
 
diff --git a/locales/zh-TW/README.md b/locales/zh-TW/README.md
index 215a3886b6..c772d6d46c 100644
--- a/locales/zh-TW/README.md
+++ b/locales/zh-TW/README.md
@@ -43,13 +43,11 @@ Zoo Code 以 Roo Code 建立的基礎持續擴充，新增了：
 - **更可靠的終端機與編輯工作流程** — 修正終端機過早完成、工作狀態競爭、上下文管理、差異更新編輯和供應商專用工具使用等問題。
 - **更完整的工作區控制** — 支援規則管理、依模式限制 MCP、多根工作區路徑控制、模型推理選項，以及完成後的變更檢閱操作。
 
-## v3.78.0 新功能
+## v3.80.0 新功能
 
-- **三款重磅新模型現已推出** — 使用全新的 Gemini 3.7 Flash、GLM 5.3 和 Qwen3.8 Max 模型，以及更新後的 DeepSeek V4 推理、定價與供應商支援。
-- **連接 NanoGPT** — 使用動態模型探索、串流與 Prompt 補全，並依速度、價格、延遲、吞吐量、工具支援和快取設定路由偏好。
-- **更可靠的供應商與任務** — 修正改善了 Azure OpenAI endpoint 設定、Kimi Code 輸出限制、任務歷史標題保留，以及 Zoo 設定的匯入/匯出。
-- Destructive Command Guard 現在支援 Intel 架構的 Mac。
-- 安全性更新修正了 `undici` 與 Mermaid 中的漏洞。
+- **更安全的檔案存取** — 使用獨立的讀取與寫入允許清單，控制 Zoo Code 可以存取的路徑。
+- **更多模型選擇** — GLM 5.3 現已透過 Z.ai 提供，Gemini provider 也新增了 Gemini 3.5 Flash Lite 和 Gemini 3.1 Flash Lite。
+- **更可靠的日常工作流程** — 改善強化了任務復原與歷史記錄、provider 串流與工具使用、設定匯入，以及不同 IDE 主題下的可讀性。
 
 ## Zoo Code 能為您做什麼？
 
diff --git a/src/CHANGELOG.md b/src/CHANGELOG.md
index 72bfdbd415..cc9721773a 100644
--- a/src/CHANGELOG.md
+++ b/src/CHANGELOG.md
@@ -1,5 +1,35 @@
 # Zoo Code Changelog
 
+## [3.80.0]
+
+### Minor Changes
+
+- Add separate Read and Write allowlists so users can restrict file operations to approved paths (#1273 by @nh2, PR #1274 by @nh2)
+- Add official GLM 5.3 support to the Z.ai API (PR #1284 by @app/zoomote)
+- Add Gemini 3.5 Flash Lite and Gemini 3.1 Flash Lite to the Gemini API provider (#1251 by @MrMody0, PR #1334 by @easonLiangWorldedtech)
+- Prevent webview gray screens and out-of-memory failures by throttling extension-state pushes (#629 by @edelauna, PR #1078 by @JunyongParkDev)
+- Recover stale delegated child tasks after an extension restart (#1100 by @aniruddhabag3, PR #1210 by @edelauna)
+- Replace the shared task-history index with atomic per-task merges to prevent lost or corrupted history (#1231 by @martin-rueegg, PR #1261 by @edelauna)
+- Mark interrupted tool calls as errors in persisted task history so resumed conversations remain consistent (#1283 by @martin-rueegg, PR #1323 by @easonLiangWorldedtech)
+- Complete OpenAI Codex prompts over the streaming transport instead of waiting indefinitely (#1242 by @Rafael-Silva-Oliveira, PR #1243 by @Rafael-Silva-Oliveira)
+- Stream LM Studio `reasoning_content` so model reasoning is preserved in responses (#1075 by @marunguy, PR #1175 by @daewoongoh)
+- Fix NanoGPT Muse Spark tool-use requests (PR #1310 by @app/zoomote)
+- Make settings import await provider-state persistence before reporting success (PR #1340 by @app/zoomote)
+- Default telemetry to opt-out and require explicit consent through the telemetry UI (PR #1069 by @edelauna)
+- Introduce an SDK-agnostic request configuration builder for consistent abort-signal support (#1007 by @easonLiangWorldedtech, PR #1008 by @easonLiangWorldedtech)
+- Delegate VS Code shell-profile reads to Terminal methods for more reliable profile detection (#634 by @marty-a11y, PR #1033 by @edelauna)
+- Increase the Semble archive download limit for larger codebase-intelligence binaries (#1305 by @navedmerchant, PR #1306 by @navedmerchant)
+- Stop estimating Zoo Gateway UI costs from unrelated default-model prices when authoritative cost data is unavailable (PR #1339 by @JamesRobert20)
+- Improve Mermaid label contrast in light IDE themes (PR #1285 by @app/zoomote)
+- Keep chat controls legible in light IDE themes (PR #1298 by @app/zoomote)
+- Fix the remaining chat controls that faded into light IDE themes (PR #1312 by @app/zoomote)
+- Improve shared-control contrast across IDE themes (PR #1333 by @app/zoomote)
+- Canonicalize provider and model identifiers in webview messages (PR #1143 by @WebMad)
+- Canonicalize model-router calls across providers (PR #1165 by @WebMad)
+- Enforce canonical provider identifiers through provider lint rules (#1294 by @WebMad, PR #1297 by @WebMad)
+- Enforce safe asynchronous handling in core tools (PR #1255 by @app/zoomote)
+- Build downloadable VSIX artifacts for pull requests (PR #1314 by @app/zoomote)
+
 ## [3.78.0]
 
 ### Minor Changes
diff --git a/src/core/webview/ClineProvider.ts b/src/core/webview/ClineProvider.ts
index 94ce829ac2..4621cb3fc4 100644
--- a/src/core/webview/ClineProvider.ts
+++ b/src/core/webview/ClineProvider.ts
@@ -303,7 +303,7 @@ export class ClineProvider
 
