fix(chat): 删除会话期间阻止发送和重复删除

Reviewed-on: #22

.gitignore

@@ -7,6 +7,7 @@ frontend/*.tsbuildinfo
 # Backend
 backend/.venv/
 backend/.venv-models/
+backend/.venv-models-cuda/
 backend/data/models/
 backend/data/attachments/
 backend/.uv-cache/

README.md

@@ -1,153 +1,136 @@
 # Notes Agent（暂命名） 团队开发说明
 
-> 本文件用于团队开发期间快速配置环境和启动项目，不是正式的项目 README。
+> 本文件用于团队开发期间快速配置环境、启动项目并了解当前实现状态，不是正式的项目 README。
 
-> 当前基线：2026-09-03。第一阶段 Web 联调前后端已经完成；第二阶段已完成 Workspace 去 Mock、Agent Trace 持久化与 SSE 恢复、stdio MCP Bridge、隔离 Plugin Host、Plugin Command/Settings，以及独立 MCP Server 配置中心 C.1（stdio、Streamable HTTP 与旧 SSE 兼容）。真实音频、Provider 协议增强、Benchmark、导出、主题包、Trace 可视化、Mermaid 与函数图像仍在后续开发；Tauri Host、Stronghold、原生多 Vault 文件系统和 Sync Server 尚未接入。
+NotesAgent 是本地优先的 AI 笔记与知识库项目。当前可运行形态为 Vue/Vite Web 前端与 FastAPI AI Core：Markdown 和附件保存在本地 Vault，SQLite 管理元数据、全文索引、向量空间、搜索历史、AI 会话、任务、Agent Trace、多模态任务及运行诊断。AI 对话已接入知识库检索，会话与消息由后端持久化并供 Web 和桌面客户端共用。
 
-## 当前目录
+截至 2026-09-05，第一阶段及第二阶段 A～F 的工程范围已经合并到 `main`。当前已完成真实 Workspace、混合检索与知识库问答、Agent/Tool/Permission、Skill/Plugin、MCP 配置与调用、模型提供商与路由、RAG Benchmark，以及本地 Embedding、音频转写和片段级声纹聚类。Tauri/Rust Host、Stronghold、原生多 Vault 文件系统、生产级 MCP 沙箱和 Sync Server 尚未接入。
+
+## 目录
 
 ```text
 NotesAgent/
 ├── frontend/      Vue 3 + TypeScript + Vite 前端
-├── backend/       FastAPI + Pydantic 后端
+├── backend/       FastAPI AI Core、SQLite 与本地模型运行管理
 ├── docs/          架构、契约、开发说明、协作规范与问题复盘
 └── server sync/   云同步服务预留目录，当前未实现
 ```
 
+## 当前能力
+
+- 工作区：打开一个后端配置的真实 Vault，编辑 Markdown，管理文件与目录。
+- 检索与问答：FTS5、sqlite-vec、RRF 与轻量词面精排；搜索历史持久化到后端 SQLite；AI 对话自动检索知识库并返回 Citation。
+- Agent 与扩展：持久化 Trace、可恢复 SSE、Tool/Permission、Skill、Plugin Command/Settings/Secret、隔离 Plugin Host。
+- MCP：独立配置 stdio、Streamable HTTP 和旧 SSE Server，发现并调用工具；生产 stdio 沙箱等待 Tauri Host。
+- 模型服务：OpenAI Chat/Compatible、OpenAI Responses、Anthropic Messages、Ollama；国内常用提供商 logo 预设、独立凭据、模型发现和自定义请求 JSON。
+- 多模态：API 优先，未配置或响应无效时回退本地；`local_only` 禁止远程调用。任务、修订、事件、来源和回退原因写入 SQLite。
+- 模型运行：默认 CPU，可选 CUDA 12.8 组件；固定模型 revision，按需启动独立子进程，交互检索优先排队，CUDA 初始化或显存失败时用同一冻结配置在 CPU 重试一次。
+- 可观测性：输入、输出、缓存命中、推理 Token 与音频用量卡片；本地运行诊断保留最近 200 条，不保存正文、文件路径、密钥或异常全文。
+- 界面偏好：设置页可即时切换全局中文/英文界面，并控制由系统词典提供的编辑器拼写检查；偏好目前保存于 Web 端设备配置，后续由 Tauri 配置存储接管。
+
+## 本地模型
+
+| 能力 | 当前模型 | 许可 | 说明 |
+| --- | --- | --- | --- |
+| 默认 Embedding | `hotchpotch/bekko-embedding-v1-a8m` | MIT | 384 维，中文检索默认选择 |
+| 可选 Embedding | `ibm-granite/granite-embedding-97m-multilingual-r2` | Apache-2.0 | 384 维，多语言备选 |
+| 音频转写与语言识别 | `Qwen/Qwen3-ASR-0.6B` | Apache-2.0 | 返回片段级时间边界 |
+| 声纹提取与匹配 | `iic/speech_eres2netv2_sv_zh-cn_16k-common` | Apache-2.0 | 192 维声纹，供相似度和片段聚类使用 |
+
+模型权重按代码中的固定 revision 下载并校验，推理阶段离线读取。当前说话人处理是能量分段、ASR 片段与 ERes2NetV2 聚类，不包含逐字强制对齐、同段多人或重叠语音分离。`HashEmbeddingProvider` 只用于确定性测试注入。
+
 ## 开发环境
 
-当前开发版需要：
+| 环境 | 要求 |
+| --- | --- |
+| Git | 较新稳定版 |
+| Node.js | 22+，推荐 24 |
+| pnpm | 10+ |
+| Python | 3.11+，推荐 3.12 |
+| uv | 较新稳定版 |
 
-| 环境 | 要求 | 说明 |
+当前 Web 联调不需要 Rust 和 Tauri。桌面端集成时再安装 Rust Toolchain 与 Tauri CLI。
-| --- | --- | --- |
-| Git | 较新稳定版 | 代码版本管理 |
-| Node.js | 22 或更高版本 | 推荐使用 Node.js 24 |
-| pnpm | 10 或更高版本 | 前端依赖与脚本管理 |
-| Python | 3.11 或更高版本 | 推荐使用 Python 3.12 |
-| uv | 较新稳定版 | 后端依赖和虚拟环境管理 |
 
-检查本机环境：
+## 初始化与启动
 
-```powershell
+安装 API 与前端依赖：
-git --version
-node --version
-pnpm --version
-python --version
-uv --version
-```
-
-当前 Web 联调不需要 Rust 和 Tauri。开始桌面端集成后，再按照 `docs/architecture/AI笔记软件技术栈说明-团队版-v2.3.md` 安装 Rust Toolchain 与 Tauri CLI。
-
-## 首次初始化
-
-### 后端
 
 ```powershell
 cd backend
 uv sync
-cd ..
+cd ../frontend
-```
-
-`uv sync` 会根据 `backend/pyproject.toml` 安装依赖，并自动创建和管理 `backend/.venv`，不需要手动创建或激活虚拟环境。
-
-### 前端
-
-```powershell
-cd frontend
 pnpm install
 cd ..
 ```
 
-## 启动开发环境
+在两个终端分别启动：
-
-前端和后端需要在两个终端中分别启动。
-
-### 终端一：启动后端
 
 ```powershell
+# 终端一
 cd backend
 uv run uvicorn app.main:app --reload --host 127.0.0.1 --port 8000
-```
 
-后端地址：
+# 终端二
-
-- 健康检查：<http://127.0.0.1:8000/health>
-- 服务状态：<http://127.0.0.1:8000/api/status>
-- API 文档：<http://127.0.0.1:8000/docs>
-- OpenAPI JSON：<http://127.0.0.1:8000/openapi.json>
-
-#### 开发环境使用外部模型
-
-在“设置 → 模型提供商”中选择 DeepSeek 或 OpenAI 预设后，直接在密码输入框填写 API Key。前端只在提交期间持有该值，不写入 Pinia 或 localStorage；AI Core 将其加密保存到本机 `backend/data/credentials/`，Provider 配置只保留内部 Credential ID。
-
-该目录同时包含本地开发用主密钥和密文，并已加入 `.gitignore`。这提供本地静态加密和完整性校验，但不能替代操作系统凭据库。开始 Tauri 桌面集成后，应将存储实现迁移到 Stronghold，保留现有 Credential API 与 Provider 接口边界。
-
-无界面或自动化环境仍可使用 `DEEPSEEK_API_KEY`、`OPENAI_API_KEY` 或 `AINOTE_CREDENTIAL_<ID>` 注入；设置页保存的本地密钥优先，环境变量仅在本地未保存对应 Credential ID 时作为回退。密钥不得写入仓库文件、README、Issue、提交信息或聊天记录。
-
-### 终端二：启动前端
-
-```powershell
 cd frontend
 pnpm dev
 ```
 
-前端地址：<http://127.0.0.1:5173>
+前端地址为 <http://127.0.0.1:5173>，Vite 将 `/api` 和 `/health` 代理到 <http://127.0.0.1:8000>。后端提供健康检查 `/health`、服务状态 `/api/status`、API 文档 `/docs` 和机器可读契约 `/openapi.json`。
 
-开发环境中，Vite 会将 `/api` 和 `/health` 请求代理到 `http://127.0.0.1:8000`。联调时应先启动后端，再启动或刷新前端。
+## 安装本地模型运行组件
+
+API 环境保留在 `backend/.venv`，模型依赖安装到独立环境。默认安装 CPU：
+
+```powershell
+./backend/scripts/install-model-runtime.ps1
+```
+
+CUDA 为 Windows 可选组件，可在“设置 → 模型提供商 → 本地模型”中安装，也可保留 CPU 环境并创建独立 CUDA 环境：
+
+```powershell
+./backend/scripts/install-model-runtime.ps1 -Device cuda -RuntimeDirectory ./backend/.venv-models-cuda
+$env:APP_MODEL_PYTHON = (Resolve-Path ./backend/.venv-models-cuda/Scripts/python.exe).Path
+```
+
+脚本固定 `torch`/`torchaudio` 2.9.1，CPU 使用官方 CPU wheel，CUDA 使用 cu128 wheel；脚本不会安装或修改 NVIDIA 驱动。模型权重需要在设置页显式下载，不会在推理时自动下载。
+
+## 模型提供商与凭据
+
+在“设置 → 模型提供商”中选择预设或创建自定义提供商。API Key 只在前端提交期间存在，不写入 Pinia 或 `localStorage`；后端将密文和开发主密钥保存到已忽略的 `backend/data/credentials/`，Provider 配置只保存 Credential ID。
+
+无界面环境可使用 `OPENAI_API_KEY`、`DEEPSEEK_API_KEY` 或 `AINOTE_CREDENTIAL_<ID>`。当前 Fernet 存储用于 Web 联调，桌面端将沿用 Credential API 边界迁移到 Stronghold。
 
 ## 测试与构建
 
-后端测试：
-
 ```powershell
 cd backend
 uv run pytest
-```
 
-前端类型检查及生产构建：
+cd ../frontend
-
+pnpm test
-```powershell
-cd frontend
 pnpm build
 ```
 
-前端单元与组件测试：
+当前回归基线为后端 559 项、前端 106 项测试通过，TypeScript 类型检查与生产构建通过。存在一条既有 Starlette/httpx 弃用提示和 Vite 大 bundle 提示；测试数量以当前分支实际输出和 CI 为准。
 
-```powershell
+## 文档
-cd frontend
-pnpm test
-```
-
-当前回归基线为后端 218 项测试、前端 32 项测试，且 TypeScript 类型检查和生产构建通过。测试数量会随功能增长，以本地实际输出和 CI 为准。
-
-构建产物位于 `frontend/dist`，该目录不提交到 Git。
-
-## 文档导航
 
 | 文档 | 用途 |
 | --- | --- |
-| [文档总索引](docs/README.md) | 文档分类、阅读顺序和维护规则 |
+| [文档总索引](docs/README.md) | 全部架构、契约、开发说明和复盘入口 |
-| [技术栈说明](docs/architecture/AI笔记软件技术栈说明-团队版-v2.3.md) | 目标架构、第二阶段技术边界与模块依赖 |
+| [前端 README](frontend/README.md) | 前端结构、运行方式和数据边界 |
-| [第二阶段分工表](docs/architecture/第二阶段团队分工表.md) | 第二阶段人员职责、任务顺序、协作关系与验收项 |
+| [后端 README](backend/README.md) | API Core、模型运行与配置 |
-| [后端接口契约](docs/contracts/后端接口契约-开发版.md) | HTTP/SSE 接口、错误和当前实现状态 |
+| [技术栈说明](docs/architecture/AI笔记软件技术栈说明-团队版-v2.3.md) | 当前技术基线、目标桌面架构与模块边界 |
-| [第二阶段接口契约](docs/contracts/第二阶段接口契约-开发版.md) | 第二阶段公共 DTO、计划接口、SSE、错误码与联调顺序 |
+| [多模态与模型运行](docs/development/多模态管线与模型运行开发说明.md) | 模型 revision、CPU/CUDA、路由、用量和接口 |
-| [AI Core 与 Agent Core](docs/development/AI-Core与Agent-Core开发说明.md) | Provider、Agent、Tool、Permission 与 Extension Core |
+| [阶段 F 收尾验收](docs/development/阶段F收尾验收记录.md) | 自动化、CPU/CUDA 真实闭环和未关闭专项 |
-| [MCP Bridge 与 Plugin Host](docs/development/MCP-Bridge与Plugin-Host开发说明.md) | stdio MCP、隔离进程、Tool 映射、状态与错误边界 |
+| [后端接口契约](docs/contracts/后端接口契约-开发版.md) | 当前 HTTP/SSE 接口说明 |
-| [Plugin Command 与 Settings](docs/development/Plugin-Command与Settings开发说明.md) | Command Registry、Settings Schema、Secret 引用与联调边界 |
+| [第二阶段接口契约](docs/contracts/第二阶段接口契约-开发版.md) | 第二阶段公共 DTO 与行为边界 |
-| [Plugin Command 与 Settings 复盘](docs/retrospectives/Plugin-Command与Settings问题与修复复盘.md) | 阶段 D 连续审阅发现的安全、事务、Schema 与运行时契约问题 |
-| [Git 使用细则](docs/guides/Git使用细则-团队开发版.md) | 分支、提交、PR、Review 与合并流程 |
-| [CI/CD 细则](docs/guides/CI-CD细则-团队开发版.md) | Gitea 流水线、质量门禁、产物、发布与回滚规则 |
-| [Agent Trace 复盘](docs/retrospectives/Agent-Core第二阶段问题与修复复盘.md) | Agent 持久化、SSE 恢复、事件契约与脱敏问题复盘 |
 
-## 日常开发注意事项
+## 开发约定
 
-- Python 依赖统一修改 `backend/pyproject.toml`，修改后执行 `uv sync`。
+- 后端依赖统一修改 `backend/pyproject.toml` 并执行 `uv sync`；模型依赖由 `backend/scripts/model-requirements.lock` 锁定。
-- 前端依赖统一使用 pnpm 安装，不要混用 npm 或 yarn。
+- 前端依赖统一使用 pnpm，不混用 npm 或 yarn。
-- `backend/.venv`、`frontend/node_modules`、`frontend/dist` 均为本地生成目录，不提交到 Git。
+- `backend/.venv*`、模型权重、`frontend/node_modules` 和 `frontend/dist` 都是本地产物，不提交 Git。
-- API 默认监听 `127.0.0.1:8000`，前端默认监听 `127.0.0.1:5173`。
+- 前端不直接访问 SQLite 或厂商模型协议；持久数据通过 FastAPI 服务读写。
-- 后端附件目录默认是 `backend/data/attachments`，可通过 `APP_ATTACHMENTS_PATH` 覆盖；该目录由桌面 Host 管理。
+- 接口或数据结构变化时，同一提交同步更新前后端类型、契约和开发说明。
-- 跨模块接口发生变化时，需要同步更新前后端类型和 `docs` 中的接口说明。
+- 当前行为以代码、测试和运行中的 `/openapi.json` 为准；规划能力必须在文档中明确标注。
-- 当前已实现接口见 `docs/contracts/后端接口契约-开发版.md`，第二阶段规划接口见 `docs/contracts/第二阶段接口契约-开发版.md`；已实现能力以 `/openapi.json` 为准。
-- 前端页面、交互、状态管理及当前阶段后续页面需求见 `docs/contracts/前端页面需求说明-开发版.md`。
-- 分支、提交、Pull Request、Review 和冲突处理规范见 `docs/guides/Git使用细则-团队开发版.md`。
-- CI 检查、产物、发布和回滚规范见 `docs/guides/CI-CD细则-团队开发版.md`。

backend/README.md

@@ -1,34 +1,103 @@
-# Notes Agent Backend
+# NotesAgent Backend
 
-FastAPI + Pydantic 的本地 AI Core / Agent Core。项目使用 uv 管理依赖和虚拟环境。
+NotesAgent Backend 是基于 Python 3.11+、FastAPI、Pydantic v2 和 SQLite 的本地 AI Core / Agent Core，使用 uv 管理 API 依赖和虚拟环境。
 
-当前实现包含 Knowledge/Retrieval、Chat、Agent、Tool/Permission、Skill/Plugin、MCP、模型提供商与多模态任务。支持 OpenAI Chat/Compatible、Responses、Anthropic Messages 和 Ollama；真实本地 Embedding、ASR、声纹模型默认 CPU，CUDA 显式选装。操作系统级 Plugin 沙箱仍属于后续阶段。
+当前实现包含 Knowledge/Retrieval、Chat、Agent、Tool/Permission、Skill/Plugin、MCP、模型提供商、RAG Benchmark、多模态任务、本地模型调度、Token/音频用量和运行诊断。数据持久化位于后端 SQLite 与 Vault；Tauri Sidecar 生命周期、Stronghold 和操作系统级 Plugin 沙箱属于后续桌面阶段。
+
+## 初始化与运行
 
 ```powershell
 uv sync
 uv run uvicorn app.main:app --reload --host 127.0.0.1 --port 8000
 ```
 
-`uv sync` 首次运行时会自动创建由 uv 管理的 `.venv`，无需手动执行 `python -m venv` 或激活环境。
+`uv sync` 会创建并管理 `backend/.venv`，无需手动激活环境。启动后可访问：
-
-启动后可访问：
 
 - 健康检查：<http://127.0.0.1:8000/health>
+- 服务状态：<http://127.0.0.1:8000/api/status>
 - API 文档：<http://127.0.0.1:8000/docs>
 - OpenAPI：<http://127.0.0.1:8000/openapi.json>
 
-运行回归测试：
+## 核心模块
+
+| 目录 | 职责 |
+| --- | --- |
+| `app/knowledge`、`app/retrieval` | Markdown 解析、FTS5、sqlite-vec、RRF、真实 Embedding 路由和 Citation |
+| `app/agent` | Agent Runtime、Tool 调用、权限与持久化 Trace |
+| `app/extensions` | Skill、Plugin Host、MCP Registry 与 stdio/HTTP/SSE Bridge |
+| `app/providers` | OpenAI Chat/Compatible、Responses、Anthropic Messages、Ollama 与能力路由 |
+| `app/local_models` | 模型目录、固定 revision 下载、独立进程、设备回退和队列调度 |
+| `app/services` | 索引、知识库上下文、聊天记录、转写、搜索历史、用量和诊断等应用服务 |
+| `app/benchmarks` | 版本化 RAG Dataset、异步评测、指标与报告 |
+
+## 模型路由
+
+Embedding、音频转写和声纹匹配遵循同一规则：
+
+1. 配置可用 API 时先调用 API；
+2. API 失败或返回无效结果时回退本地模型；
+3. 未配置 API 时直接使用本地模型；
+4. `local_only` 请求只允许本地模型；
+5. 响应和诊断记录实际来源、设备及回退原因。
+
+生产向量按 Provider、模型、revision、接口和维度隔离，切换空间后需要重建索引。Markdown 和 FTS 在模型不可用时仍可保存与查询；`HashEmbeddingProvider` 仅供测试显式注入。
+
+## 本地模型运行环境
+
+API 的 `backend/.venv` 与模型环境分离。默认安装 CPU 运行组件：
+
+```powershell
+./scripts/install-model-runtime.ps1
+```
+
+可选 CUDA 环境：
+
+```powershell
+./scripts/install-model-runtime.ps1 -Device cuda -RuntimeDirectory ./.venv-models-cuda
+$env:APP_MODEL_PYTHON = (Resolve-Path ./.venv-models-cuda/Scripts/python.exe).Path
+```
+
+脚本固定 `torch`/`torchaudio` 2.9.1，CUDA 使用 cu128 wheel，不安装驱动。其余模型依赖由 `scripts/model-requirements.lock` 锁定，包含 `qwen-asr`、`sentence-transformers`、ModelScope 和 PyAV。
+
+| 能力 | 模型 | 固定 revision | 许可 |
+| --- | --- | --- | --- |
+| 默认 Embedding | `hotchpotch/bekko-embedding-v1-a8m` | `c721113d59a1d91b447450324f51c4b3332c924a` | MIT |
+| 可选 Embedding | `ibm-granite/granite-embedding-97m-multilingual-r2` | `835ad14087e140460703cf0fae09f97d469d65c2` | Apache-2.0 |
+| 音频转写 | `Qwen/Qwen3-ASR-0.6B` | `5eb144179a02acc5e5ba31e748d22b0cf3e303b0` | Apache-2.0 |
+| 声纹匹配 | `iic/speech_eres2netv2_sv_zh-cn_16k-common` | `3317286545c587ae682dbc166831d9448780eebb` | Apache-2.0 |
+
+模型运行时默认 CPU。任务在独立子进程中按需加载并在结束后释放；队列中查询 Embedding、媒体任务、后台索引的优先级依次降低。CUDA 不可用、初始化失败或显存不足时，系统清理失败进程并以同一冻结配置在 CPU 重试一次。
+
+音频由 PyAV 解码为 16 kHz 单声道，经过能量分段、Qwen3-ASR 和 ERes2NetV2 片段聚类。当前只提供片段级时间戳，不支持逐字对齐、同段多人和重叠语音分离。
+
+## Provider 与凭据
+
+支持 OpenAI Chat/Compatible、OpenAI Responses、Anthropic Messages 和 Ollama。Provider 配置可分别绑定聊天、Embedding、转写和声纹能力，并通过受限的自定义请求 JSON 合并厂商扩展字段。
+
+API Key 可由前端设置页写入，也可通过 `OPENAI_API_KEY`、`DEEPSEEK_API_KEY` 或 `AINOTE_CREDENTIAL_<ID>` 注入。开发环境使用 Fernet 密文存储，接口不返回明文；`plugin.*` 是 Plugin Settings 的保留凭据命名空间。
+
+## 测试
 
 ```powershell
 uv run pytest
 ```
 
-阶段 F 后端基线为 472 项测试通过。Provider API Key 可通过前端设置页写入，也可用 `OPENAI_API_KEY`、`DEEPSEEK_API_KEY` 或 `AINOTE_CREDENTIAL_<ID>` 注入；不要把真实密钥写入仓库。`plugin.*` 是 Plugin Settings 的保留凭据命名空间，通用 Provider 凭据接口不能读写。
+当前基线为 562 项测试通过，另有一条既有 Starlette/httpx 弃用提示。真实模型冒烟脚本：
 
-本地模型 CPU/CUDA 安装、多模态任务、Token 用量与自定义 JSON 见 [多模态管线与模型运行开发说明](../docs/development/多模态管线与模型运行开发说明.md)。
+```powershell
+.venv/Scripts/python scripts/local-model-smoke.py bekko --download
+.venv/Scripts/python scripts/local-model-smoke.py qwen3-asr --download --audio C:/path/to/speech.wav
+.venv/Scripts/python scripts/local-model-smoke.py eres2netv2 --download --audio C:/path/to/speech.wav --reference C:/path/to/reference.wav
+```
 
-团队接口清单见 `../docs/contracts/后端接口契约-开发版.md`，机器可读契约以运行时的 `/openapi.json` 为准。
+## 相关文档
 
-AI Core 与 Agent Core 的模块边界、Mock Provider 和 Tool Calling 调试方式见 `../docs/development/AI-Core与Agent-Core开发说明.md`。
+- [后端接口契约](../docs/contracts/后端接口契约-开发版.md)
+- [第二阶段接口契约](../docs/contracts/第二阶段接口契约-开发版.md)
+- [多模态管线与模型运行](../docs/development/多模态管线与模型运行开发说明.md)
+- [阶段 F 收尾验收](../docs/development/阶段F收尾验收记录.md)
+- [AI Core 与 Agent Core](../docs/development/AI-Core与Agent-Core开发说明.md)
+- [Knowledge 与 Retrieval Core](../docs/development/Knowledge与Retrieval-Core开发说明.md)
+- [阶段 F：Embedding 与知识库问题](../docs/retrospectives/阶段F-Embedding与知识库问题与解决方案.md)
 
-Knowledge Core 与 Retrieval Core 的模块边界、数据模型、接口与检索流程见 `../docs/development/Knowledge与Retrieval-Core开发说明.md`。
+机器可读接口以运行中的 `/openapi.json` 为准。

backend/app/contracts.py

@@ -255,11 +255,57 @@ class ModelRequest(Contract):
 
 
 class ChatRequest(ModelRequest):
-    conversation_id: str | None = None
+    conversation_id: str | None = Field(default=None, min_length=1, max_length=128)
+    user_message_id: str | None = Field(default=None, min_length=1, max_length=128)
+    assistant_message_id: str | None = Field(default=None, min_length=1, max_length=128)
+    conversation_title: str | None = Field(default=None, max_length=120)
     use_rag: bool = True
     retrieval: SearchRequest | None = None
 
 
+class ConversationCreateRequest(Contract):
+    conversation_id: str | None = Field(default=None, min_length=1, max_length=128)
+    title: str = Field(min_length=1, max_length=120)
+
+    @field_validator("title")
+    @classmethod
+    def title_must_not_be_blank(cls, value: str) -> str:
+        value = value.strip()
+        if not value:
+            raise ValueError("title must not be blank")
+        return value
+
+
+class Conversation(Contract):
+    conversation_id: str
+    title: str
+    created_at: datetime
+    updated_at: datetime
+    message_count: int = 0
+
+
+class ConversationListResponse(Contract):
+    items: list[Conversation] = Field(default_factory=list)
+    page: PageMeta = Field(default_factory=PageMeta)
+
+
+class ChatMessage(Contract):
+    message_id: str
+    conversation_id: str
+    role: Literal["user", "assistant", "system"]
+    content: str
+    created_at: datetime
+    citations: list[dict[str, Any]] = Field(default_factory=list)
+    tool_calls: list[dict[str, Any]] = Field(default_factory=list)
+    thinking: str | None = None
+    usage: dict[str, Any] | None = None
+
+
+class ChatMessageListResponse(Contract):
+    items: list[ChatMessage] = Field(default_factory=list)
+    page: PageMeta = Field(default_factory=PageMeta)
+
+
 class ModelEventType(str, Enum):
     citation = "Citation"
     text_delta = "TextDelta"
@@ -1050,6 +1096,7 @@ class TranscriptEditRequest(Contract):
 
 
 class TranscriptNoteRequest(Contract):
+    update_existing: bool = False
     title: str = Field(min_length=1, max_length=200)
     folder: str | None = None
     include_timestamps: bool = True

backend/app/database/migrations.py

@@ -132,6 +132,33 @@ MIGRATIONS: list[str] = [
     """
     ALTER TABLE blocks ADD COLUMN embedding_local_only INTEGER NOT NULL DEFAULT 0;
     """,
+    # v7: application-owned chat conversations and messages, shared by web and desktop clients.
+    """
+    CREATE TABLE IF NOT EXISTS chat_conversations (
+        conversation_id TEXT PRIMARY KEY,
+        title TEXT NOT NULL,
+        created_at TEXT NOT NULL,
+        updated_at TEXT NOT NULL
+    );
+    CREATE INDEX IF NOT EXISTS idx_chat_conversations_updated
+        ON chat_conversations(updated_at DESC);
+
+    CREATE TABLE IF NOT EXISTS chat_messages (
+        message_id TEXT PRIMARY KEY,
+        conversation_id TEXT NOT NULL REFERENCES chat_conversations(conversation_id) ON DELETE CASCADE,
+        sequence INTEGER NOT NULL,
+        role TEXT NOT NULL,
+        content TEXT NOT NULL DEFAULT '',
+        thinking TEXT,
+        citations_json TEXT NOT NULL DEFAULT '[]',
+        tool_calls_json TEXT NOT NULL DEFAULT '[]',
+        usage_json TEXT,
+        created_at TEXT NOT NULL,
+        UNIQUE(conversation_id, sequence)
+    );
+    CREATE INDEX IF NOT EXISTS idx_chat_messages_conversation
+        ON chat_messages(conversation_id, sequence);
+    """,
 ]
 
 

backend/app/local_model_routes.py

@@ -1,15 +1,30 @@
+import asyncio
 from fastapi import APIRouter
+from app.services import model_diagnostics
 from app.local_models import manager
 from app.local_models.runtime import RuntimeConfig, configuration, configure, interpreter, runtime
 
 router = APIRouter(prefix="/api/local-models", tags=["Local models"])
 
 
+@router.get("/runtime-components/cuda")
+async def cuda_status():
+    from app.local_models import components
+    return await components.status()
+
+
+@router.post("/runtime-components/cuda", status_code=202)
+async def install_cuda():
+    from app.local_models import components
+    return await components.install()
+
+
 @router.get("")
 async def list_models():
-    return {**manager.describe(), "runtime_installed": interpreter().is_file(), "config": configuration(),
+    items, diagnostics = await asyncio.gather(asyncio.to_thread(manager.describe), asyncio.to_thread(model_diagnostics.recent))
+    return {**items, "runtime_installed": interpreter().is_file(), "config": configuration(),
             "active_models": list(runtime.active.values()), "queued_requests": len(runtime.waiters),
-            "last_inference": runtime.diagnostics[-1] if runtime.diagnostics else None}
+            "last_inference": diagnostics[-1] if diagnostics else None}
 
 
 @router.put("/config")
@@ -34,5 +49,5 @@ async def delete(key: str):
 
 @router.get("/diagnostics")
 async def diagnostics():
-    return {"items": runtime.diagnostics, "config": configuration(), "scope": "current_process",
+    return {"items": await asyncio.to_thread(model_diagnostics.recent), "config": configuration(), "scope": "application_last_200_attempts",
             "contains": "model_revision_device_timing_resources_only"}

backend/app/local_models/components.py

@@ -0,0 +1,111 @@
+"""User-triggered installation of the fixed optional CUDA runtime on Windows."""
+import asyncio
+import json
+import os
+import shutil
+import subprocess
+
+from app.config import BACKEND_DIR
+from app.errors import ApiError
+from app.local_models.process import ThreadedProcess
+
+ROOT = BACKEND_DIR / '.venv-models-cuda'
+state = {'status': 'unchecked', 'stage': '', 'cuda_available': None}
+task = None
+
+
+def ready():
+    return (ROOT / 'ready.json').is_file() and (ROOT / 'Scripts/python.exe').is_file()
+
+
+async def status():
+    global task
+    if state['status'] == 'unchecked':
+        state.update(status='checking', stage='检查已有 CUDA 组件')
+        task = asyncio.create_task(run(False))
+    return {**state, 'supported': os.name == 'nt', 'custom_interpreter': bool(os.getenv('APP_MODEL_PYTHON'))}
+
+
+async def install():
+    global task
+    from app.local_models.runtime import runtime
+    if os.name != 'nt':
+        raise ApiError(422, 'PLATFORM_UNSUPPORTED', '此安装入口目前支持 Windows。')
+    if task is not None and not task.done():
+        return await status()
+    if runtime.active or runtime.waiters:
+        raise ApiError(409, 'MODEL_IN_USE', '请等待本地模型任务结束后再安装组件。')
+    if state['status'] == 'installed':
+        return await status()
+    if not shutil.which('uv'):
+        raise ApiError(422, 'UV_NOT_INSTALLED', '后端未找到 uv，请先安装 uv 并重启后端。')
+    state.update(status='installing', stage='准备独立 CUDA 环境', error=None)
+    task = asyncio.create_task(run(True))
+    return await status()
+
+
+async def execute(args, timeout):
+    process = ThreadedProcess(args, env={**os.environ, 'PYTHONIOENCODING': 'utf-8'},
+                              limit=8192, creationflags=0x08000000 if os.name == 'nt' else 0)
+    process.stdin.close()
+    lines = []
+    try:
+        async with asyncio.timeout(timeout):
+            while line := await process.stdout.readline():
+                value = line.decode('utf-8', errors='replace').strip()
+                stages = {'COMPONENT:torch': '下载并安装 PyTorch CUDA（约 3 GB）',
+                          'COMPONENT:dependencies': '安装模型依赖', 'COMPONENT:verify': '验证运行组件'}
+                if value in stages:
+                    state['stage'] = stages[value]
+                lines = (lines + [value])[-4:]
+            await process.wait()
+            if process.returncode:
+                raise RuntimeError('component command failed')
+            return lines
+    finally:
+        if process.returncode is None:
+            if os.name == 'nt':
+                await asyncio.to_thread(subprocess.run, ['taskkill', '/PID', str(process.process.pid), '/T', '/F'],
+                                        stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL,
+                                        creationflags=0x08000000)
+            else:
+                process.kill()
+            await process.wait()
+        await process.close()
+
+
+async def run(download):
+    marker = ROOT / 'ready.json'
+    try:
+        if download:
+            marker.unlink(missing_ok=True)
+            await execute(['powershell.exe', '-NoProfile', '-NonInteractive', '-File',
+                           str(BACKEND_DIR / 'scripts/install-model-runtime.ps1'), '-Device', 'cuda',
+                           '-RuntimeDirectory', str(ROOT), '-QuietProgress'], 7200)
+        python = ROOT / 'Scripts/python.exe'
+        if not python.is_file():
+            state.update(status='not_installed', stage='尚未安装')
+            return
+        result = await execute([str(python), '-c',
+            'import json, torch, torchaudio, sentence_transformers, qwen_asr; '
+            'assert torch.version.cuda; '
+            'print(json.dumps({"torch":torch.__version__,"cuda_available":torch.cuda.is_available()}))'], 180)
+        info = json.loads(result[-1])
+        marker.write_text(json.dumps(info), encoding='utf-8')
+        state.update(status='installed', stage='组件已安装', error=None, **info)
+    except asyncio.CancelledError:
+        marker.unlink(missing_ok=True)
+        state.update(status='interrupted', stage='安装检查已中断，可重试')
+        raise
+    except Exception:
+        marker.unlink(missing_ok=True)
+        state.update(status='failed', stage='组件安装或验证失败',
+                     error='请检查网络、磁盘空间和 uv；可以重试。CPU 环境不受影响。')
+
+
+async def shutdown():
+    if task is not None and not task.done():
+        task.cancel()
+        await asyncio.gather(task, return_exceptions=True)
+    if state['status'] in {'checking', 'interrupted'}:
+        state['status'] = 'unchecked'

backend/app/local_models/manager.py

@@ -49,8 +49,20 @@ def task_key(key):
     return str(model_path(key)), key
 
 
+def disk_bytes(key):
+    total = 0
+    try:
+        root = model_path(key).resolve()
+        for path in root.rglob("*"):
+            if not path.is_symlink() and path.is_file() and path.resolve().is_relative_to(root):
+                total += path.stat().st_size
+    except OSError:
+        return None
+    return total
+
+
 def describe():
-    return {"items": [{**spec.public(), **read_state(key)} for key, spec in CATALOG.items()]}
+    return {"items": [{**spec.public(), **read_state(key), "disk_bytes": disk_bytes(key)} for key, spec in CATALOG.items()]}
 
 
 async def download(key):

backend/app/local_models/runtime.py

@@ -4,8 +4,10 @@ from __future__ import annotations
 import asyncio
 import json
 import os
+import time
 from contextlib import closing
 from contextvars import ContextVar
+from functools import wraps
 from pathlib import Path
 from typing import Literal
 
@@ -31,6 +33,18 @@ class RuntimeConfig(BaseModel):
 
 runtime_context = ContextVar("runtime_config", default=None)
 runtime_progress = ContextVar("runtime_progress", default=None)
+embedding_priority = ContextVar("embedding_priority", default=0)
+
+
+def background_embeddings(operation):
+    @wraps(operation)
+    async def wrapped(*args, **kwargs):
+        token = embedding_priority.set(20)
+        try:
+            return await operation(*args, **kwargs)
+        finally:
+            embedding_priority.reset(token)
+    return wrapped
 
 
 def configuration():
@@ -55,7 +69,11 @@ def configure(request):
     return request
 
 
-def interpreter():
+def interpreter(config=None):
+    from app.local_models import components
+    requested_device = (config or configuration()).device
+    if not os.getenv("APP_MODEL_PYTHON") and requested_device == "cuda" and components.ready():
+        return components.ROOT / "Scripts/python.exe"
     return Path(os.getenv("APP_MODEL_PYTHON", str(BACKEND_DIR / ".venv-models" / ("Scripts/python.exe" if os.name == "nt" else "bin/python"))))
 