 	public isViewLaunched = false
 	public settingsImportedAt?: number
-	public readonly latestAnnouncementId = "aug-2026-v3.78.0-models-nanogpt-reliability" // v3.78.0 new models, NanoGPT, and provider/task reliability
+	public readonly latestAnnouncementId = "aug-2026-v3.80.0-allowlists-models-reliability" // v3.80.0 file allowlists, models, and workflow reliability
 	public readonly providerSettingsManager: ProviderSettingsManager
 	public readonly customModesManager: CustomModesManager
 
diff --git a/src/package.json b/src/package.json
index 55035dc206..57286d6330 100644
--- a/src/package.json
+++ b/src/package.json
@@ -3,7 +3,7 @@
 	"displayName": "%extension.displayName%",
 	"description": "%extension.description%",
 	"publisher": "ZooCodeOrganization",
-	"version": "3.78.0",
+	"version": "3.80.0",
 	"icon": "assets/icons/icon.png",
 	"galleryBanner": {
 		"color": "#617A91",
@@ -466,7 +466,7 @@
 		"@lmstudio/sdk": "^1.1.1",
 		"@mistralai/mistralai": "^1.9.18",
 		"@modelcontextprotocol/sdk": "1.29.0",
-		"@qdrant/js-client-rest": "^1.14.0",
+		"@qdrant/js-client-rest": "^1.18.0",
 		"@roo-code/cloud": "workspace:^",
 		"@roo-code/core": "workspace:^",
 		"@roo-code/ipc": "workspace:^",
diff --git a/src/services/code-index/vector-store/qdrant-client.ts b/src/services/code-index/vector-store/qdrant-client.ts
index 99e6b33deb..801230bafc 100644
--- a/src/services/code-index/vector-store/qdrant-client.ts
+++ b/src/services/code-index/vector-store/qdrant-client.ts
@@ -61,6 +61,7 @@ export class QdrantVectorStore implements IVectorStore {
 				port: port,
 				prefix: urlObj.pathname === "/" ? undefined : urlObj.pathname.replace(/\/+$/, ""),
 				apiKey,
+				checkCompatibility: false,
 				headers: {
 					"User-Agent": "Zoo-Code",
 				},
@@ -71,6 +72,7 @@ export class QdrantVectorStore implements IVectorStore {
 			this.client = new QdrantClient({
 				url: parsedUrl,
 				apiKey,
+				checkCompatibility: false,
 				headers: {
 					"User-Agent": "Zoo-Code",
 				},
diff --git a/webview-ui/src/components/chat/Announcement.tsx b/webview-ui/src/components/chat/Announcement.tsx
index 0f05aa867a..6b1682bdfa 100644
--- a/webview-ui/src/components/chat/Announcement.tsx
+++ b/webview-ui/src/components/chat/Announcement.tsx
@@ -20,6 +20,7 @@ interface AnnouncementProps {
  * announcement will render. As soon as an announcement is shown, the id will be
  * updated in state. This ensures that announcements are not shown more than
  * once, even if the user doesn't close it themselves.
+ * Release-specific copy is maintained in each locale's `chat.json` announcement keys.
  */
 
 const Announcement = ({ hideAnnouncement }: AnnouncementProps) => {
diff --git a/webview-ui/src/i18n/locales/ca/chat.json b/webview-ui/src/i18n/locales/ca/chat.json
index d5b63ea886..203d54f6ae 100644
--- a/webview-ui/src/i18n/locales/ca/chat.json
+++ b/webview-ui/src/i18n/locales/ca/chat.json
@@ -344,9 +344,9 @@
 		"support": "Dona suport a Zoo Code fent-nos una estrella a <githubLink>GitHub</githubLink>.",
 		"release": {
 			"heading": "Novetats:",
-			"highlight1": "Han arribat tres grans models nous — utilitza els nous Gemini 3.7 Flash, GLM 5.3 i Qwen3.8 Max, a més de millores en el raonament, els preus i la cobertura de proveïdors de DeepSeek V4.",
-			"highlight2": "Connecta't a NanoGPT — utilitza el descobriment dinàmic de models, streaming i completions de prompts, i preferències d'encaminament per velocitat, preu, latència, rendiment, compatibilitat amb eines i memòria cau.",
-			"highlight3": "Proveïdors i tasques més fiables — les correccions milloren la configuració dels endpoints d'Azure OpenAI, els límits de sortida de Kimi Code, la conservació dels títols de l'historial de tasques i la importació/exportació de configuració de Zoo."
+			"highlight1": "Accés als fitxers més segur — utilitza llistes de permisos de lectura i escriptura separades per controlar a quines rutes pot accedir Zoo Code.",
+			"highlight2": "Més models per triar — GLM 5.3 ja està disponible mitjançant Z.ai, i Gemini 3.5 Flash Lite i Gemini 3.1 Flash Lite s'han afegit al proveïdor Gemini.",
+			"highlight3": "Fluxos de treball quotidians més fiables — les millores reforcen la recuperació i l'historial de tasques, l'streaming i l'ús d'eines dels proveïdors, la importació de configuració i la llegibilitat en diferents temes de l'IDE."