 
@@ -75,32 +93,86 @@ class Runtime:
         return any(target in paths for paths in self.active_files.values())
 
     async def infer(self, key, operation, payload, *, priority=10):
-        if read_state(key)["status"] != "installed":
+        from app.services import model_diagnostics
-            raise ProviderError("LOCAL_MODEL_NOT_INSTALLED", "请先在模型配置中下载本地模型。")
+        config = configuration().model_copy(deep=True)
-        if not interpreter().is_file():
-            raise ProviderError("LOCAL_RUNTIME_NOT_INSTALLED", "请先运行本地模型 CPU/CUDA 安装脚本。")
-        config = configuration()
         self.counter += 1
         ticket = (priority, self.counter)
         self.waiters.append(ticket)
-        process = None
+        queued_at = time.monotonic()
-        attempt = None
+        reason = None
+        from app.services.usage_service import usage_context
+        from uuid import uuid4
+        context = dict(usage_context.get() or {})
+        context.setdefault("request_id", uuid4().hex)
+        usage_token = usage_context.set(context)
         try:
-            # One resident model at a time prevents overlapping CPU/GPU allocations.
             while self.active or ticket != min(self.waiters):
                 await asyncio.sleep(0.05)
             self.waiters.remove(ticket)
             self.active[ticket] = key
             self.active_files[ticket] = {str(Path(payload[name]).resolve()) for name in ("source", "reference") if payload.get(name)}
-            # Deletion may have occurred while this request was queued.
+            queue_seconds = time.monotonic() - queued_at
-            if read_state(key)["status"] != "installed":
+            # Keep the reservation while replacing a failed CUDA process with CPU.
-                raise ProviderError("LOCAL_MODEL_NOT_INSTALLED", "模型文件已被删除。")
+            for device in (["cuda", "cpu"] if config.device == "cuda" else ["cpu"]):
-            from app.services.usage_service import UsageAttempt
+                started = time.monotonic()
-            attempt = UsageAttempt("local-models", CATALOG[key].repository, "local", operation, source="local")
+                diagnostics = dict(model=CATALOG[key].repository, revision=CATALOG[key].revision,
+                    operation=operation, source="local", requested_device=config.device,
+                    attempted_device=device, queue_seconds=queue_seconds, fallback_reason=reason, request_id=context["request_id"])
+                try:
+                    result = await self._execute(key, operation, payload, config.model_copy(update={"device": device}), diagnostics)
+                    diagnostics.update(result.get("diagnostics", {}))
+                    diagnostics.update(requested_device=config.device, status="completed")
+                    if reason:
+                        diagnostics["fallback_reason"] = reason
+                    return result["result"]
+                except asyncio.CancelledError:
+                    diagnostics.update(status="cancelled", error_code="LOCAL_MODEL_CANCELLED")
+                    raise
+                except ProviderError as exc:
+                    diagnostics.update(status="failed", error_code=exc.code)
+                    if device == "cuda" and exc.code in {"LOCAL_CUDA_INIT_FAILED", "LOCAL_CUDA_OOM"}:
+                        reason = exc.code
+                        callback = runtime_progress.get()
+                        if callback:
+                            callback({"reset": True, "progress": 0})
+                        continue
+                    raise
+                except Exception:
+                    diagnostics.update(status="failed", error_code="LOCAL_MODEL_INVALID_RESPONSE")
+                    raise ProviderError("LOCAL_MODEL_INVALID_RESPONSE", "本地模型返回无效数据。") from None
+                finally:
+                    diagnostics["requested_device"] = config.device
+                    diagnostics["elapsed_seconds"] = time.monotonic() - started
+                    self.diagnostics.append(model_diagnostics.record(**diagnostics))
+                    self.diagnostics = self.diagnostics[-100:]
+        except asyncio.CancelledError:
+            if ticket not in self.active:
+                model_diagnostics.record(model=CATALOG[key].repository, operation=operation,
+                    source="local", status="cancelled", error_code="LOCAL_QUEUE_CANCELLED",
+                    requested_device=config.device, queue_seconds=time.monotonic() - queued_at)
+            raise
+        finally:
+            if ticket in self.waiters:
+                self.waiters.remove(ticket)
+            self.active.pop(ticket, None)
+            self.active_files.pop(ticket, None)
+            usage_context.reset(usage_token)
+
+    async def _execute(self, key, operation, payload, config, diagnostics):
+        if read_state(key)["status"] != "installed":
+            raise ProviderError("LOCAL_MODEL_NOT_INSTALLED", "请先下载本地模型。")
+        executable = interpreter(config)
+        if not executable.is_file():
+            raise ProviderError("LOCAL_RUNTIME_NOT_INSTALLED", "请先安装本地模型运行环境。")
+        from app.services.usage_service import UsageAttempt
+        attempt = UsageAttempt("local-models", CATALOG[key].repository, "local", operation, source="local")
+        diagnostics.update(attempt_id=attempt.attempt_id, request_id=attempt.request_id)
+        process = None
+        try:
             env = {**os.environ, "HF_HUB_OFFLINE": "1", "TRANSFORMERS_OFFLINE": "1",
                    "HF_HUB_DISABLE_TELEMETRY": "1", "OMP_NUM_THREADS": str(config.cpu_threads),
                    "PYTHONIOENCODING": "utf-8"}
-            args = (str(interpreter()), str(Path(__file__).with_name("worker.py")))
+            args = (str(executable), str(Path(__file__).with_name("worker.py")))
             options = {"env": env, "limit": 16 * 1024 * 1024,
                        **({"creationflags": 0x08000000} if os.name == "nt" else {})}
             try:
@@ -118,8 +190,6 @@ class Runtime:
                 process.stdin.close()
                 final = None
                 while line := await process.stdout.readline():
-                    if len(line) > 16 * 1024 * 1024:
-                        raise ProviderError("LOCAL_MODEL_INVALID_RESPONSE", "本地模型输出超限。")
                     message = json.loads(line)
                     if "progress" in message:
                         callback = runtime_progress.get()
@@ -137,26 +207,19 @@ class Runtime:
                 raise ProviderError("LOCAL_MODEL_PROCESS_FAILED", "本地模型进程退出，请检查依赖与资源预算。")
             if not isinstance(result, dict):
                 raise ProviderError("LOCAL_MODEL_INVALID_RESPONSE", "本地模型进程未返回有效结果。")
+            diagnostics.update(result.get("diagnostics", {}))
             if "error_code" in result:
                 raise ProviderError(result["error_code"], result.get("message", "本地推理失败。"))
             attempt.observe(result)
             attempt.completed = True
-            self.diagnostics.append({"model": CATALOG[key].repository, "revision": CATALOG[key].revision,
+            return result
-                                     **result.get("diagnostics", {})})
-            self.diagnostics = self.diagnostics[-100:]
-            return result["result"]
         finally:
-            if ticket in self.waiters:
-                self.waiters.remove(ticket)
             if process is not None and process.returncode is None:
                 process.kill()
                 await process.wait()
             if process is not None and hasattr(process, "close"):
                 await process.close()
-            self.active.pop(ticket, None)
+            attempt.persist()
-            self.active_files.pop(ticket, None)
-            if attempt:
-                attempt.persist()
 
 
 runtime = Runtime()
@@ -188,7 +251,7 @@ class LocalEmbedding:
         config = (self._config or configuration()).model_copy(deep=True)
         token = runtime_context.set(config)
         try:
-            return await runtime.infer(config.embedding_model, "embedding", {"texts": texts}, priority=0)
+            return await runtime.infer(config.embedding_model, "embedding", {"texts": texts}, priority=embedding_priority.get())
         finally:
             runtime_context.reset(token)
 

backend/app/local_models/worker.py

@@ -76,16 +76,25 @@ def voice_embedding(model, audio, device):
         return torch.nn.functional.normalize(vector, dim=0)
 
 
+class CudaInitializationError(RuntimeError):
+    pass
+
+
 def run(request):
     import torch
     import psutil
     config, payload = request["config"], request["payload"]
     torch.set_num_threads(config["cpu_threads"])
     requested = config["device"]
-    device = "cuda:0" if requested == "cuda" and torch.cuda.is_available() else "cpu"
+    try:
-    if device != "cpu":
+        device = "cuda:0" if requested == "cuda" and torch.cuda.is_available() else "cpu"
-        total = torch.cuda.get_device_properties(0).total_memory
+        if device != "cpu":
-        torch.cuda.set_per_process_memory_fraction(min(1.0, config["gpu_memory_limit_mb"] * 1024 ** 2 / total))
+            torch.cuda.init()
+            total = torch.cuda.get_device_properties(0).total_memory
+            torch.cuda.set_per_process_memory_fraction(min(1.0, config["gpu_memory_limit_mb"] * 1024 ** 2 / total))
+    except Exception as exc:
+        raise CudaInitializationError() from exc
+    request["_actual_device"] = device
     process = psutil.Process()
     peak = [0]
     stop = threading.Event()
@@ -102,6 +111,7 @@ def run(request):
     path, operation = request["model_path"], request["operation"]
     try:
         usage = {}
+        audio_seconds = None
         if operation == "embedding":
             from sentence_transformers import SentenceTransformer
             model = SentenceTransformer(path, device=device, local_files_only=True, trust_remote_code=False,
@@ -116,6 +126,7 @@ def run(request):
                 device_map=device, attn_implementation="sdpa", max_inference_batch_size=1, max_new_tokens=512)
             loaded = time.monotonic()
             audio = decode(payload["source"])
+            audio_seconds = len(audio) / 16000
             regions = speech_regions(audio)
             language = {"zh": "Chinese", "en": "English", "ja": "Japanese", "yue": "Cantonese"}.get(payload.get("language"), payload.get("language"))
             segments = []
@@ -154,7 +165,7 @@ def run(request):
             result = {"speakers": speakers}
         else:
             raise ValueError("Unknown inference operation")
-        return {"result": result, "usage": usage, "diagnostics": {"requested_device": requested, "actual_device": device,
+        return {"result": result, "usage": usage, "audio_seconds": audio_seconds, "diagnostics": {"requested_device": requested, "actual_device": device,
                 "fallback_reason": "CUDA_UNAVAILABLE" if requested == "cuda" and device == "cpu" else None,
                 "load_seconds": loaded - started, "inference_seconds": time.monotonic() - loaded,
                 "peak_memory_bytes": max(peak[0], process.memory_info().rss), "operation": operation}}
@@ -170,6 +181,16 @@ if __name__ == "__main__":
             response = run(request)
         except (ImportError, ModuleNotFoundError):
             response = {"error_code": "LOCAL_RUNTIME_DEPENDENCY_MISSING", "message": "本地模型运行依赖不完整，请重新运行安装脚本。"}
-        except Exception:
+        except Exception as exc:
-            response = {"error_code": "LOCAL_INFERENCE_FAILED", "message": "本地推理失败，请检查媒体格式、模型和设备配置。"}
+            # Only device failures allow the host to retry once in a fresh CPU process.
+            import torch
+            cuda_failure = isinstance(exc, CudaInitializationError)
+            cuda_oom = request.get("_actual_device") == "cuda:0" and isinstance(exc, torch.cuda.OutOfMemoryError)
+            if cuda_failure or cuda_oom:
+                response = {"error_code": "LOCAL_CUDA_OOM" if cuda_oom else "LOCAL_CUDA_INIT_FAILED",
+                            "message": "CUDA 运行失败，将释放进程并重试 CPU。"}
+            else:
+                response = {"error_code": "LOCAL_INFERENCE_FAILED", "message": "本地推理失败，请检查媒体格式、模型和设备配置。"}
+    if "error_code" in response:
+        response["diagnostics"] = {"requested_device": request["config"]["device"], "actual_device": request.get("_actual_device", "unknown")}
     sys.stdout.buffer.write((json.dumps(response, ensure_ascii=False, allow_nan=False) + "\n").encode("utf-8"))

backend/app/main.py

@@ -26,6 +26,8 @@ async def lifespan(_: FastAPI):
         yield
     finally:
         await transcription_service.shutdown()
+        from app.local_models import components
+        await components.shutdown()
         from app.local_models import manager
         for _, key in list(manager._downloads):
             await manager.cancel_download(key)

backend/app/media_routes.py

@@ -3,6 +3,7 @@ from __future__ import annotations
 
 import asyncio
 import json
+import hashlib
 from contextlib import closing
 from pathlib import Path
 from uuid import uuid4
@@ -22,14 +23,17 @@ MEDIA_SUFFIXES = {".wav", ".mp3", ".flac", ".ogg", ".m4a", ".mp4", ".webm", ".tx
 
 
 @router.post("/attachments", status_code=201)
-async def upload_attachment(request: Request, filename: str = Query(min_length=1, max_length=255)):
+async def upload_attachment(request: Request, filename: str = Query(min_length=1, max_length=255),
+                            idempotency_key: str | None = Header(None, min_length=16, max_length=100, pattern=r"^[a-zA-Z0-9_-]+$")):
     suffix = Path(filename).suffix.lower()
     if suffix not in MEDIA_SUFFIXES:
         raise ApiError(422, "UNSUPPORTED_MEDIA", "Unsupported attachment extension.")
-    attachment_id = f"media_{uuid4().hex}{suffix}"
+    identity = hashlib.sha256(idempotency_key.encode()).hexdigest() if idempotency_key else uuid4().hex
+    attachment_id = f"media_{identity}{suffix}"
     destination = attachment_path(attachment_id)
     destination.parent.mkdir(parents=True, exist_ok=True)
-    temporary = destination.with_suffix(destination.suffix + ".upload")
+    temporary = destination.with_suffix(destination.suffix + f".{uuid4().hex}.upload")
+    digest = hashlib.sha256()
     size = 0
     try:
         with temporary.open("xb") as stream:
@@ -37,10 +41,40 @@ async def upload_attachment(request: Request, filename: str = Query(min_length=1
                 size += len(chunk)
                 if size > MAX_UPLOAD_BYTES:
                     raise ApiError(413, "ATTACHMENT_TOO_LARGE", "Attachment exceeds 25 MiB.")
+                digest.update(chunk)
                 stream.write(chunk)
         if not size:
             raise ApiError(422, "EMPTY_ATTACHMENT", "Attachment is empty.")
-        temporary.replace(destination)
+        content_hash = digest.hexdigest()
+        if idempotency_key:
+            with closing(connect()) as conn:
+                conn.execute("CREATE TABLE IF NOT EXISTS media_upload_idempotency (idempotency_key TEXT PRIMARY KEY, attachment_id TEXT NOT NULL, filename TEXT NOT NULL, content_hash TEXT NOT NULL)")
+                conn.execute("BEGIN IMMEDIATE")
+                try:
+                    row = conn.execute("SELECT attachment_id,filename,content_hash FROM media_upload_idempotency WHERE idempotency_key=?", (idempotency_key,)).fetchone()
+                    if row:
+                        if row["filename"] != Path(filename).name or row["content_hash"] != content_hash:
+                            raise ApiError(409, "IDEMPOTENCY_CONFLICT", "同一上传标识不能用于不同附件。")
+                        existing = attachment_path(row["attachment_id"])
+                        if not existing.is_file() or hashlib.sha256(existing.read_bytes()).hexdigest() != content_hash:
+                            raise ApiError(409, "IDEMPOTENCY_EXPIRED", "该上传标识对应的附件已不存在，请开始一次新提交。")
+                        attachment_id = row["attachment_id"]
+                    else:
+                        if destination.exists() and hashlib.sha256(destination.read_bytes()).hexdigest() != content_hash:
+                            raise ApiError(409, "IDEMPOTENCY_CONFLICT", "同一上传标识不能用于不同附件。")
+                        if not destination.exists():
+                            temporary.replace(destination)
+                        conn.execute("INSERT INTO media_upload_idempotency VALUES (?,?,?,?)",
+                                     (idempotency_key, attachment_id, Path(filename).name, content_hash))
+                    conn.execute("COMMIT")
+                except BaseException:
+                    conn.execute("ROLLBACK")
+                    raise
+        elif destination.exists():
+            if hashlib.sha256(destination.read_bytes()).digest() != digest.digest():
+                raise ApiError(409, "IDEMPOTENCY_CONFLICT", "同一上传标识不能用于不同附件。")
+        else:
+            temporary.replace(destination)
     finally:
         temporary.unlink(missing_ok=True)
     return {"attachment_id": attachment_id, "filename": Path(filename).name, "size": size}

backend/app/provider_preview_routes.py

@@ -1,12 +1,67 @@
 from fastapi import APIRouter
-from pydantic import BaseModel
+from pydantic import BaseModel, Field
 from app.contracts import ProviderCreateRequest, ProviderConfig, ModelRequest, Message, MessageRole
 from app.providers.factory import ProviderFactory
-from app.request_overrides import apply_overrides
+from app.request_overrides import RequestOverride, apply_overrides
 
 router = APIRouter(prefix="/api/providers", tags=["Providers"])
 
 
+class RulesTransfer(BaseModel):
+    version: int = Field(default=1, ge=1, le=1)
+    request_overrides: list[RequestOverride] = Field(max_length=100)
+
+
+@router.post("/request-rules/validate")
+async def validate_rules(request: RulesTransfer):
+    return request
+
+
+class ProbeRequest(BaseModel):
+    provider: ProviderCreateRequest
+    stream: bool = True
+
+
+@router.post("/request-probe")
+async def probe(request: ProbeRequest):
+    """Explicit user-triggered inference; no vault context, tools or media uploads."""
+    import asyncio
+    from contextlib import aclosing
+    from app.container import container
+    from app.errors import ApiError
+    from app.providers.base import ProviderError
+    from app.providers.factory import UnsupportedProviderError
+    config = ProviderConfig(provider_id="request-probe", **request.provider.model_dump())
+    if not config.default_model:
+        raise ApiError(422, "MODEL_REQUIRED", "请填写要验证的模型 ID。")
+    try:
+        adapter = container.provider_factory.build(config)
+        model_request = ModelRequest(provider_id=config.provider_id, model=config.default_model,
+            messages=[Message(role=MessageRole.user, content="Reply with OK.")], max_tokens=32)
+        received = False
+        async with asyncio.timeout(45):
+            if request.stream:
+                async with aclosing(adapter.stream(model_request)) as events:
+                    async for event in events:
+                        if event.event.value in {"TextDelta", "ThinkingDelta"}:
+                            received = received or bool(str(event.data.get("text") or "").strip())
+                        if event.event.value == "Error":
+                            raise ProviderError("PROVIDER_PROBE_FAILED", "模型返回了错误事件。")
+            else:
+                response = await adapter.complete(model_request)
+                received = bool(response.text and response.text.strip())
+        if not received:
+            raise ApiError(422, "PROVIDER_EMPTY_RESPONSE", "请求未返回有效文本，不能标记验证通过。")
+    except ProviderError as exc:
+        raise ApiError(502, exc.code, "推理验证失败，请检查模型、凭据和自定义参数。") from exc
+    except TimeoutError as exc:
+        raise ApiError(504, "PROVIDER_TIMEOUT", "推理验证超时。") from exc
+    except UnsupportedProviderError as exc:
+        raise ApiError(422, "PROVIDER_TYPE_UNSUPPORTED", "该协议不支持推理验证。") from exc
+    return {"success": True, "stream": request.stream, "model": config.default_model,
+            "message": "当前请求配置已通过实际推理验证。"}
+
+
 class PreviewRequest(BaseModel):
     provider: ProviderCreateRequest
     stream: bool = True

backend/app/providers/routing.py

@@ -6,6 +6,8 @@ available only for explicitly injected tests and protocol fixtures.
 from __future__ import annotations
 
 import hashlib
+import asyncio
+import time
 import json
 import math
 from dataclasses import dataclass, field, replace
@@ -179,6 +181,7 @@ class ModelRoutingService:
         payload = apply_overrides(kwargs.get(field, {}), provider.request_overrides, capability)
         kwargs[field] = payload if field == "json" else {key: json.dumps(value) if isinstance(value, (dict, list, bool)) or value is None else value for key, value in payload.items()}
         attempt = UsageAttempt(binding.provider_id, binding.model, provider.provider_type.value, capability)
+        started = time.monotonic()
         try:
             async with httpx.AsyncClient(timeout=30, transport=self.transport) as client:
                 async with client.stream("POST", url, headers=headers, **kwargs) as response:
@@ -202,6 +205,11 @@ class ModelRoutingService:
             raise invalid_response() from exc
         finally:
             attempt.persist()
+            from app.services.model_diagnostics import record
+            task = asyncio.current_task()
+            status = "completed" if attempt.completed else ("cancelled" if task and task.cancelling() else "failed")
+            record(model=binding.model, operation=capability, source="api", status=status,
+                   attempt_id=attempt.attempt_id, request_id=attempt.request_id, elapsed_seconds=time.monotonic() - started)
         if not isinstance(data, dict) or data.get("error"):
             raise invalid_response()
         return data, url
@@ -255,6 +263,9 @@ class ModelRoutingService:
                                        model_id="api-" + hashlib.sha256(identity.encode()).hexdigest())
             except ProviderError as exc:
                 reason = exc.code
+                from app.services.model_diagnostics import record
+                record(model=binding.model, source="api", status="fallback", error_code=reason,
+                       fallback_reason=reason, operation="model_routing")
         from app.local_models.runtime import LocalEmbedding
         local_embedding = self.local_embedding.snapshot() if isinstance(self.local_embedding, LocalEmbedding) else self.local_embedding
         try:
@@ -314,6 +325,9 @@ class ModelRoutingService:
                 return RoutedTranscript(text=text, source="api", segments=segments)
             except ProviderError as exc:
                 reason = exc.code
+                from app.services.model_diagnostics import record
+                record(model=binding.model, source="api", status="fallback", error_code=reason,
+                       fallback_reason=reason, operation="model_routing")
         try:
             text = await self.local_speech.transcribe(source, language)
             if isinstance(text, RoutedTranscript):
@@ -346,6 +360,9 @@ class ModelRoutingService:
                 return SpeakerMatchResult(score=score, source="api")
             except ProviderError as exc:
                 reason = exc.code
+                from app.services.model_diagnostics import record
+                record(model=binding.model, source="api", status="fallback", error_code=reason,
+                       fallback_reason=reason, operation="model_routing")
         try:
             score = await self.local_speech.match(source, reference)
             if not finite_number(score) or not 0 <= score <= 1:

backend/app/routes.py

@@ -1,4 +1,5 @@
 import asyncio
+import json
 from collections.abc import AsyncIterator
 from contextlib import aclosing
 from datetime import datetime, timezone
@@ -15,6 +16,10 @@ from app.contracts import (
     AgentRunListResponse,
     AgentTraceResponse,
     ChatRequest,
+    ChatMessageListResponse,
+    Conversation,
+    ConversationCreateRequest,
+    ConversationListResponse,
     BenchmarkDatasetListResponse,
     BenchmarkEventType,
     BenchmarkKind,
@@ -321,6 +326,40 @@ async def clear_search_history() -> dict[str, list[str]]:
     return {"queries": []}
 
 
+@router.get("/chat/conversations", response_model=ConversationListResponse, tags=["Chat"])
+async def list_chat_conversations(
+    limit: int = Query(default=50, ge=1, le=100), offset: int = Query(default=0, ge=0)
+) -> ConversationListResponse:
+    from app.services import chat_history
+    items, total = chat_history.list_conversations(limit, offset)
+    return ConversationListResponse(items=items, page=PageMeta(total=total, limit=limit, offset=offset))
+
+
+@router.post("/chat/conversations", response_model=Conversation, status_code=201, tags=["Chat"])
+async def create_chat_conversation(request: ConversationCreateRequest) -> Conversation:
+    from app.services import chat_history
+    return chat_history.create(request.title, request.conversation_id)
+
+
+@router.get("/chat/conversations/{conversation_id}/messages", response_model=ChatMessageListResponse, tags=["Chat"])
+async def list_chat_messages(
+    conversation_id: str,
+    limit: int = Query(default=500, ge=1, le=1000),
+    offset: int = Query(default=0, ge=0),
+) -> ChatMessageListResponse:
+    from app.services import chat_history
+    items, total = chat_history.list_messages(conversation_id, limit, offset)
+    return ChatMessageListResponse(items=items, page=PageMeta(total=total, limit=limit, offset=offset))
+
+
+@router.delete("/chat/conversations/{conversation_id}", response_model=OperationResponse, tags=["Chat"])
+async def delete_chat_conversation(conversation_id: str) -> OperationResponse:
+    from app.services import chat_history
+    if not chat_history.delete(conversation_id):
+        raise ApiError(404, "CONVERSATION_NOT_FOUND", "conversation not found", {"conversation_id": conversation_id})
+    return OperationResponse(status="completed", resource_id=conversation_id, message="deleted")
+
+
 @router.post(
     "/chat",
     response_class=StreamingResponse,
@@ -333,14 +372,38 @@ async def clear_search_history() -> dict[str, list[str]]:
     tags=["Chat"],
 )
 async def chat(request: ChatRequest) -> StreamingResponse:
+    from app.services import chat_history
+
+    conversation_id = request.conversation_id
+    assistant_message_id = request.assistant_message_id or f"message_{uuid4().hex}"
+    if conversation_id:
+        user_message = next(
+            (message for message in reversed(request.messages) if message.role.value == "user" and message.content.strip()),
+            None,
+        )
+        if user_message is not None:
+            chat_history.append_message(
+                conversation_id,
+                message_id=request.user_message_id or f"message_{uuid4().hex}",
+                role="user",
+                content=user_message.content,
+                title=request.conversation_title or user_message.content[:30],
+            )
     provider = provider_or_404(request.provider_id)
 
     async def stream() -> AsyncIterator[str]:
         sequence = 0
+        assistant_content = ""
+        assistant_thinking = ""
+        citations: list[dict] = []
+        tool_calls: list[dict] = []
+        argument_buffers: dict[str, str] = {}
+        usage: dict | None = None
         try:
             from app.services.chat_context import prepare
-            grounded_request, citations = await prepare(request)
+            grounded_request, grounded_citations = await prepare(request)
-            for citation in citations:
+            for citation in grounded_citations:
+                citations.append(citation)
                 event = ModelEvent(event=ModelEventType.citation, sequence=sequence,
                                    data=citation, timestamp=utc_now())
                 sequence += 1
@@ -349,13 +412,58 @@ async def chat(request: ChatRequest) -> StreamingResponse:
                 async for event in events:
                     event = event.model_copy(update={"sequence": sequence})
                     sequence += 1
+                    if event.event == ModelEventType.text_delta:
+                        assistant_content += str(event.data.get("text", ""))
+                    elif event.event == ModelEventType.thinking_delta:
+                        assistant_thinking += str(event.data.get("text", ""))
+                    elif event.event == ModelEventType.tool_call_start:
+                        tool_calls.append({
+                            "tool_call_id": str(event.data.get("tool_call_id", "")),
+                            "name": str(event.data.get("name", "unknown")),
+                            "parameters": event.data.get("arguments") if isinstance(event.data.get("arguments"), dict) else {},
+                            "status": "running",
+                        })
+                    elif event.event == ModelEventType.tool_call_delta:
+                        call_id = str(event.data.get("tool_call_id", ""))
+                        call = next((item for item in tool_calls if item["tool_call_id"] == call_id), None)
+                        if call is not None:
+                            delta = event.data.get("arguments_delta")
+                            if isinstance(delta, str):
+                                argument_buffers[call_id] = argument_buffers.get(call_id, "") + delta
+                                try:
+                                    parsed_arguments = json.loads(argument_buffers[call_id])
+                                    if isinstance(parsed_arguments, dict):
+                                        call["parameters"] = parsed_arguments
+                                except ValueError:
+                                    pass
+                            arguments = event.data.get("arguments")
+                            if isinstance(arguments, dict):
+                                call["parameters"].update(arguments)
+                    elif event.event == ModelEventType.tool_call_end:
+                        call_id = str(event.data.get("tool_call_id", ""))
+                        call = next((item for item in tool_calls if item["tool_call_id"] == call_id), None)
+                        if call is not None:
+                            call["status"] = "completed"
+                    elif event.event == ModelEventType.usage:
+                        input_tokens = int(event.data.get("input_tokens", 0))
+                        output_tokens = int(event.data.get("output_tokens", 0))
+                        usage = {"input_tokens": input_tokens, "output_tokens": output_tokens,
+                                 "total_tokens": input_tokens + output_tokens}
+                    elif event.event == ModelEventType.error:
+                        if assistant_content:
+                            assistant_content += "\n\n"
+                        assistant_content += str(event.data.get("message", "Model generation failed."))
                     yield as_sse(event.event.value, event.model_dump_json())
         except Exception as exc:
+            failure_message = exc.message if isinstance(exc, ApiError) else "知识库检索或模型生成失败，请检查服务状态。"
+            if assistant_content:
+                assistant_content += "\n\n"
+            assistant_content += failure_message
             error = ModelEvent(
                 event=ModelEventType.error,
                 sequence=sequence,
                 data={"code": exc.code if isinstance(exc, ApiError) else "CHAT_FAILED",
-                      "message": exc.message if isinstance(exc, ApiError) else "知识库检索或模型生成失败，请检查服务状态。"},
+                      "message": failure_message},
                 timestamp=utc_now(),
             )
             done = ModelEvent(
@@ -364,6 +472,18 @@ async def chat(request: ChatRequest) -> StreamingResponse:
             )
             yield as_sse(error.event.value, error.model_dump_json())
             yield as_sse(done.event.value, done.model_dump_json())
+        finally:
+            if conversation_id and (assistant_content or assistant_thinking or citations or tool_calls):
+                chat_history.append_message(
+                    conversation_id,
+                    message_id=assistant_message_id,
+                    role="assistant",
+                    content=assistant_content,
+                    thinking=assistant_thinking or None,
+                    citations=citations,
+                    tool_calls=tool_calls,
+                    usage=usage,
+                )
 
     return StreamingResponse(stream(), media_type="text/event-stream")
 

backend/app/services/chat_history.py

@@ -0,0 +1,187 @@
+from __future__ import annotations
+
+from contextlib import closing
+from datetime import datetime, timezone
+import json
+import sqlite3
+from typing import Any
+from uuid import uuid4
+
+from app.contracts import ChatMessage, Conversation
+from app.database.db import connect, transaction
+from app.errors import ApiError
+
+
+def _now() -> datetime:
+    return datetime.now(timezone.utc)
+
+
+def _conversation(row) -> Conversation:
+    return Conversation(
+        conversation_id=row["conversation_id"],
+        title=row["title"],
+        created_at=datetime.fromisoformat(row["created_at"]),
+        updated_at=datetime.fromisoformat(row["updated_at"]),
+        message_count=row["message_count"],
+    )
+
+
+def _message(row) -> ChatMessage:
+    citations = json.loads(row["citations_json"])
+    for citation in citations:
+        if isinstance(citation.get("heading_path"), list):
+            citation["heading_path"] = " / ".join(str(part) for part in citation["heading_path"])
+    return ChatMessage(
+        message_id=row["message_id"],
+        conversation_id=row["conversation_id"],
+        role=row["role"],
+        content=row["content"],
+        thinking=row["thinking"],
+        citations=citations,
+        tool_calls=json.loads(row["tool_calls_json"]),
+        usage=json.loads(row["usage_json"]) if row["usage_json"] else None,
+        created_at=datetime.fromisoformat(row["created_at"]),
+    )
+
+
+def create(title: str, conversation_id: str | None = None) -> Conversation:
+    conversation_id = conversation_id or f"conversation_{uuid4().hex}"
+    now = _now().isoformat()
+    with closing(connect()) as conn, transaction(conn):
+        try:
+            conn.execute(
+                "INSERT INTO chat_conversations(conversation_id,title,created_at,updated_at) VALUES(?,?,?,?)",
+                (conversation_id, title.strip(), now, now),
+            )
+        except sqlite3.IntegrityError as exc:
+            raise ApiError(409, "CONVERSATION_ALREADY_EXISTS", "conversation already exists", {"conversation_id": conversation_id}) from exc
+    result = get(conversation_id)
+    assert result is not None
+    return result
+
+
+def get(conversation_id: str) -> Conversation | None:
+    with closing(connect()) as conn:
+        row = conn.execute(
+            """SELECT c.*, COUNT(m.message_id) AS message_count
+               FROM chat_conversations c LEFT JOIN chat_messages m USING(conversation_id)
+               WHERE c.conversation_id=? GROUP BY c.conversation_id""",
+            (conversation_id,),
+        ).fetchone()
+        return _conversation(row) if row else None
+
+
+def list_conversations(limit: int, offset: int) -> tuple[list[Conversation], int]:
+    with closing(connect()) as conn:
+        total = conn.execute("SELECT COUNT(*) FROM chat_conversations").fetchone()[0]
+        rows = conn.execute(
+            """SELECT c.*, COUNT(m.message_id) AS message_count
+               FROM chat_conversations c LEFT JOIN chat_messages m USING(conversation_id)
+               GROUP BY c.conversation_id ORDER BY c.updated_at DESC LIMIT ? OFFSET ?""",
+            (limit, offset),
+        ).fetchall()
+        return [_conversation(row) for row in rows], total
+
+
+def list_messages(conversation_id: str, limit: int, offset: int) -> tuple[list[ChatMessage], int]:
+    if get(conversation_id) is None:
+        raise ApiError(404, "CONVERSATION_NOT_FOUND", "conversation not found", {"conversation_id": conversation_id})
+    with closing(connect()) as conn:
+        total = conn.execute("SELECT COUNT(*) FROM chat_messages WHERE conversation_id=?", (conversation_id,)).fetchone()[0]
+        rows = conn.execute(
+            "SELECT * FROM chat_messages WHERE conversation_id=? ORDER BY sequence LIMIT ? OFFSET ?",
+            (conversation_id, limit, offset),
+        ).fetchall()
+        return [_message(row) for row in rows], total
+
+
+def delete(conversation_id: str) -> bool:
+    with closing(connect()) as conn, transaction(conn):
+        return conn.execute("DELETE FROM chat_conversations WHERE conversation_id=?", (conversation_id,)).rowcount > 0
+
+
+def append_message(
+    conversation_id: str,
+    *,
+    message_id: str,
+    role: str,
+    content: str,
+    title: str | None = None,
+    thinking: str | None = None,
+    citations: list[dict[str, Any]] | None = None,
+    tool_calls: list[dict[str, Any]] | None = None,
+    usage: dict[str, Any] | None = None,
+) -> None:
+    now = _now().isoformat()
+    clean_title = (title or "").strip() or content[:30].strip() or "New conversation"
+    with closing(connect()) as conn:
+        conn.execute("BEGIN IMMEDIATE")
+        try:
+            _append_message_in_transaction(
+                conn, conversation_id, message_id=message_id, role=role, content=content,
+                title=clean_title, thinking=thinking, citations=citations, tool_calls=tool_calls,
+                usage=usage, now=now,
+            )
+            conn.execute("COMMIT")
+        except BaseException:
+            if conn.in_transaction:
+                conn.execute("ROLLBACK")
+            raise
+
+
+def _append_message_in_transaction(
+    conn,
+    conversation_id: str,
+    *,
+    message_id: str,
+    role: str,
+    content: str,
+    title: str,
+    thinking: str | None,
+    citations: list[dict[str, Any]] | None,
+    tool_calls: list[dict[str, Any]] | None,
+    usage: dict[str, Any] | None,
+    now: str,
+) -> None:
+    conversation = conn.execute(
+        "SELECT 1 FROM chat_conversations WHERE conversation_id=?", (conversation_id,)
+    ).fetchone()
+    if conversation is None:
+        # A stream may finish after deletion. Check under BEGIN IMMEDIATE so
+        # deletion and assistant persistence cannot recreate an orphaned chat.
+        if role == "assistant":
+            return
+        conn.execute(
+            "INSERT INTO chat_conversations(conversation_id,title,created_at,updated_at) VALUES(?,?,?,?)",
+            (conversation_id, title, now, now),
+        )
+    count = conn.execute(
+        "SELECT COUNT(*) FROM chat_messages WHERE conversation_id=?", (conversation_id,)
+    ).fetchone()[0]
+    if count == 0:
+        conn.execute(
+            "UPDATE chat_conversations SET title=? WHERE conversation_id=?",
+            (title, conversation_id),
+        )
+    existing = conn.execute(
+        "SELECT conversation_id FROM chat_messages WHERE message_id=?", (message_id,)
+    ).fetchone()
+    if existing:
+        if existing["conversation_id"] != conversation_id:
+            raise ApiError(409, "MESSAGE_ID_CONFLICT", "message id belongs to another conversation")
+        return
+    sequence = conn.execute(
+        "SELECT COALESCE(MAX(sequence), -1) + 1 FROM chat_messages WHERE conversation_id=?",
+        (conversation_id,),
+    ).fetchone()[0]
+    conn.execute(
+        """INSERT INTO chat_messages(message_id,conversation_id,sequence,role,content,thinking,citations_json,tool_calls_json,usage_json,created_at)
+           VALUES(?,?,?,?,?,?,?,?,?,?)""",
+        (message_id, conversation_id, sequence, role, content, thinking,
+         json.dumps(citations or [], ensure_ascii=False), json.dumps(tool_calls or [], ensure_ascii=False),
+         json.dumps(usage, ensure_ascii=False) if usage is not None else None, now),
+    )
+    conn.execute(
+        "UPDATE chat_conversations SET updated_at=? WHERE conversation_id=?",
+        (now, conversation_id),
+    )

backend/app/services/media_notes.py

@@ -19,8 +19,10 @@ async def create_transcript_note(job_id, options):
         job = require_job(job_id)
         if job.status != "completed":
             raise ApiError(409, "TRANSCRIPT_NOT_READY", "Only completed transcripts can become notes.")
-        options_hash = hashlib.sha256(options.model_dump_json().encode()).hexdigest()
+        options_hash = hashlib.sha256(options.model_copy(update={"update_existing": False}).model_dump_json(exclude={"update_existing"}).encode()).hexdigest()
         with closing(connect()) as conn:
+            conn.execute("CREATE TABLE IF NOT EXISTS media_note_baselines (note_id TEXT PRIMARY KEY, content_hash TEXT NOT NULL)")
+            previous = conn.execute("SELECT m.note_id,b.content_hash FROM media_notes m LEFT JOIN media_note_baselines b ON b.note_id=m.note_id WHERE m.job_id=? AND m.options_hash=? ORDER BY m.revision DESC LIMIT 1", (job_id, options_hash)).fetchone()
             row = conn.execute("SELECT note_id FROM media_notes WHERE job_id=? AND revision=? AND options_hash=?",
                                (job_id, job.revision, options_hash)).fetchone()
         if row:
@@ -44,15 +46,34 @@ async def create_transcript_note(job_id, options):
         if job.local_only:
             # Persist the indexing policy in the Vault, including later rebuilds.
             lines = ["---", "embedding_local_only: true", "---", "", *lines]
-        try:
+        markdown = "\n".join(lines)
-            note = await note_service.create_note(title=title, markdown="\n".join(lines), folder=options.folder, tags=["转写"])
+        if options.update_existing:
-        except ApiError as exc:
+            if previous is None or previous[1] is None:
-            if exc.code != "RESOURCE_CONFLICT" or "note_id" not in exc.details:
+                raise ApiError(409, "NOTE_UPDATE_BASELINE_MISSING", "没有可安全更新的导出记录，请先创建新笔记。")
-                raise
+            current = await note_service.get_note(previous[0])
-            # Recover a crash between successful note creation and linking the job.
+            if current is None:
-            note = await note_service.get_note(exc.details["note_id"])
+                raise ApiError(404, "RESOURCE_NOT_FOUND", "已导出笔记不存在。")
-            if note is None or marker not in note.markdown:
+            # Recover a successful update if linking failed after the Vault write.
-                raise
+            if current.markdown == markdown:
+                note = current
+            else:
+                note = await note_service.update_note(previous[0], markdown=markdown, expected_content_hash=previous[1])
+        else:
+            note = await _create_note(title, markdown, options, marker)
         with closing(connect()) as conn, transaction(conn):
             conn.execute("INSERT OR IGNORE INTO media_notes VALUES (?,?,?,?)", (job_id, job.revision, options_hash, note.note_id))
+            conn.execute("INSERT OR REPLACE INTO media_note_baselines VALUES (?,?)", (note.note_id, hashlib.sha256(markdown.encode()).hexdigest()))
         return note
+
+
+async def _create_note(title, markdown, options, marker):
+    try:
+        note = await note_service.create_note(title=title, markdown=markdown, folder=options.folder, tags=["转写"])
+    except ApiError as exc:
+        if exc.code != "RESOURCE_CONFLICT" or "note_id" not in exc.details:
+            raise
+        # Recover a crash between successful note creation and linking the job.
+        note = await note_service.get_note(exc.details["note_id"])
+        if note is None or marker not in note.markdown:
+            raise
+    return note

backend/app/services/model_diagnostics.py

@@ -0,0 +1,37 @@
+"""Bounded, durable diagnostics. No payloads, paths, exception text or credentials."""
+import json
+import logging
+import math
+from contextlib import closing
+from datetime import datetime, timezone
+
+from app.database.db import connect, transaction
+
+TEXT = {"model", "revision", "operation", "source", "requested_device", "actual_device",
+        "attempted_device", "fallback_reason", "error_code", "status", "request_id", "attempt_id"}
+NUMBERS = {"load_seconds", "inference_seconds", "elapsed_seconds", "peak_memory_bytes", "queue_seconds"}
+
+
+def connection():
+    conn = connect()
+    conn.execute("CREATE TABLE IF NOT EXISTS model_diagnostics (id INTEGER PRIMARY KEY AUTOINCREMENT, record_json TEXT NOT NULL)")
+    return conn
+
+
+def record(**values):
+    safe = {key: value[:240] for key, value in values.items() if key in TEXT and isinstance(value, str)}
+    safe.update({key: value for key, value in values.items()
+                 if key in NUMBERS and type(value) in (float, int) and math.isfinite(value) and value >= 0})
+    safe["timestamp"] = datetime.now(timezone.utc).isoformat()
+    try:
+        with closing(connection()) as conn, transaction(conn):
+            conn.execute("INSERT INTO model_diagnostics(record_json) VALUES (?)", (json.dumps(safe),))
+            conn.execute("DELETE FROM model_diagnostics WHERE id NOT IN (SELECT id FROM model_diagnostics ORDER BY id DESC LIMIT 200)")
+    except Exception:
+        logging.getLogger(__name__).warning("Model diagnostic persistence failed")
+    return safe
+
+
+def recent():
+    with closing(connection()) as conn:
+        return [json.loads(row[0]) for row in conn.execute("SELECT record_json FROM model_diagnostics ORDER BY id")]

backend/app/services/note_service.py

@@ -17,7 +17,7 @@ from app.contracts import Note, NoteBlock, NoteSummary
 from app.database.db import connect, transaction
 from app.errors import ApiError
 from app.knowledge.parser import ParsedNote, parse_note
-from app.local_models.runtime import LocalEmbedding
+from app.local_models.runtime import LocalEmbedding, background_embeddings
 from app.retrieval import routed_vectors
 from app.retrieval.vectorstore import SqliteVecStore, VectorRecord
 from app.services.coordination import serialized_vault_mutation
@@ -77,6 +77,7 @@ def _delete_markdown(rel_path: str) -> None:
 PreparedIndex = tuple[list[list[float]], routed_vectors.RemoteEmbeddings | None]
 