 		},
 		"cloudAgents": {
 			"heading": "Novetats al núvol:",
diff --git a/webview-ui/src/i18n/locales/de/chat.json b/webview-ui/src/i18n/locales/de/chat.json
index ad94856234..590914a9ee 100644
--- a/webview-ui/src/i18n/locales/de/chat.json
+++ b/webview-ui/src/i18n/locales/de/chat.json
@@ -344,9 +344,9 @@
 		"support": "Bitte unterstütze Zoo Code, indem du uns auf <githubLink>GitHub</githubLink> einen Stern gibst.",
 		"release": {
 			"heading": "Was ist neu:",
-			"highlight1": "Drei bedeutende neue Modelle sind da — nutze die brandneuen Modelle Gemini 3.7 Flash, GLM 5.3 und Qwen3.8 Max sowie aktualisiertes Reasoning, Preise und Anbieterabdeckung für DeepSeek V4.",
-			"highlight2": "Verbinde dich mit NanoGPT — nutze dynamische Modellerkennung, Streaming und Prompt-Vervollständigung sowie Routing-Einstellungen für Geschwindigkeit, Preis, Latenz, Durchsatz, Tool-Unterstützung und Caching.",
-			"highlight3": "Zuverlässigere Anbieter und Aufgaben — Korrekturen verbessern die Einrichtung von Azure-OpenAI-Endpunkten, Kimi-Code-Ausgabelimits, die Beibehaltung von Titeln im Aufgabenverlauf sowie den Import und Export von Zoo-Einstellungen."
+			"highlight1": "Sicherer Dateizugriff — nutze getrennte Lese- und Schreib-Positivlisten, um zu steuern, auf welche Pfade Zoo Code zugreifen darf.",
+			"highlight2": "Mehr Modellauswahl — GLM 5.3 ist jetzt über Z.ai verfügbar; außerdem wurden Gemini 3.5 Flash Lite und Gemini 3.1 Flash Lite zum Gemini-Anbieter hinzugefügt.",
+			"highlight3": "Zuverlässigere tägliche Workflows — Verbesserungen stärken die Wiederherstellung und den Verlauf von Aufgaben, Streaming und Tool-Nutzung bei Anbietern, Einstellungsimporte und die Lesbarkeit in verschiedenen IDE-Themes."
 		},
 		"cloudAgents": {
 			"heading": "Neu in der Cloud:",
diff --git a/webview-ui/src/i18n/locales/en/chat.json b/webview-ui/src/i18n/locales/en/chat.json
index 89f6c2f488..1caacde55f 100644
--- a/webview-ui/src/i18n/locales/en/chat.json
+++ b/webview-ui/src/i18n/locales/en/chat.json
@@ -363,9 +363,9 @@
 		"support": "Please support Zoo Code by starring us on <githubLink>GitHub</githubLink>.",
 		"release": {
 			"heading": "What's New:",
-			"highlight1": "Three major new models have arrived — use the brand-new Gemini 3.7 Flash, GLM 5.3, and Qwen3.8 Max models, plus updated DeepSeek V4 reasoning, pricing, and provider coverage.",
-			"highlight2": "Connect to NanoGPT — use dynamic model discovery, streaming and prompt completions, and routing preferences for speed, price, latency, throughput, tool support, and caching.",
-			"highlight3": "More reliable providers and tasks — fixes improve Azure OpenAI endpoint setup, Kimi Code output limits, task-history title preservation, and Zoo settings import/export."
+			"highlight1": "Safer file access — use separate Read and Write allowlists to control which paths Zoo Code can access.",
+			"highlight2": "More model choice — GLM 5.3 is now available through Z.ai, with Gemini 3.5 Flash Lite and Gemini 3.1 Flash Lite added to the Gemini provider.",
+			"highlight3": "More reliable everyday workflows — improvements strengthen task recovery and history, provider streaming and tool use, settings imports, and readability across IDE themes."
 		},
 		"cloudAgents": {
 			"heading": "New in the Cloud:",
diff --git a/webview-ui/src/i18n/locales/es/chat.json b/webview-ui/src/i18n/locales/es/chat.json
index 7876845932..527d78aed5 100644
--- a/webview-ui/src/i18n/locales/es/chat.json
+++ b/webview-ui/src/i18n/locales/es/chat.json
@@ -344,9 +344,9 @@
 		"support": "Apoya a Zoo Code dándonos una estrella en <githubLink>GitHub</githubLink>.",
 		"release": {
 			"heading": "Novedades:",
-			"highlight1": "Han llegado tres importantes modelos nuevos — usa los flamantes Gemini 3.7 Flash, GLM 5.3 y Qwen3.8 Max, además de mejoras en el razonamiento, los precios y la cobertura de proveedores de DeepSeek V4.",
-			"highlight2": "Conéctate a NanoGPT — usa descubrimiento dinámico de modelos, streaming y completado de prompts, y preferencias de enrutamiento por velocidad, precio, latencia, rendimiento, compatibilidad con herramientas y caché.",
-			"highlight3": "Proveedores y tareas más fiables — las correcciones mejoran la configuración de endpoints de Azure OpenAI, los límites de salida de Kimi Code, la conservación de títulos del historial de tareas y la importación/exportación de ajustes de Zoo."
+			"highlight1": "Acceso a archivos más seguro — usa listas de permitidos separadas para lectura y escritura a fin de controlar a qué rutas puede acceder Zoo Code.",
+			"highlight2": "Más modelos para elegir — GLM 5.3 ya está disponible mediante Z.ai, y Gemini 3.5 Flash Lite y Gemini 3.1 Flash Lite se han añadido al proveedor de Gemini.",
+			"highlight3": "Flujos de trabajo diarios más fiables — las mejoras refuerzan la recuperación y el historial de tareas, el streaming y el uso de herramientas de los proveedores, la importación de ajustes y la legibilidad en distintos temas del IDE."