 
+@background_embeddings
 async def prepare_note_index(parsed: ParsedNote, *, strict=False) -> PreparedIndex:
     """Compute vectors before opening a write transaction (including API I/O)."""
     texts = [block.content for block in parsed.blocks]
@@ -180,13 +181,18 @@ async def get_note(note_id: str) -> Note | None:
 
 @serialized_vault_mutation
 async def update_note(
-    note_id: str, *, title: str | None = None, markdown: str | None = None, tags: list[str] | None = None
+    note_id: str, *, title: str | None = None, markdown: str | None = None, tags: list[str] | None = None, expected_content_hash: str | None = None
 ) -> Note:
     record = repository.get_note_record(note_id)
     if record is None:
         raise ApiError(404, "RESOURCE_NOT_FOUND", "note not found", {"note_id": note_id})
 
     old_md = _read_markdown(record.file_path)
+    if expected_content_hash is not None:
+        import hashlib
+        if hashlib.sha256(old_md.encode()).hexdigest() != expected_content_hash:
+            raise ApiError(409, "NOTE_CONTENT_CONFLICT", "笔记已被编辑，请保留现有内容或导出为新笔记。")
+
     new_md = old_md if markdown is None else markdown
     # PATCH 语义：tags=None 保持原标签；[] 清空；非空列表替换（区别于 create 的 frontmatter 推导）
     effective_tags = record.tags if tags is None else tags

backend/app/services/transcription_service.py

@@ -134,6 +134,10 @@ async def _execute(job_id, request, routing=None):
     from app.contracts import TranscriptSegment
     token = runtime_context.set(RuntimeConfig.model_validate(job.model_snapshot.get("local_runtime", {})))
     def progress(message):
+        if message.get("reset"):
+            job.segments = []; job.progress = 0
+            save(job, "AttemptRestarted")
+            return
         job.progress = max(0.0, min(0.99, message["progress"]))
         job.segments.append(TranscriptSegment.model_validate(message["segment"]))
         save(job, "SegmentReady")

backend/app/services/usage_service.py

@@ -3,6 +3,7 @@ from __future__ import annotations
 
 import json
 import logging
+import math
 from contextlib import closing
 from contextvars import ContextVar
 from datetime import datetime, timezone
@@ -53,6 +54,7 @@ class UsageAttempt:
         self.capability, self.source = capability, source
         self.started_at = datetime.now(timezone.utc).isoformat()
         self.raw = {}
+        self.audio_seconds = None
         self.completed = False
         context = usage_context.get() or {}
         self.request_id = context.get("request_id") or uuid4().hex
@@ -61,6 +63,9 @@ class UsageAttempt:
     def observe(self, data):
         if not isinstance(data, dict):
             return
+        duration = data.get("audio_seconds", data.get("duration"))
+        if self.capability in {"transcription", "speaker_matching"} and type(duration) in (int, float) and math.isfinite(duration) and 0 <= duration <= 7200:
+            self.audio_seconds = max(self.audio_seconds or 0, duration)
         values = [data.get("usage"), (data.get("message") or {}).get("usage") if isinstance(data.get("message"), dict) else None,
                   (data.get("response") or {}).get("usage") if isinstance(data.get("response"), dict) else None]
         if self.protocol == "ollama":
@@ -87,7 +92,7 @@ class UsageAttempt:
             miss = inputs - hit
         if hit is not None and inputs is not None and hit > inputs:
             hit, miss = None, None
-        return dict(input_tokens=inputs, output_tokens=outputs,
+        return dict(audio_seconds=self.audio_seconds, input_tokens=inputs, output_tokens=outputs,
                     total_tokens=inputs + outputs if inputs is not None and outputs is not None else first("total_tokens"),
                     cache_hit_tokens=hit, cache_miss_tokens=miss, cache_write_tokens=write,
                     reasoning_tokens=first("output_tokens_details.reasoning_tokens", "completion_tokens_details.reasoning_tokens"))
@@ -103,7 +108,7 @@ class UsageAttempt:
 
 
 def aggregate(start, end, provider_id=None, model=None, source=None):
-    query = "SELECT counters_json,completed FROM model_usage WHERE started_at>=? AND started_at<?"
+    query = "SELECT counters_json,completed,capability FROM model_usage WHERE started_at>=? AND started_at<?"
     args = [start.astimezone(timezone.utc).isoformat(), end.astimezone(timezone.utc).isoformat()]
     for column, value in (("provider_id", provider_id), ("model", model), ("source", source)):
         if value:
@@ -115,8 +120,14 @@ def aggregate(start, end, provider_id=None, model=None, source=None):
     totals = {key: None for key in METRICS}
     coverage = {key: 0 for key in METRICS}
     hits, eligible_input, cache_requests = 0, 0, 0
+    audio_requests, audio_covered, audio_seconds = 0, 0, None
     for row in rows:
+        if row[2] in {"transcription", "speaker_matching"}:
+            audio_requests += 1
         counts = json.loads(row[0])
+        if counts.get("audio_seconds") is not None:
+            audio_covered += 1
+            audio_seconds = (audio_seconds or 0) + counts["audio_seconds"]
         for key in METRICS:
             if counts.get(key) is not None:
                 totals[key] = (totals[key] or 0) + counts[key]
@@ -125,7 +136,7 @@ def aggregate(start, end, provider_id=None, model=None, source=None):
             hits += counts["cache_hit_tokens"]
             eligible_input += counts["input_tokens"] if counts.get("input_tokens") is not None else counts["cache_hit_tokens"] + counts["cache_miss_tokens"]
             cache_requests += 1
-    return {"totals": totals, "coverage": coverage, "request_count": len(rows),
+    return {"audio_request_count": audio_requests, "audio_seconds": audio_seconds, "audio_covered_requests": audio_covered, "totals": totals, "coverage": coverage, "request_count": len(rows),
             "complete_requests": sum(row[1] for row in rows), "cache_covered_requests": cache_requests,
             "cache_hit_rate": hits / eligible_input if eligible_input else None,
             "options": [dict(row) for row in options], "start": start, "end": end,

backend/scripts/install-model-runtime.ps1

@@ -1,9 +1,12 @@
 param(
-    [ValidateSet('cpu', 'cuda')][string]$Device = 'cpu'
+    [ValidateSet('cpu', 'cuda')][string]$Device = 'cpu',
+    [string]$RuntimeDirectory = '',
+    [switch]$QuietProgress
 )
 $ErrorActionPreference = 'Stop'
+$uvOptions = if ($QuietProgress) { @('--quiet') } else { @() }
 $backendRoot = Split-Path $PSScriptRoot -Parent
-$runtimeRoot = Join-Path $backendRoot '.venv-models'
+$runtimeRoot = if ($RuntimeDirectory) { [IO.Path]::GetFullPath($RuntimeDirectory) } else { Join-Path $backendRoot '.venv-models' }
 $runtimePython = Join-Path $runtimeRoot 'Scripts/python.exe'
 if (!(Test-Path -LiteralPath $runtimePython)) {
     & uv venv --python 3.12 $runtimeRoot
@@ -11,9 +14,14 @@ if (!(Test-Path -LiteralPath $runtimePython)) {
 }
 # CPU is the default. CUDA wheels include the runtime, not the NVIDIA driver.
 $torchIndex = if ($Device -eq 'cuda') { 'https://download.pytorch.org/whl/cu128' } else { 'https://download.pytorch.org/whl/cpu' }
-& uv pip install --python $runtimePython --index-url $torchIndex 'torch==2.9.1' 'torchaudio==2.9.1'
+$wheelVariant = if ($Device -eq 'cuda') { 'cu128' } else { 'cpu' }
+# Pin the local version too: ==2.9.1 alone also accepts an already-installed CPU wheel.
+Write-Output 'COMPONENT:torch'
+& uv @uvOptions pip install --python $runtimePython --index-url $torchIndex "torch==2.9.1+$wheelVariant" "torchaudio==2.9.1+$wheelVariant"
 if ($LASTEXITCODE -ne 0) { throw 'PyTorch 安装失败' }
-& uv pip install --python $runtimePython -r (Join-Path $PSScriptRoot 'model-requirements.lock') -c (Join-Path $PSScriptRoot 'model-requirements.txt')
+Write-Output 'COMPONENT:dependencies'
+& uv @uvOptions pip install --python $runtimePython -r (Join-Path $PSScriptRoot 'model-requirements.lock') -c (Join-Path $PSScriptRoot 'model-requirements.txt')
 if ($LASTEXITCODE -ne 0) { throw '模型依赖安装失败' }
+Write-Output 'COMPONENT:verify'
 & $runtimePython -c 'import torch; print({"torch":torch.__version__,"cuda_available":torch.cuda.is_available()})'
 if ($LASTEXITCODE -ne 0) { throw '模型运行环境检查失败' }

backend/tests/test_chat_history.py

@@ -0,0 +1,121 @@
+import asyncio
+from datetime import datetime, timezone
+from types import SimpleNamespace
+
+from fastapi.testclient import TestClient
+import pytest
+
+from app.contracts import ChatRequest, ModelEvent, ModelEventType
+from app.main import app
+from app.services import chat_history
+
+
+def test_chat_history_survives_new_connections_and_deletes_messages() -> None:
+    conversation = chat_history.create("Persistent chat", "conversation-1")
+    chat_history.append_message(
+        conversation.conversation_id,
+        message_id="user-1",
+        role="user",
+        content="question",
+    )
+    chat_history.append_message(
+        conversation.conversation_id,
+        message_id="assistant-1",
+        role="assistant",
+        content="answer",
+        citations=[{"note_id": "note-1", "heading_path": ["Heading"]}],
+        usage={"input_tokens": 2, "output_tokens": 1, "total_tokens": 3},
+    )
+
+    listed, total = chat_history.list_conversations(50, 0)
+    messages, message_total = chat_history.list_messages("conversation-1", 50, 0)
+    assert total == 1
+    assert listed[0].message_count == 2
+    assert message_total == 2
+    assert messages[1].citations[0]["note_id"] == "note-1"
+    assert messages[1].usage["total_tokens"] == 3
+
+    assert chat_history.delete("conversation-1") is True
+    assert chat_history.list_conversations(50, 0)[1] == 0
+
+
+def test_chat_stream_persists_user_and_assistant_messages(monkeypatch) -> None:
+    from app import routes
+
+    class Adapter:
+        async def stream(self, _request):
+            now = datetime.now(timezone.utc)
+            yield ModelEvent(event=ModelEventType.text_delta, data={"text": "persisted answer"}, timestamp=now)
+            yield ModelEvent(event=ModelEventType.usage, data={"input_tokens": 4, "output_tokens": 2}, timestamp=now)
+            yield ModelEvent(event=ModelEventType.done, timestamp=now)
+
+    monkeypatch.setattr(routes, "provider_or_404", lambda _provider_id: SimpleNamespace(adapter=Adapter()))
+    payload = {
+        "provider_id": "configured",
+        "model": "model",
+        "conversation_id": "conversation-stream",
+        "user_message_id": "user-stream",
+        "assistant_message_id": "assistant-stream",
+        "conversation_title": "Persist this",
+        "use_rag": False,
+        "messages": [{"role": "user", "content": "question"}],
+    }
+    with TestClient(app) as client:
+        with client.stream("POST", "/api/chat", json=payload) as response:
+            assert response.status_code == 200
+            assert "persisted answer" in "".join(response.iter_text())
+        messages = client.get("/api/chat/conversations/conversation-stream/messages").json()["items"]
+        conversations = client.get("/api/chat/conversations").json()["items"]
+        assert [message["content"] for message in messages] == ["question", "persisted answer"]
+        assert messages[1]["usage"]["total_tokens"] == 6
+        assert conversations[0]["title"] == "Persist this"
+        assert conversations[0]["message_count"] == 2
+
+
+def test_chat_conversation_crud_api() -> None:
+    with TestClient(app) as client:
+        created = client.post("/api/chat/conversations", json={"conversation_id": "crud", "title": "CRUD"})
+        assert created.status_code == 201
+        assert client.get("/api/chat/conversations").json()["page"]["total"] == 1
+        assert client.get("/api/chat/conversations/crud/messages").json()["items"] == []
+        assert client.delete("/api/chat/conversations/crud").status_code == 200
+        missing = client.get("/api/chat/conversations/crud/messages")
+        assert missing.status_code == 404
+        assert missing.json()["error"]["code"] == "CONVERSATION_NOT_FOUND"
+
+
+@pytest.mark.parametrize("close_early", [True, False])
+@pytest.mark.parametrize("deleted", [True, False])
+def test_stream_finalization_respects_conversation_deletion(monkeypatch, close_early, deleted) -> None:
+    from app import routes
+
+    class Adapter:
+        async def stream(self, _request):
+            now = datetime.now(timezone.utc)
+            yield ModelEvent(event=ModelEventType.text_delta, data={"text": "partial answer"}, timestamp=now)
+            yield ModelEvent(event=ModelEventType.done, timestamp=now)
+
+    monkeypatch.setattr(routes, "provider_or_404", lambda _: SimpleNamespace(adapter=Adapter()))
+
+    async def scenario():
+        response = await routes.chat(ChatRequest(
+            provider_id="configured", model="model", conversation_id="stream",
+            use_rag=False, messages=[{"role": "user", "content": "question"}],
+        ))
+        await anext(response.body_iterator)
+        if deleted:
+            assert chat_history.delete("stream")
+        if close_early:
+            await response.body_iterator.aclose()
+        else:
+            async for _ in response.body_iterator:
+                pass
+        if deleted:
+            assert chat_history.get("stream") is None
+            assert chat_history.list_conversations(50, 0)[1] == 0
+        else:
+            messages, total = chat_history.list_messages("stream", 50, 0)
+            assert total == 2
+            assert [message.content for message in messages] == ["question", "partial answer"]
+
+    asyncio.run(scenario())

backend/tests/test_local_models.py

@@ -47,7 +47,7 @@ def test_local_model_missing_is_explicit():
 def test_cancel_reaps_active_model_process(monkeypatch):
     import app.local_models.runtime as module
     monkeypatch.setattr(module,'read_state',lambda key:{'status':'installed'})
-    monkeypatch.setattr(module,'interpreter',lambda:Path(sys.executable))
+    monkeypatch.setattr(module,'interpreter',lambda *_:Path(sys.executable))
     class Input:
         def write(self, value):
             request = json.loads(value)
@@ -92,7 +92,7 @@ def test_subprocess_fallback_runs_and_reaps_real_worker(monkeypatch, tmp_path, c
     import app.local_models.process as process_module
 
     monkeypatch.setattr(module, 'read_state', lambda key: {'status': 'installed'})
-    monkeypatch.setattr(module, 'interpreter', lambda: Path(sys.executable))
+    monkeypatch.setattr(module, 'interpreter', lambda *_: Path(sys.executable))
     worker = tmp_path / 'worker.py'
     worker.write_text(
         'import json,sys,time\n'