 		},
 		"cloudAgents": {
 			"heading": "Novedades en la Nube:",
diff --git a/webview-ui/src/i18n/locales/fr/chat.json b/webview-ui/src/i18n/locales/fr/chat.json
index 032d43ce27..638b2c0227 100644
--- a/webview-ui/src/i18n/locales/fr/chat.json
+++ b/webview-ui/src/i18n/locales/fr/chat.json
@@ -344,9 +344,9 @@
 		"support": "Soutiens Zoo Code en nous donnant une étoile sur <githubLink>GitHub</githubLink>.",
 		"release": {
 			"heading": "Nouveautés :",
-			"highlight1": "Trois nouveaux modèles majeurs sont arrivés — utilise les tout nouveaux Gemini 3.7 Flash, GLM 5.3 et Qwen3.8 Max, ainsi que les améliorations du raisonnement, des tarifs et de la couverture des providers de DeepSeek V4.",
-			"highlight2": "Connecte-toi à NanoGPT — profite de la découverte dynamique des modèles, du streaming et de la complétion des prompts, avec des préférences de routage pour la vitesse, le prix, la latence, le débit, la prise en charge des outils et le cache.",
-			"highlight3": "Providers et tâches plus fiables — les correctifs améliorent la configuration des endpoints Azure OpenAI, les limites de sortie de Kimi Code, la conservation des titres dans l'historique des tâches et l'import/export des paramètres Zoo."
+			"highlight1": "Accès aux fichiers plus sûr — utilise des listes d'autorisation distinctes pour la lecture et l'écriture afin de contrôler les chemins auxquels Zoo Code peut accéder.",
+			"highlight2": "Plus de modèles au choix — GLM 5.3 est désormais disponible via Z.ai, et Gemini 3.5 Flash Lite ainsi que Gemini 3.1 Flash Lite ont été ajoutés au provider Gemini.",
+			"highlight3": "Des workflows quotidiens plus fiables — les améliorations renforcent la récupération et l'historique des tâches, le streaming et l'utilisation des outils par les providers, l'import des paramètres et la lisibilité dans les différents thèmes d'IDE."
 		},
 		"cloudAgents": {
 			"heading": "Nouveautés dans le Cloud :",
diff --git a/webview-ui/src/i18n/locales/hi/chat.json b/webview-ui/src/i18n/locales/hi/chat.json
index 94a805f328..31270bc937 100644
--- a/webview-ui/src/i18n/locales/hi/chat.json
+++ b/webview-ui/src/i18n/locales/hi/chat.json
@@ -344,9 +344,9 @@
 		"support": "कृपया <githubLink>GitHub</githubLink> पर हमें स्टार देकर Zoo Code का समर्थन करें।",
 		"release": {
 			"heading": "नया क्या है:",
-			"highlight1": "तीन बड़े नए models आ गए हैं — बिल्कुल नए Gemini 3.7 Flash, GLM 5.3 और Qwen3.8 Max models का उपयोग करें, साथ ही DeepSeek V4 के updated reasoning, pricing और provider coverage का लाभ लें।",
-			"highlight2": "NanoGPT से connect करें — dynamic model discovery, streaming और prompt completions के साथ speed, price, latency, throughput, tool support और caching के लिए routing preferences का उपयोग करें।",
-			"highlight3": "अधिक reliable providers और tasks — fixes Azure OpenAI endpoint setup, Kimi Code output limits, task-history title preservation और Zoo settings import/export को बेहतर बनाते हैं।"
+			"highlight1": "ज़्यादा सुरक्षित file access — Zoo Code किन paths को access कर सकता है, इसे नियंत्रित करने के लिए अलग-अलग Read और Write allowlists का उपयोग करें।",
+			"highlight2": "Models के ज़्यादा विकल्प — GLM 5.3 अब Z.ai के माध्यम से उपलब्ध है, और Gemini provider में Gemini 3.5 Flash Lite और Gemini 3.1 Flash Lite जोड़े गए हैं।",
+			"highlight3": "रोज़मर्रा के ज़्यादा reliable workflows — सुधार task recovery और history, provider streaming और tool use, settings imports तथा IDE themes में readability को बेहतर बनाते हैं।"
 		},
 		"cloudAgents": {
 			"heading": "क्लाउड में नया:",
diff --git a/webview-ui/src/i18n/locales/id/chat.json b/webview-ui/src/i18n/locales/id/chat.json
index d58d80db00..3b11773652 100644
--- a/webview-ui/src/i18n/locales/id/chat.json
+++ b/webview-ui/src/i18n/locales/id/chat.json
@@ -373,9 +373,9 @@
 		"support": "Dukung Zoo Code dengan memberi kami bintang di <githubLink>GitHub</githubLink>.",
 		"release": {
 			"heading": "Yang Baru:",
-			"highlight1": "Tiga model baru utama telah hadir — gunakan model terbaru Gemini 3.7 Flash, GLM 5.3, dan Qwen3.8 Max, ditambah pembaruan reasoning, harga, dan cakupan provider DeepSeek V4.",
-			"highlight2": "Hubungkan ke NanoGPT — gunakan penemuan model dinamis, streaming dan penyelesaian prompt, serta preferensi routing untuk kecepatan, harga, latensi, throughput, dukungan tool, dan caching.",
-			"highlight3": "Provider dan task yang lebih andal — perbaikan meningkatkan pengaturan endpoint Azure OpenAI, batas output Kimi Code, penyimpanan judul riwayat task, serta impor/ekspor pengaturan Zoo."