backend/tests/test_multimodal_finalization.py

@@ -0,0 +1,223 @@
+"""Finalization regressions: device recovery, durable facts and guarded writes."""
+import asyncio
+import json
+import sys
+from contextlib import closing
+from datetime import datetime, timedelta, timezone
+from pathlib import Path
+
+import pytest
+from fastapi.testclient import TestClient
+
+from app.errors import ApiError
+from app.providers.base import ProviderError
+
+
+@pytest.mark.parametrize('code,retries', [('LOCAL_CUDA_OOM', True), ('LOCAL_CUDA_INIT_FAILED', True),
+                                          ('LOCAL_INFERENCE_FAILED', False), ('LOCAL_RUNTIME_DEPENDENCY_MISSING', False)])
+def test_cuda_retries_only_device_failures_in_reaped_process(monkeypatch, code, retries):
+    import app.local_models.runtime as module
+    from app.services import model_diagnostics
+    from app.services.usage_service import connection
+    monkeypatch.setattr(module, 'configuration', lambda: module.RuntimeConfig(device='cuda'))
+    monkeypatch.setattr(module, 'read_state', lambda key: {'status': 'installed'})
+    monkeypatch.setattr(module, 'interpreter', lambda *_: Path(sys.executable))
+    events = []
+
+    class Process:
+        def __init__(self):
+            from types import SimpleNamespace
+            self.stdin = SimpleNamespace(write=self.write, drain=self.drain, close=lambda: None)
+            self.stdout = asyncio.StreamReader()
+            self.returncode = None
+            self.device = None
+        def write(self, raw):
+            self.device = json.loads(raw)['config']['device']
+            events.append('start-' + self.device)
+            result = {'error_code': code} if self.device == 'cuda' else {'result': [[1, 0]], 'usage': {'input_tokens': 2}, 'diagnostics': {'actual_device': 'cpu'}}
+            self.stdout.feed_data((json.dumps(result) + '\n').encode())
+            self.stdout.feed_eof()
+        async def drain(self):
+            pass
+        async def close(self):
+            pass
+        async def wait(self):
+            self.returncode = 0
+            events.append('reaped-' + self.device)
+        def kill(self):
+            self.returncode = -9
+
+    async def spawn(*args, **kwargs):
+        if events:
+            assert events[-1] == 'reaped-cuda'
+        return Process()
+    monkeypatch.setattr(module.asyncio, 'create_subprocess_exec', spawn)
+
+    async def scenario():
+        runtime = module.Runtime()
+        if retries:
+            assert await runtime.infer('bekko', 'embedding', {'texts': ['private text']}) == [[1, 0]]
+        else:
+            with pytest.raises(ProviderError) as error:
+                await runtime.infer('bekko', 'embedding', {'texts': ['private text']})
+            assert error.value.code == code
+        assert not runtime.active and not runtime.waiters
+    asyncio.run(scenario())
+    assert events == (['start-cuda', 'reaped-cuda', 'start-cpu', 'reaped-cpu'] if retries else ['start-cuda', 'reaped-cuda'])
+    records = model_diagnostics.recent()
+    assert records[0]['error_code'] == code
+    assert 'private text' not in json.dumps(records)
+    if retries:
+        assert records[-1]['requested_device'] == 'cuda' and records[-1]['actual_device'] == 'cpu'
+        assert records[-1]['fallback_reason'] == code
+        assert records[0]['request_id'] == records[1]['request_id']
+        assert records[0]['attempt_id'] != records[1]['attempt_id']
+    with closing(connection()) as conn:
+        assert conn.execute('SELECT COUNT(*) FROM model_usage').fetchone()[0] == (2 if retries else 1)
+
+
+def test_cpu_failure_does_not_loop_and_interactive_precedes_index(monkeypatch):
+    import app.local_models.runtime as module
+    async def scenario():
+        runtime = module.Runtime()
+        entered, release = asyncio.Event(), asyncio.Event()
+        order = []
+        async def execute(key, operation, payload, config, diagnostics):
+            order.append(payload['name'])
+            if payload['name'] == 'running':
+                entered.set()
+                await release.wait()
+            return {'result': []}
+        monkeypatch.setattr(runtime, '_execute', execute)
+        first = asyncio.create_task(runtime.infer('bekko', 'embedding', {'name': 'running'}))
+        await entered.wait()
+        background = asyncio.create_task(runtime.infer('bekko', 'embedding', {'name': 'index'}, priority=20))
+        query = asyncio.create_task(runtime.infer('bekko', 'embedding', {'name': 'query'}, priority=0))
+        await asyncio.sleep(0)
+        release.set()
+        await asyncio.gather(first, background, query)
+        assert order == ['running', 'query', 'index']
+        calls = []
+        async def failed(key, operation, payload, config, diagnostics):
+            calls.append(config.device)
+            raise ProviderError('LOCAL_CUDA_OOM', 'simulated')
+        monkeypatch.setattr(runtime, '_execute', failed)
+        monkeypatch.setattr(module, 'configuration', lambda: module.RuntimeConfig(device='cuda'))
+        with pytest.raises(ProviderError):
+            await runtime.infer('bekko', 'embedding', {})
+        assert calls == ['cuda', 'cpu'] and not runtime.active
+    asyncio.run(scenario())
+
+
+def test_durable_diagnostics_are_bounded_and_disk_size_is_real():
+    from app.services import model_diagnostics
+    from app.local_models import manager
+    for index in range(205):
+        model_diagnostics.record(model='bekko', status='failed', error_code='TEST', payload='secret', elapsed_seconds=index)
+    records = model_diagnostics.recent()
+    assert len(records) == 200 and records[0]['elapsed_seconds'] == 5
+    assert 'secret' not in json.dumps(records)
+    path = manager.model_path('bekko')
+    path.mkdir(parents=True)
+    (path / 'weights.partial').write_bytes(b'1234567')
+    assert manager.disk_bytes('bekko') == 7
+
+
+def test_upload_key_replay_and_content_conflict():
+    from app.main import app
+    with TestClient(app) as client:
+        headers = {'Idempotency-Key': 'stable-upload-123456'}
+        first = client.post('/api/media/attachments?filename=lecture.txt', content=b'original', headers=headers)
+        again = client.post('/api/media/attachments?filename=lecture.txt', content=b'original', headers=headers)
+        assert first.status_code == again.status_code == 201
+        assert first.json()['attachment_id'] == again.json()['attachment_id']
+        assert client.post('/api/media/attachments?filename=lecture.txt', content=b'changed', headers=headers).status_code == 409
+        changed_name = client.post('/api/media/attachments?filename=lecture.md', content=b'original', headers=headers)
+        assert changed_name.status_code == 409 and changed_name.json()['error']['code'] == 'IDEMPOTENCY_CONFLICT'
+        assert client.get('/api/media/attachments/' + first.json()['attachment_id']).content == b'original'
+
+
+def test_updated_transcript_note_keeps_identity_and_rejects_user_edits():
+    from app.contracts import TranscriptNoteRequest, TranscriptEditRequest, IndexRebuildRequest
+    from app.services import transcription_service as jobs, note_service, index_service
+    from app.services.media_notes import create_transcript_note
+    from app.services.attachment_service import attachment_path
+    path = attachment_path('lecture.txt')
+    path.parent.mkdir(parents=True, exist_ok=True)
+    path.write_text('original', encoding='utf-8')
+    async def scenario():
+        job = await jobs.create_transcription('lecture.txt', local_only=True)
+        options = TranscriptNoteRequest(title='Lecture')
+        first = await create_transcript_note(job.job_id, options)
+        await index_service.rebuild(IndexRebuildRequest())
+        jobs.edit(job.job_id, TranscriptEditRequest(revision=1, text='revised'))
+        update = options.model_copy(update={'update_existing': True})
+        second = await create_transcript_note(job.job_id, update)
+        assert first.note_id == second.note_id and 'revised' in second.markdown
+        assert 'embedding_local_only: true' in second.markdown
+        again = await create_transcript_note(job.job_id, update)
+        assert again.note_id == first.note_id
+        await note_service.update_note(first.note_id, markdown='User edits')
+        jobs.edit(job.job_id, TranscriptEditRequest(revision=2, text='third revision'))
+        with pytest.raises(ApiError) as error:
+            await create_transcript_note(job.job_id, update)
+        assert error.value.code == 'NOTE_CONTENT_CONFLICT'
+        assert (await note_service.get_note(first.note_id)).markdown == 'User edits'
+        copy = await create_transcript_note(job.job_id, options)
+        assert copy.note_id != first.note_id
+    asyncio.run(scenario())
+
+
+def test_audio_usage_is_separate_and_unknown_durations_stay_null():
+    from app.services.usage_service import UsageAttempt, aggregate
+    now = datetime.now(timezone.utc)
+    first = UsageAttempt('local', 'asr', 'local', 'transcription', source='local')
+    first.observe({'audio_seconds': 2.25, 'usage': {}})
+    first.persist(); first.persist()
+    unknown = UsageAttempt('remote', 'asr', 'openai_compatible', 'transcription')
+    unknown.persist()
+    result = aggregate(now - timedelta(days=1), now + timedelta(days=1))
+    assert result['audio_request_count'] == 2 and result['audio_covered_requests'] == 1
+    assert result['audio_seconds'] == 2.25 and result['totals']['input_tokens'] is None
+    remote = aggregate(now - timedelta(days=1), now + timedelta(days=1), source='api')
+    assert remote['audio_seconds'] is None
+
+
+def test_request_rule_import_rejects_credentials_and_host_fields():
+    from app.main import app
+    with TestClient(app) as client:
+        path = '/api/providers/request-rules/validate'
+        body = {'version': 1, 'request_overrides': [{'body': {'enable_thinking': False}}]}
+        assert client.post(path, json=body).status_code == 200
+        for bad in ({'api_key': 'secret'}, {'nested': {'authorization': 'secret'}}, {'stream': False}):
+            body['request_overrides'][0]['body'] = bad
+            assert client.post(path, json=body).status_code == 422
+
+
+@pytest.mark.parametrize('stream', [False, True])
+def test_inference_probe_uses_adapter_body_and_no_vault_context(monkeypatch, stream):
+    import httpx
+    from app.container import container
+    from app.main import app
+    original = container.provider_factory.build
+    requests = []
+    def respond(request):
+        data = json.loads(request.content)
+        requests.append(data)
+        assert data['enable_thinking'] is False and data['stream'] == stream
+        assert data['messages'] == [{'role': 'user', 'content': 'Reply with OK.'}]
+        assert not data.get('tools')
+        if stream:
+            return httpx.Response(200, text='data: {"choices":[{"delta":{"content":"OK"},"finish_reason":null}]}\n\ndata: [DONE]\n\n')
+        return httpx.Response(200, json={'choices': [{'message': {'role': 'assistant', 'content': 'OK'}, 'finish_reason': 'stop'}]})
+    def build(config):
+        adapter = original(config)
+        adapter.transport = httpx.MockTransport(respond)
+        return adapter
+    monkeypatch.setattr(container.provider_factory, 'build', build)
+    with TestClient(app) as client:
+        response = client.post('/api/providers/request-probe', json={'stream': stream, 'provider': {
+            'name': 'Probe', 'provider_type': 'openai_compatible', 'base_url': 'https://fixture.invalid/v1',
+            'default_model': 'test', 'request_overrides': [{'body': {'enable_thinking': False}}]}})
+        assert response.status_code == 200, response.text
+        assert len(requests) == 1

backend/tests/test_runtime_components.py

@@ -0,0 +1,86 @@
+import asyncio
+import json
+import os
+
+import pytest
+
+from app.errors import ApiError
+from app.local_models import components, runtime
+
+
+@pytest.fixture(autouse=True)
+def isolate(monkeypatch, tmp_path):
+    monkeypatch.setattr(components, 'ROOT', tmp_path / 'cuda')
+    monkeypatch.setattr(components, 'state', {'status': 'unchecked', 'stage': '', 'cuda_available': None})
+    monkeypatch.setattr(components, 'task', None)
+
+
+def test_status_checks_without_installing_and_detects_existing_cuda(monkeypatch):
+    python = components.ROOT / 'Scripts/python.exe'
+    python.parent.mkdir(parents=True)
+    python.touch()
+    calls = []
+    async def execute(args, timeout):
+        calls.append(args)
+        return [json.dumps({'torch': '2.9.1+cu128', 'cuda_available': True})]
+    monkeypatch.setattr(components, 'execute', execute)
+    async def scenario():
+        assert (await components.status())['status'] == 'checking'
+        await components.task
+        assert (await components.status())['status'] == 'installed'
+        assert len(calls) == 1 and calls[0][0] == str(python)
+        assert components.ready()
+    asyncio.run(scenario())
+
+
+@pytest.mark.skipif(os.name != 'nt', reason='Windows installer')
+def test_install_deduplicates_and_failure_can_retry(monkeypatch):
+    monkeypatch.setattr(components.shutil, 'which', lambda name: 'uv.exe')
+    async def scenario():
+        entered, release = asyncio.Event(), asyncio.Event()
+        calls = []
+        async def execute(args, timeout):
+            calls.append(args)
+            entered.set()
+            await release.wait()
+            raise RuntimeError('private exception')
+        monkeypatch.setattr(components, 'execute', execute)
+        await components.install()
+        await entered.wait()
+        first = components.task
+        await components.install()
+        assert first is components.task
+        release.set()
+        await first
+        assert components.state['status'] == 'failed'
+        assert 'private exception' not in str(components.state)
+        await components.install()
+        await components.task
+        assert len(calls) == 2 and '-RuntimeDirectory' in calls[0]
+        assert not components.ready()
+    asyncio.run(scenario())
+
+
+@pytest.mark.skipif(os.name != 'nt', reason='Windows installer')
+def test_install_refuses_active_inference(monkeypatch):
+    monkeypatch.setattr(runtime.runtime, 'active', {1: 'bekko'})
+    async def scenario():
+        with pytest.raises(ApiError) as exc:
+            await components.install()
+        assert exc.value.code == 'MODEL_IN_USE'
+    asyncio.run(scenario())
+
+
+def test_interpreter_keeps_cpu_default_and_respects_explicit_override(monkeypatch):
+    monkeypatch.delenv('APP_MODEL_PYTHON', raising=False)
+    python = components.ROOT / 'Scripts/python.exe'
+    python.parent.mkdir(parents=True)
+    python.touch()
+    (components.ROOT / 'ready.json').write_text('{}')
+    monkeypatch.setattr(runtime, 'configuration', lambda: runtime.RuntimeConfig(device='cpu'))
+    assert runtime.interpreter() != python
+    # A queued attempt keeps its frozen device even after the saved setting changes.
+    assert runtime.interpreter(runtime.RuntimeConfig(device='cuda')) == python
+    assert runtime.interpreter(runtime.RuntimeConfig(device='cpu')) != python
+    monkeypatch.setenv('APP_MODEL_PYTHON', 'explicit-python.exe')
+    assert str(runtime.interpreter()) == 'explicit-python.exe'

docs/README.md

@@ -2,6 +2,10 @@
 
 本目录集中保存团队开发期间需要长期维护的架构、接口、实现、协作和问题复盘文档。文档按用途分类，避免设计约束、开发记录与故障复盘混放。
 
+当前文档基线为 2026-09-05：第一阶段和第二阶段 A～F 工程范围已经合并到 `main`，当前可运行形态仍为 Vue/Vite Web 前端与 FastAPI AI Core。Tauri/Rust Host、Stronghold、原生多 Vault 文件系统、生产级 MCP 沙箱和 Sync Server 尚未接入。
+
+仓库入口文档：[项目 README](../README.md)、[前端 README](../frontend/README.md)、[后端 README](../backend/README.md)。
+
 ## 目录分类
 
 | 目录 | 内容 | 适用场景 |
@@ -29,7 +33,7 @@
 ## development：开发说明
 
 - [多模态管线与模型运行开发说明](development/多模态管线与模型运行开发说明.md)
-
+- [阶段 F 收尾验收记录](development/阶段F收尾验收记录.md)
 - [AI Core 与 Agent Core 开发说明](development/AI-Core与Agent-Core开发说明.md)
 - [Knowledge 与 Retrieval Core 开发说明](development/Knowledge与Retrieval-Core开发说明.md)
 - [Benchmark 开发说明](development/Benchmark开发说明.md)

docs/architecture/AI笔记软件技术栈说明-团队版-v2.3.md

@@ -5,7 +5,7 @@
 > 适用范围：桌面客户端、本地知识库、RAG、Agent、Skill、多模型接入、多模态处理与可选云同步  
 > 目标读者：前端、Rust 桌面端、Python AI Core、算法、测试与后续接手项目的开发成员
 
-> 实施状态更新：2026-09-04。本文同时包含目标架构、当前实现和第二阶段接口基线。第一阶段已完成 Vue Web 联调前端、FastAPI、Knowledge/Retrieval、Agent/Tool/Permission、Skill/Plugin 声明式运行时、Mock/OpenAI-Compatible/Ollama Provider、DeepSeek/OpenAI 预设、模型发现及开发阶段 Fernet 凭据存储。Web Workspace 已通过 FastAPI 接入后端配置的真实单 Vault；第二阶段 Agent Trace 持久化、分页快照、可恢复 SSE、stdio MCP Bridge、隔离 Plugin Host、Plugin Command 与 Plugin Settings/Secret Contract 已完成。阶段 E 已完成 Responses/Anthropic 协议、国内 logo 预设、Provider 配置恢复和 Embedding/转写/声纹 API 路由；本地语音模型仍为阶段 F 接口预留。RAG Benchmark 检索评测（Dataset 加载、异步运行、SSE 进度、指标聚合与报告）已完成，Agent Benchmark 暂缓。后续继续接入真实音频处理、文档导出、主题包、Trace 可视化、Mermaid 和函数图像。Tauri/Rust Host、Stronghold、原生多 Vault 文件系统和 Sync Server 仍未实现。
+> 实施状态更新：2026-09-05。本文同时包含目标架构、当前实现和第二阶段接口基线。第一阶段及第二阶段 A～F 工程范围已经合并到 `main`：Vue Web 联调前端、FastAPI、真实单 Vault、Knowledge/Retrieval、知识库 Chat、Agent/Tool/Permission、Skill/Plugin、MCP 配置与调用、Provider 多协议与国内 logo 预设、RAG Benchmark、本地 Embedding、音频转写、片段级声纹聚类、CPU/CUDA 运行管理及用量诊断均已实现。当前生产 Embedding 使用固定 revision 的 Bekko A8M，Granite 97M Multilingual r2 可选；音频本地链路使用 Qwen3-ASR-0.6B 与 ERes2NetV2。Tauri/Rust Host、Stronghold、原生多 Vault 文件系统、生产级 MCP 沙箱和 Sync Server 仍未实现；逐字强制对齐、重叠语音分离及带标注长音频质量验收尚未完成。
 
 ---
 
@@ -47,17 +47,17 @@
 | 元数据 | SQLite | 笔记元数据、Block、标签、会话、Trace、任务、索引状态 |
 | 全文检索 | SQLite FTS5 | 关键词、标题、术语、标签等文本检索 |
 | 向量检索 | sqlite-vec + VectorStore | 本地语义检索 |
-| Embedding | 可插拔 EmbeddingProvider，默认本地 BGE-M3 类模型 | 为 Note Block 生成向量 |
+| Embedding | 可插拔 EmbeddingProvider；默认 Bekko A8M，可选 Granite 97M Multilingual r2 | 为 Note Block 生成 384 维向量，并按模型空间隔离索引 |
-| Reranker | BGE reranker 类 Cross-Encoder | 对候选检索结果进行精排 |
+| Reranker | 当前 `LexicalReranker`；保留 `RerankerProvider` 替换边界 | 对 RRF 候选进行词面重叠与原始分数加权精排 |
 | Agent | 自研 Agent Runtime | 模型推理、工具选择、工具调用、结果回灌、运行控制 |
 | Skill | 自研声明式 Skill Runtime | 复用提示词、工具集合、权限和检索配置 |
 | Plugin | 自研 Plugin Runtime + Plugin Manifest + MCP Bridge | 扩展程序能力、Tool、外部服务集成和受控 UI Contribution |
 | Theme | Theme Manifest + Design Token + 受限 CSS | 本地主题包导入、预览、启停与社区格式兼容 |
 | LLM | 自研 Provider Adapter | 统一不同模型服务商的输入、输出、Streaming 与 Tool Calling |
 | 模型协议 | OpenAI Responses / Chat Completions compatible / Anthropic Messages / Ollama | 用户自定义模型接入 |
-| ASR | faster-whisper | 音频转写 |
+| ASR | Qwen3-ASR-0.6B | 本地音频转写与语言识别，返回片段级时间边界 |
-| 说话人分离 | pyannote.audio | 课堂、会议等多人音频中的说话人区分 |
+| 声纹匹配与片段聚类 | ERes2NetV2 中文声纹模型 | 两段音频相似度与转写片段 speaker 聚类 |
-| 情感识别 | emotion2vec | 可选音频分析能力 |
+| 情感识别 | Provider 接口预留，尚未选择运行模型 | 后续可选音频分析能力 |
 | 文档导出 | Document AST + Exporter Adapter | Markdown 到 HTML、PDF、DOCX，并保留图表、公式和代码块 |
 | 密钥存储 | 当前 Fernet 开发存储；目标 Tauri Stronghold | Web 联调期避免明文落盘，桌面集成后保存模型 API Key 和同步凭证 |
 | 云同步 | 独立 Sync Server：FastAPI + PostgreSQL + S3/MinIO | 可选自托管，多设备 Vault 同步、版本管理和设备管理 |
@@ -675,13 +675,15 @@ class EmbeddingProvider(Protocol):
     async def embed_query(self, query: str) -> list[float]: ...
 ```
 
-目标默认配置使用本地 BGE-M3 类模型。当前第一阶段实现是 128 维 `HashEmbeddingProvider`，只用于离线跑通向量存储、索引更新和 Hybrid 链路，不代表真实语义召回质量。第二阶段接入真实 Embedding 时继续实现相同接口，上层 Retrieval Core 不依赖具体模型运行时。
+生产默认配置使用 `hotchpotch/bekko-embedding-v1-a8m`，可选 `ibm-granite/granite-embedding-97m-multilingual-r2`，两者均输出 384 维向量。模型权重使用代码目录中审阅过的固定 revision，下载后校验，推理阶段离线读取。`HashEmbeddingProvider` 仅供测试显式注入，不进入生产检索回退。
+
+Embedding 支持 Provider API 与本地模型路由：配置可用 API 时优先调用，响应失败或无效时回退本地模型；未配置 API 时直接使用本地模型；`local_only` 禁止远程调用。索引和查询冻结同一份模型与设备配置，并记录实际来源和回退原因。
 
 索引记录需要保存 embedding model id、模型版本、向量维度和归一化方式。用户更换模型或任一索引兼容字段变化后，索引服务必须将旧向量标记为不可用并要求重建，禁止把不同模型生成的向量写入同一索引空间。
 
 ### 9.5 RRF 与 Reranker
 
-FTS5 和 Vector Search 分别产生候选集合，经 RRF 进行排名融合。融合后的候选交给 BGE reranker 类 Cross-Encoder 进行精排。
+FTS5 和 Vector Search 分别产生候选集合，经 RRF 进行排名融合。当前 `LexicalReranker` 使用词面重叠和归一化原始分数做确定性精排；`RerankerProvider` 接口保留后续替换 Cross-Encoder 的边界，当前阶段没有随本地模型运行环境安装独立 Reranker 权重。
 
 初始参数可以采用：
 
@@ -1268,7 +1270,7 @@ enabled
 → 返回连接测试结果
 ```
 
-当前设置页已经提供 OpenAI、DeepSeek 与 Ollama 预设，保存后通过 `/api/providers/{provider_id}/models` 自动发现模型。Credential API 只返回配置状态，不提供任何明文读取接口。
+当前设置页提供 OpenAI、DeepSeek、通义千问、Kimi、智谱、豆包、腾讯混元、百度千帆、MiniMax、阶跃星辰、硅基流动和 Ollama 等带 logo 预设，也支持自定义兼容服务。保存后通过 `/api/providers/{provider_id}/models` 自动发现模型；聊天、Embedding、转写和声纹能力可独立绑定。Credential API 只返回配置状态，不提供任何明文读取接口。
 
 日志中不记录完整 API Key。请求异常信息在进入前端前过滤 Authorization Header 和密钥片段。
 
@@ -1282,16 +1284,19 @@ enabled
 
 ```mermaid
 flowchart LR
-    A["Audio"] --> D["pyannote.audio"]
+    A["Audio"] --> X["PyAV Decode · 16 kHz Mono"]
-    D --> S["Speaker Segments"]
+    X --> V["Energy Segmentation"]
-    S --> W["faster-whisper"]
+    V --> W["Qwen3-ASR-0.6B"]
-    W --> T["Timestamped Transcript"]
+    V --> S["ERes2NetV2 Embeddings"]
+    W --> T["Segment Transcript"]
+    S --> SC["Speaker Clustering"]
+    SC --> T
     T --> C["Content Structuring"]
     C --> M["Markdown Note"]
     M --> I["Index Pipeline"]
 ```
 
-pyannote.audio 生成说话人区间；faster-whisper 对各区间进行转写。最终 Transcript 至少包含：
+PyAV 将音轨解码为 16 kHz 单声道，能量分段后由 Qwen3-ASR-0.6B 转写；ERes2NetV2 为片段提取 192 维声纹并按相似度聚类。最终 Transcript 至少包含：
 
 ```text
 speaker
@@ -1317,7 +1322,9 @@ status
 error
 ```
 
-`pyannote.audio` 和 `faster-whisper` 通过独立 Adapter 加载，模型下载、设备选择、精度、批量大小和缓存目录由配置管理。缺少说话人模型时可以只返回时间戳转写，但必须明确标记 diarization 不可用；模型失败不能生成伪造的 completed 结果。
+本地模型由独立运行环境和子进程按需加载，模型下载、固定 revision、设备、超时、内存预算和缓存目录由配置管理。转写与声纹也可以绑定 Provider API；API 无配置或返回无效时回退本地，`local_only` 请求禁止远程调用。缺少声纹模型时只能返回没有 speaker 的片段并明确标记 diarization 不可用，模型失败不能生成伪造的 completed 结果。
+
+当前时间信息为能量分段产生的片段级边界，不是逐字强制对齐。ERes2NetV2 聚类不能处理同一片段内多人或重叠发言，因此不将当前能力描述为完整说话人分离。
 
 ### 14.2 OCR
 
@@ -1325,9 +1332,9 @@ OCR 作为 Media Pipeline 的输入适配能力，用于图片笔记、白板照
 
 OCR 引擎在当前技术栈中尚未固定，调用接口先定义为 `OCRProvider`，具体实现完成 PoC 后确定。
 
-### 14.3 emotion2vec
+### 14.3 音频情感分析
 
-emotion2vec 作为音频扩展分析模块。输出可以附加到音频段元数据，不参与核心 RAG 索引和 Agent 启动流程。
+音频情感分析保留 Provider/Adapter 扩展位置，当前阶段未选定或集成本地运行模型。未来输出可以附加到片段元数据，但不参与核心 RAG 索引和 Agent 启动流程。
 
 ### 14.4 Document AST 与多格式导出
 
@@ -1630,7 +1637,7 @@ Agent 中间运行状态
 设备级性能配置
 ```
 
-例如 Client A 使用本地 BGE-M3，Client B 使用另一种 Embedding Provider。服务器只同步 Markdown。Client B 收到文件后按照自己的 Embedding 配置生成向量，并写入本机 VectorStore。
+例如 Client A 使用本地 Bekko A8M，Client B 使用 Granite 或 API Embedding Provider。服务器只同步 Markdown。Client B 收到文件后按照自己的 Embedding 配置生成向量，并写入本机对应的隔离向量空间。
 
 ### 16.5 多设备同步流程
 
@@ -2010,7 +2017,7 @@ SQLite
 Git
 ```
 
-Rust Toolchain 与 Tauri CLI 只在桌面容器阶段安装。当前轻量 Embedding/Reranker 不要求 CUDA；接入 faster-whisper、pyannote.audio 或真实本地模型时再按所选运行时增加 CPU/GPU 依赖。
+Rust Toolchain 与 Tauri CLI 只在桌面容器阶段安装。本地模型依赖与 API 环境分离，默认使用 CPU；Windows 可以通过设置页或 `backend/scripts/install-model-runtime.ps1 -Device cuda` 显式安装 PyTorch 2.9.1 cu128 运行组件。CPU 与 CUDA 环境可并存，安装脚本不安装或修改 NVIDIA 驱动，也不要求 vLLM 或 FlashAttention。
 
 ### 19.2 本地开发
 
@@ -2233,10 +2240,11 @@ Agent Runtime
 
 ```text
 Audio
-→ pyannote.audio
+→ PyAV 解码为 16 kHz 单声道
-→ Speaker Segments
+→ 能量分段
-→ faster-whisper
+→ Qwen3-ASR-0.6B 片段转写
-→ Timestamped Transcript
+→ ERes2NetV2 声纹提取与片段聚类
+→ 片段级时间戳 Transcript
 → Content Structuring
 → Markdown
 → Note Core
@@ -2333,14 +2341,17 @@ Markdown Workspace
 
 第一阶段 Plugin Runtime 已完成安装、启用、停用、权限和声明式 Tool 注册，建立 Skill 调用 Plugin Tool 的基础链路。Command、Settings 和 MCP 执行不计入第一阶段完成项。
 
-截至 2026-09-03，上述第一阶段后端链路和 Web 联调前端均已完成；第二阶段的 Workspace 去 Mock 联调、Agent Trace 持久化/恢复接口、stdio MCP Bridge / Plugin Host，以及 Plugin Command/Settings 前后端闭环也已完成。Plugin 详情页现已提供 Host 状态、重启、动态设置、Secret 管理和命令执行，全局命令面板可加载 Plugin Command。当前验证基线为后端 136 项测试、前端 32 项测试、TypeScript 类型检查及生产构建通过。向量链路当前使用 `HashEmbeddingProvider` 验证工程正确性，真实 Embedding 召回质量不属于该测试结论。
+截至 2026-09-05，上述第一阶段链路和第二阶段 A～F 工程范围均已合并。除 Workspace、Agent Trace、MCP、Plugin Command/Settings 外，当前还包含 Provider 多协议与国内预设、真实 Embedding 路由与隔离向量空间、RAG Benchmark、Qwen3-ASR 转写、ERes2NetV2 声纹匹配与片段聚类、CPU/CUDA 组件管理、请求 JSON、Token/音频用量及运行诊断。阶段 F 合并验证基线为后端 559 项测试、前端 103 项测试、TypeScript 类型检查及生产构建通过。
+
+Bekko A8M 与 Granite 97M Multilingual r2 在 8 篇短文、6 条改写查询的小样本冒烟中均得到 Hit@1、Recall@5、MRR 1.0；该结果只证明中文检索闭环可运行，不足以区分质量优劣。CPU/CUDA 已完成短音频到知识库检索的真实闭环；带标注课程长音频的 WER/CER、DER、重叠语音和吞吐仍需专项验收。
 
 第二阶段在既有 Contract 上接入：
 
 ```text
 Multimodal
-├── faster-whisper
+├── Qwen3-ASR-0.6B
-└── pyannote.audio
+├── ERes2NetV2 片段声纹聚类
+└── API → 本地模型回退路由
 