+			"highlight1": "Akses file yang lebih aman — gunakan allowlist Baca dan Tulis yang terpisah untuk mengontrol path yang dapat diakses Zoo Code.",
+			"highlight2": "Lebih banyak pilihan model — GLM 5.3 kini tersedia melalui Z.ai, serta Gemini 3.5 Flash Lite dan Gemini 3.1 Flash Lite telah ditambahkan ke provider Gemini.",
+			"highlight3": "Workflow sehari-hari yang lebih andal — peningkatan memperkuat pemulihan dan riwayat task, streaming dan penggunaan tool provider, impor pengaturan, serta keterbacaan di berbagai tema IDE."
 		},
 		"cloudAgents": {
 			"heading": "Baru di Cloud:",
diff --git a/webview-ui/src/i18n/locales/it/chat.json b/webview-ui/src/i18n/locales/it/chat.json
index 7c4c657b35..f473b9e454 100644
--- a/webview-ui/src/i18n/locales/it/chat.json
+++ b/webview-ui/src/i18n/locales/it/chat.json
@@ -344,9 +344,9 @@
 		"support": "Sostieni Zoo Code lasciandoci una stella su <githubLink>GitHub</githubLink>.",
 		"release": {
 			"heading": "Novità:",
-			"highlight1": "Sono arrivati tre importanti nuovi modelli — usa i nuovissimi Gemini 3.7 Flash, GLM 5.3 e Qwen3.8 Max, oltre agli aggiornamenti di reasoning, prezzi e copertura dei provider per DeepSeek V4.",
-			"highlight2": "Connettiti a NanoGPT — usa la scoperta dinamica dei modelli, streaming e completamenti dei prompt, con preferenze di routing per velocità, prezzo, latenza, throughput, supporto agli strumenti e caching.",
-			"highlight3": "Provider e task più affidabili — le correzioni migliorano la configurazione degli endpoint Azure OpenAI, i limiti di output di Kimi Code, la conservazione dei titoli nella cronologia dei task e l'importazione/esportazione delle impostazioni Zoo."
+			"highlight1": "Accesso ai file più sicuro — usa allowlist separate per lettura e scrittura per controllare i percorsi a cui Zoo Code può accedere.",
+			"highlight2": "Più modelli tra cui scegliere — GLM 5.3 è ora disponibile tramite Z.ai, mentre Gemini 3.5 Flash Lite e Gemini 3.1 Flash Lite sono stati aggiunti al provider Gemini.",
+			"highlight3": "Workflow quotidiani più affidabili — i miglioramenti rafforzano il recupero e la cronologia dei task, lo streaming e l'uso degli strumenti dei provider, l'importazione delle impostazioni e la leggibilità nei vari temi dell'IDE."
 		},
 		"cloudAgents": {
 			"heading": "Novità nel Cloud:",
diff --git a/webview-ui/src/i18n/locales/ja/chat.json b/webview-ui/src/i18n/locales/ja/chat.json
index f2f17b3fa5..470e66bd65 100644
--- a/webview-ui/src/i18n/locales/ja/chat.json
+++ b/webview-ui/src/i18n/locales/ja/chat.json
@@ -344,9 +344,9 @@
 		"support": "<githubLink>GitHub</githubLink>でスターを付けて Zoo Code を応援してください。",
 		"release": {
 			"heading": "新機能:",
-			"highlight1": "注目の新モデルが3つ登場 — 最新の Gemini 3.7 Flash、GLM 5.3、Qwen3.8 Max に加え、DeepSeek V4 の reasoning、価格、プロバイダー対応も更新されました。",
-			"highlight2": "NanoGPT に接続 — 動的なモデル検出、ストリーミング、Prompt 補完に加え、速度、価格、レイテンシ、スループット、ツール対応、キャッシュに基づくルーティング設定を利用できます。",
-			"highlight3": "プロバイダーとタスクの信頼性を向上 — Azure OpenAI エンドポイント設定、Kimi Code の出力上限、タスク履歴タイトルの保持、Zoo 設定のインポート/エクスポートを改善しました。"
+			"highlight1": "より安全なファイルアクセス — 読み取り用と書き込み用の許可リストを分けて、Zoo Code がアクセスできるパスを制御できます。",
+			"highlight2": "モデルの選択肢がさらに充実 — GLM 5.3 が Z.ai で利用可能になり、Gemini プロバイダーに Gemini 3.5 Flash Lite と Gemini 3.1 Flash Lite が追加されました。",
+			"highlight3": "日常のワークフローをより確実に — タスクの復元と履歴、プロバイダーのストリーミングとツール使用、設定のインポート、IDE テーマ全体での読みやすさを改善しました。"
 		},
 		"cloudAgents": {
 			"heading": "クラウドの新機能:",
diff --git a/webview-ui/src/i18n/locales/ko/chat.json b/webview-ui/src/i18n/locales/ko/chat.json
index 090bb1a706..c988d469dd 100644
--- a/webview-ui/src/i18n/locales/ko/chat.json
+++ b/webview-ui/src/i18n/locales/ko/chat.json
@@ -344,9 +344,9 @@
 		"support": "<githubLink>GitHub</githubLink>에서 별표를 눌러 Zoo Code를 응원해 주세요.",
 		"release": {
 			"heading": "새로운 기능:",
-			"highlight1": "주요 신규 모델 3종이 출시되었습니다 — 완전히 새로운 Gemini 3.7 Flash, GLM 5.3, Qwen3.8 Max 모델과 업데이트된 DeepSeek V4 reasoning, 가격 및 provider 지원을 사용해 보세요.",
-			"highlight2": "NanoGPT에 연결하세요 — 동적 모델 검색, 스트리밍 및 prompt completion과 함께 속도, 가격, 지연 시간, 처리량, 도구 지원, 캐싱을 기준으로 한 라우팅 설정을 사용할 수 있습니다.",
-			"highlight3": "더 안정적인 provider와 task — Azure OpenAI endpoint 설정, Kimi Code 출력 제한, task 기록 제목 보존, Zoo 설정 가져오기/내보내기가 개선되었습니다."
+			"highlight1": "더 안전한 파일 접근 — 읽기와 쓰기 허용 목록을 분리해 Zoo Code가 접근할 수 있는 경로를 제어하세요.",
+			"highlight2": "더 다양한 모델 선택 — 이제 Z.ai에서 GLM 5.3을 사용할 수 있으며 Gemini provider에 Gemini 3.5 Flash Lite와 Gemini 3.1 Flash Lite가 추가되었습니다.",
+			"highlight3": "더 안정적인 일상 workflow — task 복구와 기록, provider 스트리밍과 tool 사용, 설정 가져오기, 다양한 IDE 테마에서의 가독성이 개선되었습니다."
 		},
 		"cloudAgents": {
 			"heading": "클라우드의 새로운 기능:",
diff --git a/webview-ui/src/i18n/locales/nl/chat.json b/webview-ui/src/i18n/locales/nl/chat.json
index 0f1ce14084..e6f388281e 100644
--- a/webview-ui/src/i18n/locales/nl/chat.json
+++ b/webview-ui/src/i18n/locales/nl/chat.json
@@ -317,9 +317,9 @@
 		"support": "Steun Zoo Code door ons een ster te geven op <githubLink>GitHub</githubLink>.",
 		"release": {
 			"heading": "Wat is nieuw:",
-			"highlight1": "Drie belangrijke nieuwe modellen zijn gearriveerd — gebruik de gloednieuwe Gemini 3.7 Flash-, GLM 5.3- en Qwen3.8 Max-modellen, plus bijgewerkte reasoning, prijzen en providerondersteuning voor DeepSeek V4.",
-			"highlight2": "Maak verbinding met NanoGPT — gebruik dynamische modeldetectie, streaming en promptaanvulling, plus routeringsvoorkeuren voor snelheid, prijs, latentie, doorvoer, toolondersteuning en caching.",
-			"highlight3": "Betrouwbaardere providers en taken — oplossingen verbeteren de instelling van Azure OpenAI-endpoints, Kimi Code-uitvoerlimieten, het behouden van titels in de taakgeschiedenis en het importeren/exporteren van Zoo-instellingen."
+			"highlight1": "Veiligere bestandstoegang — gebruik afzonderlijke toelatingslijsten voor lezen en schrijven om te bepalen tot welke paden Zoo Code toegang heeft.",
+			"highlight2": "Meer modelkeuze — GLM 5.3 is nu beschikbaar via Z.ai en Gemini 3.5 Flash Lite en Gemini 3.1 Flash Lite zijn toegevoegd aan de Gemini-provider.",
+			"highlight3": "Betrouwbaardere dagelijkse workflows — verbeteringen versterken taakherstel en -geschiedenis, streaming en toolgebruik van providers, het importeren van instellingen en de leesbaarheid in verschillende IDE-thema's."
 		},
 		"cloudAgents": {
 			"heading": "Nieuw in de Cloud:",
diff --git a/webview-ui/src/i18n/locales/pl/chat.json b/webview-ui/src/i18n/locales/pl/chat.json
index 53ae2013e1..39c2d1c9cd 100644
--- a/webview-ui/src/i18n/locales/pl/chat.json
+++ b/webview-ui/src/i18n/locales/pl/chat.json
@@ -344,9 +344,9 @@
 		"support": "Wesprzyj Zoo Code, dając nam gwiazdkę na <githubLink>GitHub</githubLink>.",
 		"release": {
 			"heading": "Co nowego:",
-			"highlight1": "Pojawiły się trzy ważne nowe modele — korzystaj z zupełnie nowych Gemini 3.7 Flash, GLM 5.3 i Qwen3.8 Max oraz zaktualizowanego reasoning, cen i obsługi providerów dla DeepSeek V4.",
-			"highlight2": "Połącz się z NanoGPT — korzystaj z dynamicznego wykrywania modeli, streamingu i uzupełniania promptów oraz preferencji routingu według szybkości, ceny, opóźnienia, przepustowości, obsługi narzędzi i cache.",
-			"highlight3": "Bardziej niezawodni providerzy i zadania — poprawki usprawniają konfigurację endpointów Azure OpenAI, limity wyjścia Kimi Code, zachowywanie tytułów historii zadań oraz import/eksport ustawień Zoo."
+			"highlight1": "Bezpieczniejszy dostęp do plików — używaj oddzielnych list dozwolonych ścieżek odczytu i zapisu, aby kontrolować, do których ścieżek Zoo Code ma dostęp.",
+			"highlight2": "Większy wybór modeli — GLM 5.3 jest teraz dostępny przez Z.ai, a Gemini 3.5 Flash Lite i Gemini 3.1 Flash Lite dodano do providera Gemini.",
+			"highlight3": "Bardziej niezawodne codzienne workflow — ulepszenia wzmacniają odzyskiwanie zadań i historię, streaming i użycie narzędzi przez providerów, import ustawień oraz czytelność w różnych motywach IDE."
 		},
 		"cloudAgents": {
 			"heading": "Nowości w chmurze:",
diff --git a/webview-ui/src/i18n/locales/pt-BR/chat.json b/webview-ui/src/i18n/locales/pt-BR/chat.json
index 4769341a7b..9dc67a627c 100644
--- a/webview-ui/src/i18n/locales/pt-BR/chat.json
+++ b/webview-ui/src/i18n/locales/pt-BR/chat.json
@@ -344,9 +344,9 @@
 		"support": "Apoie o Zoo Code nos dando uma estrela no <githubLink>GitHub</githubLink>.",
 		"release": {
 			"heading": "Novidades:",
-			"highlight1": "Três novos modelos importantes chegaram — use os novíssimos Gemini 3.7 Flash, GLM 5.3 e Qwen3.8 Max, além das atualizações de reasoning, preços e cobertura de providers do DeepSeek V4.",
-			"highlight2": "Conecte-se ao NanoGPT — use descoberta dinâmica de modelos, streaming e conclusão de prompts, com preferências de roteamento por velocidade, preço, latência, throughput, suporte a ferramentas e cache.",
-			"highlight3": "Providers e tarefas mais confiáveis — as correções melhoram a configuração de endpoints do Azure OpenAI, os limites de saída do Kimi Code, a preservação de títulos no histórico de tarefas e a importação/exportação das configurações do Zoo."
+			"highlight1": "Acesso a arquivos mais seguro — use listas de permissões separadas para leitura e gravação a fim de controlar quais caminhos o Zoo Code pode acessar.",
+			"highlight2": "Mais opções de modelos — o GLM 5.3 agora está disponível pela Z.ai, e o Gemini 3.5 Flash Lite e o Gemini 3.1 Flash Lite foram adicionados ao provider Gemini.",
+			"highlight3": "Workflows cotidianos mais confiáveis — as melhorias reforçam a recuperação e o histórico de tarefas, o streaming e o uso de ferramentas dos providers, a importação de configurações e a legibilidade em diferentes temas da IDE."
 		},
 		"cloudAgents": {
 			"heading": "Novidades na Nuvem:",
diff --git a/webview-ui/src/i18n/locales/ru/chat.json b/webview-ui/src/i18n/locales/ru/chat.json
index de34a0e0b8..7eb863904f 100644
--- a/webview-ui/src/i18n/locales/ru/chat.json
+++ b/webview-ui/src/i18n/locales/ru/chat.json
@@ -318,9 +318,9 @@
 		"support": "Поддержите Zoo Code, поставив нам звезду на <githubLink>GitHub</githubLink>.",
 		"release": {
 			"heading": "Что нового:",
-			"highlight1": "Появились три важных новых модели — используйте совершенно новые Gemini 3.7 Flash, GLM 5.3 и Qwen3.8 Max, а также обновлённые reasoning, цены и поддержку провайдеров для DeepSeek V4.",
-			"highlight2": "Подключайтесь к NanoGPT — используйте динамическое обнаружение моделей, streaming и дополнение Prompt, а также настройки маршрутизации по скорости, цене, задержке, пропускной способности, поддержке инструментов и кешированию.",
-			"highlight3": "Более надёжные провайдеры и задачи — исправления улучшают настройку endpoint Azure OpenAI, лимиты вывода Kimi Code, сохранение заголовков истории задач и импорт/экспорт настроек Zoo."
+			"highlight1": "Более безопасный доступ к файлам — используйте отдельные списки разрешённых путей для чтения и записи, чтобы контролировать, к каким путям может обращаться Zoo Code.",
+			"highlight2": "Больше моделей на выбор — GLM 5.3 теперь доступна через Z.ai, а Gemini 3.5 Flash Lite и Gemini 3.1 Flash Lite добавлены в провайдер Gemini.",
+			"highlight3": "Более надёжные повседневные процессы — улучшены восстановление и история задач, streaming и использование инструментов провайдерами, импорт настроек и читаемость в разных темах IDE."
 		},
 		"cloudAgents": {
 			"heading": "Новое в облаке:",
diff --git a/webview-ui/src/i18n/locales/tr/chat.json b/webview-ui/src/i18n/locales/tr/chat.json
index 752e5bff9f..b6d43b4b12 100644
--- a/webview-ui/src/i18n/locales/tr/chat.json
+++ b/webview-ui/src/i18n/locales/tr/chat.json
@@ -345,9 +345,9 @@
 		"support": "Lütfen <githubLink>GitHub</githubLink>'da bize yıldız vererek Zoo Code'u destekle.",
 		"release": {
 			"heading": "Yenilikler:",
-			"highlight1": "Üç önemli yeni model geldi — yepyeni Gemini 3.7 Flash, GLM 5.3 ve Qwen3.8 Max modellerini, ayrıca güncellenmiş DeepSeek V4 reasoning, fiyatlandırma ve provider kapsamını kullan.",
-			"highlight2": "NanoGPT'ye bağlan — dinamik model keşfi, streaming ve Prompt tamamlama ile hız, fiyat, gecikme, throughput, araç desteği ve caching için yönlendirme tercihlerini kullan.",
-			"highlight3": "Daha güvenilir provider'lar ve görevler — düzeltmeler Azure OpenAI endpoint kurulumunu, Kimi Code çıktı sınırlarını, görev geçmişi başlıklarının korunmasını ve Zoo ayarlarının içe/dışa aktarımını iyileştiriyor."
+			"highlight1": "Daha güvenli dosya erişimi — Zoo Code'un hangi yollara erişebileceğini kontrol etmek için ayrı Okuma ve Yazma izin listeleri kullan.",
+			"highlight2": "Daha fazla model seçeneği — GLM 5.3 artık Z.ai üzerinden kullanılabilir; Gemini 3.5 Flash Lite ve Gemini 3.1 Flash Lite da Gemini provider'ına eklendi.",
+			"highlight3": "Daha güvenilir günlük iş akışları — iyileştirmeler görev kurtarma ve geçmişini, provider streaming ve araç kullanımını, ayar içe aktarmayı ve IDE temalarındaki okunabilirliği güçlendiriyor."
 		},
 		"cloudAgents": {
 			"heading": "Cloud'daki yenilikler:",
diff --git a/webview-ui/src/i18n/locales/vi/chat.json b/webview-ui/src/i18n/locales/vi/chat.json
index 3caa0e8d3d..e6c7ded31a 100644
--- a/webview-ui/src/i18n/locales/vi/chat.json
+++ b/webview-ui/src/i18n/locales/vi/chat.json
@@ -345,9 +345,9 @@
 		"support": "Hãy ủng hộ Zoo Code bằng cách tặng sao cho chúng tôi trên <githubLink>GitHub</githubLink>.",
 		"release": {
 			"heading": "Có gì mới:",
-			"highlight1": "Ba model mới quan trọng đã xuất hiện — sử dụng các model hoàn toàn mới Gemini 3.7 Flash, GLM 5.3 và Qwen3.8 Max, cùng reasoning, giá và phạm vi provider được cập nhật cho DeepSeek V4.",
-			"highlight2": "Kết nối với NanoGPT — sử dụng khám phá model động, streaming và hoàn thành Prompt, cùng tùy chọn định tuyến theo tốc độ, giá, độ trễ, throughput, hỗ trợ tool và caching.",
-			"highlight3": "Provider và task đáng tin cậy hơn — các bản sửa lỗi cải thiện thiết lập endpoint Azure OpenAI, giới hạn đầu ra Kimi Code, giữ nguyên tiêu đề lịch sử task và nhập/xuất cài đặt Zoo."
+			"highlight1": "Truy cập file an toàn hơn — sử dụng allowlist Đọc và Ghi riêng biệt để kiểm soát các path mà Zoo Code có thể truy cập.",
+			"highlight2": "Nhiều lựa chọn model hơn — GLM 5.3 hiện có qua Z.ai, đồng thời Gemini 3.5 Flash Lite và Gemini 3.1 Flash Lite đã được thêm vào provider Gemini.",
+			"highlight3": "Workflow hằng ngày đáng tin cậy hơn — các cải tiến tăng cường khôi phục và lịch sử task, streaming và sử dụng tool của provider, nhập cài đặt và khả năng đọc trên các theme IDE."
 		},
 		"cloudAgents": {
 			"heading": "Mới trên Cloud:",
diff --git a/webview-ui/src/i18n/locales/zh-CN/chat.json b/webview-ui/src/i18n/locales/zh-CN/chat.json
index 76989eb473..bc8f5dba93 100644
--- a/webview-ui/src/i18n/locales/zh-CN/chat.json
+++ b/webview-ui/src/i18n/locales/zh-CN/chat.json
@@ -345,9 +345,9 @@
 		"support": "请在 <githubLink>GitHub</githubLink> 上为 Zoo Code 点星支持我们。",
 		"release": {
 			"heading": "新增功能：",
-			"highlight1": "三款重磅新模型现已推出 — 使用全新的 Gemini 3.7 Flash、GLM 5.3 和 Qwen3.8 Max 模型，以及更新后的 DeepSeek V4 推理、定价和提供商覆盖。",
-			"highlight2": "连接 NanoGPT — 使用动态模型发现、流式传输和 Prompt 补全，并按速度、价格、延迟、吞吐量、工具支持和缓存设置路由偏好。",
-			"highlight3": "更可靠的提供商和任务 — 修复改进了 Azure OpenAI endpoint 设置、Kimi Code 输出限制、任务历史标题保留以及 Zoo 设置的导入/导出。"
+			"highlight1": "更安全的文件访问 — 使用独立的读取和写入允许列表，控制 Zoo Code 可以访问的路径。",
+			"highlight2": "更多模型选择 — GLM 5.3 现已通过 Z.ai 提供，Gemini provider 也新增了 Gemini 3.5 Flash Lite 和 Gemini 3.1 Flash Lite。",
+			"highlight3": "更可靠的日常工作流 — 改进增强了任务恢复与历史记录、provider 流式传输与工具使用、设置导入，以及不同 IDE 主题下的可读性。"
 		},
 		"cloudAgents": {
 			"heading": "云端新功能：",
diff --git a/webview-ui/src/i18n/locales/zh-TW/chat.json b/webview-ui/src/i18n/locales/zh-TW/chat.json
index e096c27af4..d2fe37a774 100644
--- a/webview-ui/src/i18n/locales/zh-TW/chat.json
+++ b/webview-ui/src/i18n/locales/zh-TW/chat.json
@@ -366,9 +366,9 @@
 		"support": "請在 <githubLink>GitHub</githubLink> 上為 Zoo Code 加星支持我們。",
 		"release": {
 			"heading": "新增功能：",
-			"highlight1": "三款重磅新模型現已推出 — 使用全新的 Gemini 3.7 Flash、GLM 5.3 和 Qwen3.8 Max 模型，以及更新後的 DeepSeek V4 推理、定價與供應商支援。",
-			"highlight2": "連接 NanoGPT — 使用動態模型探索、串流與 Prompt 補全，並依速度、價格、延遲、吞吐量、工具支援和快取設定路由偏好。",
-			"highlight3": "更可靠的供應商與任務 — 修正改善了 Azure OpenAI endpoint 設定、Kimi Code 輸出限制、任務歷史標題保留，以及 Zoo 設定的匯入/匯出。"
+			"highlight1": "更安全的檔案存取 — 使用獨立的讀取與寫入允許清單，控制 Zoo Code 可以存取的路徑。",
+			"highlight2": "更多模型選擇 — GLM 5.3 現已透過 Z.ai 提供，Gemini provider 也新增了 Gemini 3.5 Flash Lite 和 Gemini 3.1 Flash Lite。",
+			"highlight3": "更可靠的日常工作流程 — 改善強化了任務復原與歷史記錄、provider 串流與工具使用、設定匯入，以及不同 IDE 主題下的可讀性。"
 		},
 		"cloudAgents": {
 			"heading": "雲端的新功能：",