 Extension / Model
 ├── MCP Bridge（stdio 首版已实现）
@@ -2365,7 +2376,7 @@ Frontend Extension
 └── Plugin Settings UI
 ```
 
-上述列表描述第二阶段技术范围，其中 stdio MCP Bridge、Plugin Command Contribution 和 Plugin Settings Contribution 后端 Contract 已实现，其余能力以各自开发说明的状态为准。每项功能必须继续经过现有 Service、Contract、Permission 和 Adapter 边界，不因 Demo 需要在 Vue 组件、Router 或 Agent Runtime 中直接绑定第三方协议。
+上述列表描述第二阶段技术范围。多模态、MCP Bridge、Plugin Command/Settings、RAG Benchmark 和 Provider 增强已经实现；Agent Benchmark、内容导出、Mermaid/函数图像完整编辑导出及社区主题包仍以各自开发说明的状态为准。每项功能必须继续经过现有 Service、Contract、Permission 和 Adapter 边界，不因 Demo 需要在 Vue 组件、Router 或 Agent Runtime 中直接绑定第三方协议。
 
 第三阶段处理：
 
@@ -2417,11 +2428,11 @@ Sync Server 按独立服务开发和部署，不进入桌面客户端核心启
 
 ## 25. 当前技术基线摘要
 
-目标桌面端采用 Tauri 2、Rust、Vue 3 和 TypeScript；当前可运行形态是 Vue/Vite Web 前端加 FastAPI。用户笔记以 Markdown 和 Assets 保存在本地 Vault，SQLite 已管理笔记元数据、全文索引、向量索引、任务及 Agent Trace；Provider/Extension Registry 当前仍为内存实现。
+目标桌面端采用 Tauri 2、Rust、Vue 3 和 TypeScript；当前可运行形态是 Vue/Vite Web 前端加 FastAPI。用户笔记以 Markdown 和 Assets 保存在本地 Vault，SQLite 已管理笔记元数据、全文索引、向量索引、搜索历史、Provider 配置、任务、多模态记录及 Agent Trace；MCP Server 配置持久化到后端数据目录，运行时 Tool/Extension Registry 在进程启动后按持久配置重建。
 
-Python AI Core 未来作为 Tauri Sidecar 运行，当前由开发命令独立启动，FastAPI 提供本地接口。Knowledge Core 管理笔记结构；Retrieval Core 当前通过 FTS5、`HashEmbeddingProvider`、sqlite-vec、RRF 和轻量 Reranker 跑通混合检索，真实 Embedding 与正式 Benchmark 仍待第二阶段后续接入；Agent Runtime 使用 Tool Registry 操作知识库和任务，并已持久化可供前端可视化与 Benchmark 共用的 Agent Trace Contract；Skill Runtime 将提示词、工具、权限和检索参数组装为可复用 Agent 配置。
+Python AI Core 未来作为 Tauri Sidecar 运行，当前由开发命令独立启动，FastAPI 提供本地接口。Knowledge Core 管理笔记结构；Retrieval Core 通过 FTS5、sqlite-vec、RRF 和 `LexicalReranker` 运行混合检索，生产 Embedding 默认使用 Bekko A8M、可选 Granite 97M Multilingual r2 或 Provider API，并按模型空间隔离索引；`HashEmbeddingProvider` 仅用于测试。RAG Benchmark 已接入版本化 Dataset、异步运行、SSE 进度、指标和报告。Agent Runtime 使用 Tool Registry 操作知识库和任务，并持久化可供前端和 Benchmark 共用的 Agent Trace Contract；Skill Runtime 将提示词、工具、权限和检索参数组装为可复用 Agent 配置。
 
-当前 Plugin Runtime 支持 Manifest、生命周期、声明式白名单 Tool Contribution、Plugin Command 与 Plugin Settings/Secret，并已通过 stdio MCP Bridge 接入独立进程 Tool、专用 MCP Command Target、Host 状态与重启接口。Provider Adapter 当前实现 Mock、OpenAI Chat/OpenAI-Compatible 与 Ollama，OpenAI Responses、Anthropic Messages 等协议仍待第二阶段后续完善。多模态目标方案使用 faster-whisper、pyannote.audio 和可选 emotion2vec；当前只读取 Host 预生成 transcript。
+当前 Plugin Runtime 支持 Manifest、生命周期、声明式白名单 Tool Contribution、Plugin Command 与 Plugin Settings/Secret，并已通过 MCP Bridge 接入 stdio、Streamable HTTP 和旧 SSE Server。Provider Adapter 已实现 OpenAI Chat/OpenAI-Compatible、OpenAI Responses、Anthropic Messages 与 Ollama，并支持模型发现、独立凭据和受限自定义请求 JSON。多模态本地链路使用 PyAV、Qwen3-ASR-0.6B 和 ERes2NetV2，默认 CPU、CUDA 显式选装；API 无配置或无效时回退本地模型。当前声纹聚类只处理片段，不等同于完整说话人分离。
 
 第二阶段内容输出以 Document AST、Exporter Adapter、Mermaid Renderer 和 Function Plot Renderer 为共同边界，支持 HTML、PDF、DOCX 与静态图导出。Theme Package 使用 Manifest、Design Token 和受限 CSS 实现本地导入；联网主题市场不属于本阶段核心依赖。API Key 在 Web 联调期由 Fernet 开发存储加密保存，桌面版迁移到 Tauri Stronghold。多设备同步的目标方案为独立、可自托管的 Sync Server，目前尚未实现；本地核心功能不依赖 Sync Server。
 

docs/contracts/第二阶段接口契约-开发版.md

@@ -1573,3 +1573,19 @@ frontend/src/
 ### Benchmark Embedding 运行归属（阶段 E 集成修复）
 
 `config_snapshot.local_embedding` 仅表示本地基线；`config_snapshot.embedding` 为 `{ "policy": "per_case", "details": "cases[].embedding" }`。报告与 CaseCompleted 事件的逐样本 `embedding` 包含实际 source（api/local/not_used/unavailable）、model_id、dimensions，以及可选 version、fallback_reason、requested_route、route_version、attempted_space。requested_route 仅含提供商引用、模型、相对端点和维度，不包含 API Key 或凭据引用。FTS 不使用 Embedding，标记 not_used；远程失败或索引不完整回退时记录实际本地模型及原因。
+
+### 阶段 F 收尾接口补充（2026-09-05）
+
+CUDA 组件：`GET /api/local-models/runtime-components/cuda` 返回 status、stage、supported、custom_interpreter、cuda_available、可选 torch/error。status 为 checking/not_installed/installing/installed/failed/interrupted；读取只检查现有环境，不下载安装。`POST` 同路径明确触发后台安装，返回 202；重复请求复用当前安装任务。正在推理/排队返回 409 MODEL_IN_USE，缺少 uv 返回 422 UV_NOT_INSTALLED，不支持的平台返回 422 PLATFORM_UNSUPPORTED。阶段进度不冒充字节百分比。关闭后端时回收安装进程树，重启后重新验证环境。
+
+| 接口/字段 | 行为 |
+| --- | --- |
+| `POST /api/media/attachments` | 可选 `Idempotency-Key` Header，16–100 位字母、数字、下划线或连字符。后端持久保存键、文件名、attachment_id 和内容摘要；同键同文件同内容返回同 attachment_id，文件名（含扩展名）或内容不一致返回 409 `IDEMPOTENCY_CONFLICT`。对应附件已清理时返回 409 `IDEMPOTENCY_EXPIRED`，客户端需开始新提交。上传仍受 25 MiB 限制。 |
+| `TranscriptNoteRequest.update_existing` | 默认 false；true 时将新修订安全写入相同导出选项对应的笔记。无基线返回 409 `NOTE_UPDATE_BASELINE_MISSING`；正文改变返回 409 `NOTE_CONTENT_CONFLICT`。同修订重复调用保持幂等。 |
+| 本地模型 `disk_bytes` | 权重目录实际字节数；无法读取为 null。与下载 bytes/total 分开。 |
+| `GET /api/local-models/diagnostics` | `scope=application_last_200_attempts`，应用 SQLite 中最近 200 条诊断，包含调用及回退事件。未实际开始推理时不伪造 actual_device。 |
+| `GET /api/usage` | 增加 `audio_request_count`、可空的 `audio_seconds`、`audio_covered_requests`，适用原有时间/提供商/模型/来源过滤。次数按 transcription/speaker_matching 实际 attempt；未报告时长不估算。 |
+| `POST /api/providers/request-rules/validate` | 输入/输出 `{version:1, request_overrides:[...]}`；最多 100 条，复用请求扩展校验，不保存提供商。 |
+| `POST /api/providers/request-probe` | 输入 `{provider:ProviderCreateRequest, stream:boolean}`；固定短消息真实聊天推理，45 秒超时。成功返回 success/stream/model/message；空响应 422、供应商错误 502、超时 504。只使用 credential_id，不接收明文密钥。 |
+
+请求预览新增 capability 选择（chat/embedding/transcription/speaker_matching），仍只返回隐藏正文的请求体。实际扩展字段是否被供应商接受，以推理响应为准。

docs/development/多模态管线与模型运行开发说明.md

@@ -93,6 +93,30 @@ POST /api/providers/request-preview 不联网，隐藏正文/文件且不包含
 
 ## 验证记录
 
+### 阶段 F 收尾行为（2026-09-05）
+
+CUDA 页面安装入口：**设置 → 模型提供商 → 本地模型 → CUDA 运行组件（可选）**。未安装时显示“下载并安装 CUDA 组件”；安装中展示真实阶段和不定进度条，失败可重试。后端仅运行项目内固定安装脚本，写入独立 `.venv-models-cuda`，检查依赖及 CUDA wheel 后才标记就绪。已有环境会先检查；成功后选择 CUDA 并保存运行设置即可使用，CPU 环境保留。显式 `APP_MODEL_PYTHON` 继续优先，页面提示覆盖关系。当前页面安装支持 Windows，需后端能找到 uv；不会自动安装显卡驱动。
+
+- 模型卡片读取权重目录的实际文件大小，包含未完成下载的文件；下载进度与磁盘占用分别显示。
+- 本地任务串行执行；等待队列中交互检索优先级为 0，媒体任务为 10，后台笔记索引为 20。同级 FIFO，不抢占已运行任务。
+- 默认 CPU。选择 CUDA 后，设备不可用直接使用 CPU；CUDA 初始化失败或显存不足时先释放原子进程，再用冻结的同一任务配置重试 CPU 一次。其他错误不触发设备重试；用户取消不会启动后续尝试。重试会清除上一尝试的部分转写片段。
+- 安装脚本固定 CPU/CUDA wheel 为 `2.9.1+cpu` / `2.9.1+cu128`，避免已有 CPU wheel 被误认为满足 CUDA 安装。可用 `-RuntimeDirectory` 指定独立环境，后端通过 `APP_MODEL_PYTHON` 选择；不自动更换显卡驱动。
+- 运行诊断写入应用 SQLite，保留最近 200 条，覆盖本地成功、失败、取消及能力 API 调用/回退事件。仅保留模型、设备、数值耗时、资源、状态码及请求标识，不保存输入、文件路径、密钥或异常全文。排队取消不记作实际模型用量；设备重试有独立 attempt，共享逻辑 request_id。
+- 前端同一次提交在响应丢失后复用上传和任务幂等键；收到附件 ID 后只重试创建任务。“重新处理为新任务”明确创建新标识。客户端待提交状态仅在当前页面内存中，已接收任务和结果由后端持久化。
+- 转写修订可选择“更新已导出笔记”。后端在 Vault 写锁内校验上次导出内容摘要，保留 note_id 和本地索引限制。用户编辑过正文时返回冲突，不覆盖；旧记录没有摘要时需先创建新笔记。重建索引保留导出基线与关联。
+- 用量卡片单列音频实际调用次数、已报告时长和覆盖次数；时长不换算为 Token。重试分别计数，历史未知数据保持“未提供”。
+- 请求 JSON 可导入、导出和恢复默认。文件格式为 `{ "version": 1, "request_overrides": [...] }`，只包含扩展规则；服务端复用受保护字段与凭据校验，导入成功仍需保存提供商才生效。
+- 请求预览不联网。聊天“发送测试推理请求”使用当前草稿、已保存的凭据引用和固定短消息，支持流式/非流式，不读取知识库、工具或附件，并计入真实用量。更改模型、连接、规则或 JSON 有效性后，旧结果和迟到响应失效；媒体规则继续通过真实媒体操作验收。
+
+独立 CUDA 环境示例（不改变默认 CPU 环境）：
+
+```powershell
+./backend/scripts/install-model-runtime.ps1 -Device cuda -RuntimeDirectory ./backend/.venv-models-cuda
+$env:APP_MODEL_PYTHON = (Resolve-Path ./backend/.venv-models-cuda/Scripts/python.exe).Path
+```
+
+设置环境变量后需从同一终端重启后端；CPU 默认仍可用。模型权重与运行环境不提交仓库。
+
 ### 2026-09-04 联调修复补充
 
 Windows 热重载不支持异步子进程时使用线程管道兼容路径。本地 Embedding 进入推理前冻结模型与设备配置，向量空间标识来自同一快照；普通 API 成功及失败回退策略保持不变。

docs/development/阶段F收尾验收记录.md

@@ -0,0 +1,55 @@
+# 阶段 F 收尾验收记录
+
+日期：2026-09-05。对应 `feat/multimodal-finalization`，基于 `6bdba2c`（阶段 F 主分支合并）。
+
+## 完成范围
+
+本轮补齐运行诊断持久化、真实磁盘占用、后台索引优先级、CUDA 设备失败时 CPU 单次重试、上传与任务重试幂等、跨修订安全更新笔记、音频用量分项，以及请求 JSON 导入/导出/重置和实际聊天推理验证。原有 API 优先、无配置/无效响应使用本地模型、local_only 禁止远程调用的流程继续保留。
+
+具体行为见[开发说明](多模态管线与模型运行开发说明.md)，接口见[开发版契约](../contracts/第二阶段接口契约-开发版.md)，故障与修复见[问题记录 F-13～F-15](../retrospectives/阶段F-Embedding与知识库问题与解决方案.md)。
+
+## 自动化与页面验证
+
+| 项目 | 结果 |
+| --- | --- |
+| 后端全量 `python -m pytest -q -p no:cacheprovider` | 559 通过；1 条已有 Starlette/httpx 弃用提示 |
+| 前端全量 `npm test -- --run` | 29 个文件、103 项通过 |
+| 类型与生产构建 `npm run build` | vue-tsc 与 Vite 构建通过，仍有既有大 bundle 提示 |
+| `git diff --check` | 通过 |
+| 真实页面 | 模型卡片读取实际大小；音频统计显示真实缺失；提供商表单展示请求编辑、恢复默认、导入/导出及推理验证入口 |
+| 请求 Adapter 验证 | 隔离 HTTP Transport 检查最终流式/非流式请求和扩展字段，不访问外部供应商 |
+| 失败恢复 | 初始化/OOM 故障注入、CPU 再失败、进程回收、队列顺序、重复提交、修订冲突和旧结果失效均覆盖 |
+
+## CPU / CUDA 真实模型闭环
+
+Windows、Python 3.12。保留原 `.venv-models` CPU 环境，独立安装 `.venv-models-cuda`；安装后检查 `torch=2.9.1+cu128`、`cuda_available=True`。显卡为 NVIDIA GeForce RTX 4060 Laptop GPU。
+
+CPU 与 CUDA 分别创建隔离 Vault、附件目录和 SQLite，只读取固定 revision 权重；运行短中文音频 → Qwen3-ASR → ERes2NetV2 片段聚类 → Markdown 笔记 → Bekko 语义检索 → 修订更新。两次均返回已完成，检索命中同一笔记，更新保留 note_id 和 `embedding_local_only: true`，结束后本地运行队列无活跃任务。CPU 实际设备为 `cpu`，CUDA 各次实际设备为 `cuda:0`。
+
+| CUDA 环节 | 权重 revision | 加载 / 推理耗时 |
+| --- | --- | --- |
+| Qwen3-ASR-0.6B | `5eb144179a02acc5e5ba31e748d22b0cf3e303b0` | 30.375 / 3.234 秒 |
+| ERes2NetV2 片段聚类 | `3317286545c587ae682dbc166831d9448780eebb` | 5.735 / 0.578 秒 |
+| Bekko 首次笔记索引 | `c721113d59a1d91b447450324f51c4b3332c924a` | 19.860 / 0.656 秒 |
+
+这些是单次功能冒烟观察值；运行期间有其他验证任务，不用于宣称吞吐或 CPU/GPU 性能倍率。短样本只产生 1 个片段和 1 个 speaker，不能验证多人重叠语音质量。CUDA OOM 恢复使用故障注入，并非实机显存耗尽测试。
+
+## 中文 Embedding 小样本对照
+
+固定 8 篇人工构造的短文，主题为线性代数、死锁、Python 函数、语义检索、光合作用、备份及两个无关干扰项（晚餐、篮球）。6 条改写查询，各有一个预期相关文档；对全部文档做余弦排序。
+
+| 模型 | revision | Hit@1 / Recall@5 / MRR |
+| --- | --- | --- |
+| Bekko A8M | `c721113d59a1d91b447450324f51c4b3332c924a` | 1.0 / 1.0 / 1.0 |
+| Granite 97M Multilingual r2 | `835ad14087e140460703cf0fae09f97d469d65c2` | 1.0 / 1.0 / 1.0 |
+
+两者在这 6 条查询上的目标排名均为 1。该结果仅证明中文检索冒烟可运行，样本量不足以区分模型优劣；继续保留 Bekko 默认、Granite 可选。
+
+## 未关闭的专项验收
+
+- 带参考转写和说话人标注的真实课程长录音尚未提供，不能报告 CER/WER、DER、阈值或长音频吞吐达标。
+- 现阶段时间戳为片段级；逐字强制对齐、同段多人/重叠语音仍未实现，不将片段聚类视为完整说话人分离。
+- 外部供应商特殊 JSON 的兼容性，需要在目标账号和模型上点击实际推理验证；离线协议通过不替代厂商验收。
+- Tauri/Rust Host 和生产 MCP 沙箱按后续阶段安排；本轮数据持久化在后端 SQLite/Vault，为桌面集成保留稳定接口。
+
+结论：阶段 F 本轮工程收尾已实现并完成 CPU/CUDA 功能验收；上述质量及外部服务专项保持待验收状态，不标记为全部通过。分支仍需独立审阅后决定合并。

docs/guides/代码注释与TODO约定.md

@@ -37,7 +37,7 @@
 | Extension | 扩展安装状态尚未持久化；MCP Host、进程隔离、签名与来源校验属于第二阶段 |
 | AI Core | 音频转写当前只读取文本或 Host 预生成旁路文本，后续接入本地 ASR 队列 |
 | Desktop | Web Workspace 已连接 FastAPI 单 Vault；后续由 Tauri IPC 增加原生目录选择、多 Vault 和文件监听 |
-| Editor / Chat | 待补文件冲突合并、受控链接对话框及会话持久化 |
+| Editor / Chat | 待补文件冲突合并及受控链接对话框；会话持久化已接入后端 SQLite |
 | Performance | Shiki 已复用单例，后续按首屏指标评估延迟加载或 Web Worker |
 
 TODO 完成后应删除对应代码注释并同步更新本索引；若工作超过一个提交，应建立 Issue，并在 Issue 中引用代码位置，而不是在源码中记录长篇设计讨论。

docs/retrospectives/阶段F-Embedding与知识库问题与解决方案.md

@@ -138,6 +138,54 @@ embedding_local_only: true
 
 ## 9. 工程经验
 
+### F-18：设备快照与迟到导入错误未完全隔离
+
+问题：推理任务虽然冻结了 RuntimeConfig，但启动子进程时又从数据库读取最新 device 来选择 Python 环境；排队期间修改设置会改变已提交任务的运行环境，CUDA → CPU 重试也可能继续使用 CUDA 环境。请求规则的迟到成功响应已失效，但迟到失败仍会把旧错误显示到新草稿。
+
+实际方案：`interpreter` 接收本次 attempt 的冻结配置，`_execute` 在检查前解析一次可执行路径并复用；CUDA attempt 使用已验证的独立 CUDA 环境，CPU attempt 使用默认 CPU 环境，显式 APP_MODEL_PYTHON 仍保持最高优先级。导入异常与成功响应使用同一 generation 条件，只允许当前操作更新界面。
+
+验证：增加保存设置变化后仍按显式 attempt 选择环境、CPU 重试环境，以及旧导入失败晚于新编辑的回归。最终后端 559 项、前端 103 项和生产构建通过。
+
+### F-16：CUDA 选装只有脚本，前端缺少安装入口
+
+问题：上一轮完成了独立 CUDA 环境安装和 GPU 实测，但页面只有设备下拉框及脚本说明。用户无法从前端下载组件，工程收尾遗漏了可操作入口。
+
+实际方案：增加独立组件卡片和 GET/POST 状态、安装接口；展示环境检查、下载 PyTorch、安装依赖、验证等真实阶段，失败允许重试。固定脚本、目录和参数，默认 CPU 环境不变；安装成功后 CUDA 模式自动选择已验证环境，显式 Python 覆盖仍优先。推理期间拒绝安装，重复点击不产生多个任务，后端关闭时回收安装子进程树。
+
+验证：21 项后端相关测试及新增前端组件测试通过，类型检查和构建通过；真实页面已显示本机 `2.9.1+cu128` 组件就绪，默认 CPU 未改变。本轮复用已安装组件验证识别，未重复下载 3 GB 安装包；下载入口、重复请求和失败重试由隔离测试覆盖。
+
+### F-17：幂等键跨扩展名与请求规则导入竞态
+
+问题：附件 ID 原先由幂等键哈希和扩展名共同生成，相同键更换扩展名可以创建第二份附件。请求规则导入等待服务端验证期间，用户的新编辑可能被迟到的导入响应覆盖。
+
+实际方案：后端持久映射幂等键、文件名、附件 ID 和内容摘要，并在 SQLite 写锁中完成查重；文件名或内容变化均返回冲突，已清理的旧附件要求开始新提交。请求规则编辑器为导入和每次草稿变化递增 generation，只接受仍对应当前草稿的响应。
+
+验证：增加同键跨扩展名冲突，以及导入后继续编辑、迟到响应不覆盖的测试。相关后端 17 项、前端 15 项通过；最终全量后端 559 项、前端 102 项及生产构建通过。
+
+### F-13：CUDA 失败重试与诊断无法追溯
+
+问题：设备不可用时能够使用 CPU，但 CUDA 初始化失败、显存不足会直接使任务失败；诊断只留在进程内存中，重启后无法解释当时的失败和回退。
+
+实际方案：在同一队列占位内完成 CUDA → CPU 单次重试，先回收失败子进程再启动 CPU。只接受初始化失败、CUDA OOM 两类重试原因，普通模型错误不扩大重试范围。转写部分结果随尝试重置，取消仍终止流程。诊断按白名单写入 SQLite，保留最近 200 条，记录请求设备、实际设备、尝试设备、状态和耗时；未知设备不冒充实际使用设备。
+
+验证：故障注入覆盖初始化失败、OOM、普通错误、CPU 再失败、资源释放顺序、队列顺序和请求用量归属。Windows RTX 4060 Laptop 实机安装 `torch 2.9.1+cu128`，ASR、片段声纹和 Embedding 的实际设备均为 `cuda:0`，完成笔记生成、检索与修订更新。实机正常 CUDA 路径通过；OOM 回退是确定性注入验证，未人为耗尽用户显存。
+
+### F-14：响应丢失重复上传与转写笔记无法安全更新
+
+问题：前端每次点击都生成新幂等键，上传或创建任务已成功但响应丢失时，重试可能制造重复附件/任务。已有导出幂等只能返回相同修订，缺少新修订更新原笔记的保护机制。
+
+实际方案：同一次页面提交冻结文件与选项，复用上传/任务键，已获得的附件 ID 继续使用；主动重新处理才重置标识。后端重复上传校验内容摘要。导出基线保存正文摘要，跨修订更新在 Vault 写锁内核对基线，用户编辑冲突返回 409，允许改为创建新笔记；旧无基线记录不强行覆盖。索引重建不丢失基线，本地限制继续随笔记持久化。
+
+验证：覆盖上传响应丢失、任务响应丢失、主动重跑、重复键内容冲突、修订更新保持 note_id、重复导出、重建恢复和用户正文冲突。CPU/CUDA 两次真实本地管线均在隔离 Vault/SQLite 中通过检索与修订闭环，不写入用户笔记库。
+
+### F-15：运行管理与请求配置验收缺项
+
+问题：下载计数不能反映实际占用，后台索引与交互查询同优先级；音频调用没有独立时长统计；请求规则缺少导入/导出/恢复默认和真实推理验证，草稿改变后旧验证结果可能误导用户。
+
+实际方案：磁盘大小读取目录文件，查询/媒体/后台索引分别排队；音频次数、已报告时长与 Token 分开聚合，保留覆盖数。规则文件由服务端验证后替换草稿，保存后生效；验证按钮使用固定短消息走实际 Adapter。草稿变化使预览与验证失效，包括无效 JSON 和迟到响应。
+
+验证：增加实际目录统计、音频缺失值与去重、规则拒绝受保护字段、隔离 HTTP 协议测试及前端迟到响应测试。最终后端 555 项、前端 100 项通过。真实供应商兼容性仍须使用目标账号验证；本轮不将 MockTransport 协议测试称为厂商实测。
+
 ### F-12：普通分割线与元数据头部消歧
 
 F-11 修复后，`---` 和 `---\n\n# Title\n\n正文` 等合法 Markdown 被误判为未闭合 frontmatter，原先能够保存的笔记被拒绝；库中已有此类文件时全量重建也会失败。

frontend/README.md

@@ -0,0 +1,80 @@
+# NotesAgent Frontend
+
+NotesAgent Frontend 是基于 Vue 3、TypeScript、Vite、Pinia、Vue Router、Milkdown 和 CodeMirror 6 的 Web 联调前端。当前页面调用 FastAPI 真实接口，不使用业务 Mock 作为运行时回退；测试文件中的 mock 只用于隔离单元和组件测试。
+
+## 初始化与运行
+
+```powershell
+pnpm install
+pnpm dev
+```
+
+开发地址为 <http://127.0.0.1:5173>。Vite 将 `/api` 和 `/health` 转发到 <http://127.0.0.1:8000>，因此联调前需要先启动后端。
+
+## 页面与能力
+
+| 路由 | 当前能力 |
+| --- | --- |
+| `/`、`/workspace` | 选择当前 Vault、浏览目录、编辑和保存 Markdown |
+| `/search` | 全文、向量和混合检索；从后端读取并清空搜索历史 |
+| `/chat` | 流式 AI 对话、知识库上下文与 Citation |
+| `/agent/runs/:runId?` | 创建 Agent 运行，查看可恢复 Trace 与 Tool/Permission 事件 |
+| `/media` | 上传音频、创建/取消/重试转写、修订结果并生成知识库笔记 |
+| `/tasks` | 管理用户、笔记和 Agent 产生的任务 |
+| `/extensions/skills` | Skill 安装、启停与配置 |
+| `/extensions/mcp` | stdio、Streamable HTTP、旧 SSE Server 配置与工具发现 |
+| `/extensions/plugins` | Plugin Host、Command、Settings、Secret 与 MCP 状态 |
+| `/themes` | 内置 Design Token 主题和编辑器显示偏好 |
+| `/settings` | Provider、模型路由、本地模型、CPU/CUDA 组件、请求 JSON、用量与诊断；全局中英文和拼写检查设置 |
+
+## 技术结构
+
+| 目录 | 职责 |
+| --- | --- |
+| `src/features` | 按页面和业务域组织的 Vue 组件 |
+| `src/stores` | Pinia 状态与页面编排 |
+| `src/services` | FastAPI HTTP/SSE 客户端和 DTO 转换 |
+| `src/contracts` | 与后端契约对应的 TypeScript 类型 |
+| `src/components` | 应用壳、命令面板和共享组件 |
+| `src/utils` | Markdown 清洗、Shiki 高亮等纯工具 |
+| `src/styles` | Design Token、布局、主题和动效 |
+
+编辑器使用 Milkdown/Crepe 与 CodeMirror 6；Markdown 展示使用 marked、DOMPurify 和 Shiki。Provider logo 位于 `src/assets/providers`，授权与来源说明随目录保存。
+
+语言设置会即时更新主导航、页面标题和各功能页面，并同步更新文档与编辑器的 `lang`。拼写检查使用浏览器或桌面 WebView 提供的本地词典，开关会即时作用于可视化 Markdown、源码编辑器以及普通文本输入；JSON、密码等结构化或敏感输入保持关闭。
+
+## 数据边界
+
+- 笔记、附件、搜索历史、任务、Trace、模型配置和多模态结果都通过 FastAPI 读写。
+- AI 对话生成和知识库检索通过 FastAPI；会话列表、用户消息、流式助手结果、引用和 Token 用量保存在后端 SQLite，刷新页面后可恢复。
+- API Key 只存在于密码输入和提交请求期间，不进入 Pinia 或 `localStorage`。
+- 页面内存可以保存尚未提交的临时状态；后端已经接收的任务和结果由 SQLite/Vault 持久化。
+- 主题、编辑器偏好、侧栏状态和最近 Vault 路径目前保存在浏览器 `localStorage`；它们是设备界面偏好，不作为笔记或模型业务数据。Tauri 集成时由桌面配置存储接管。
+- 前端不直接访问 SQLite，不拼装第三方模型协议；Provider Adapter 和请求覆盖规则由后端执行。
+- 后端不可用时页面显示连接或操作错误，不生成演示数据替代真实结果。
+
+当前仍运行在 Web/Vite 环境。后续 Tauri 集成将复用现有 Service/Contract 边界，并由 Rust Host 接管窗口、Vault 选择、Sidecar、Stronghold 和生产沙箱。
+
+## 模型设置
+
+设置页支持带 logo 的提供商预设、模型发现、聊天/Embedding/转写/声纹能力绑定，以及按 capability、model 和 stream 条件匹配的自定义请求 JSON。请求预览不联网；“发送测试推理请求”使用当前草稿和已保存凭据执行真实短请求。
+
+本地模型页显示固定 revision、许可、下载状态和实际磁盘占用。CPU 是默认运行方式；Windows 可从页面安装独立 CUDA 12.8 组件，安装过程不修改显卡驱动。当前模型选型详见[多模态管线与模型运行](../docs/development/多模态管线与模型运行开发说明.md)。
+
+## 测试与构建
+
+```powershell
+pnpm test
+pnpm type-check
+pnpm build
+```
+
+当前基线为 30 个测试文件、106 项测试通过，TypeScript 类型检查与 Vite 生产构建通过；构建仍有既有大 bundle 提示。产物位于 `dist`，不提交 Git。
+
+## 开发约定
+
+- 依赖统一使用 pnpm 管理，不混用 npm 或 yarn。
+- 新接口先更新 `src/contracts` 与 `src/services`，页面和 Store 不直接散落 `fetch` 协议细节。
+- 异步页面需要处理加载、空数据、后端错误、重复提交和迟到响应。
+- 功能行为或契约变化时，同一提交同步更新测试和相关文档。
+- 页面需求见[前端页面需求说明](../docs/contracts/前端页面需求说明-开发版.md)，后端行为以运行时 `/openapi.json` 为准。

frontend/src/api.ts

@@ -1,3 +1,5 @@
+import { t } from '@/i18n'
+
 export interface ServiceStatus {
   name: string
   version: string
@@ -10,7 +12,7 @@ const apiBaseUrl = import.meta.env.VITE_API_BASE_URL ?? ''
 export async function getServiceStatus(): Promise<ServiceStatus> {
   const response = await fetch(`${apiBaseUrl}/api/status`)
   if (!response.ok) {
-    throw new Error(`后端请求失败：HTTP ${response.status}`)
+    throw new Error(`${t('后端请求失败：', 'Backend request failed: ')}HTTP ${response.status}`)
   }
   return response.json() as Promise<ServiceStatus>
 }

frontend/src/components/common/CommandPalette.vue

@@ -8,6 +8,7 @@ import * as workspaceService from '@/services/workspaceService'
 import * as pluginService from '@/services/pluginService'
 import type { PluginCommand, PluginCommandEffect } from '@/contracts'
 import { usePluginStore } from '@/stores/plugin'
+import { t } from '@/i18n'
 
 const router = useRouter()
 const editorStore = useEditorStore()
@@ -25,15 +26,15 @@ const selectionSnapshot = ref<string | null>(null)
 interface Command { id: string; label: string; hint: string; run: () => void | Promise<void> }
 
 const builtinCommands = computed<Command[]>(() => [
-  { id: 'workspace', label: '打开工作区', hint: '导航', run: () => router.push('/workspace') },
+  { id: 'workspace', label: t('打开工作区', 'Open workspace'), hint: t('导航', 'Navigation'), run: () => router.push('/workspace') },
-  { id: 'search', label: '全局搜索', hint: '导航', run: () => router.push('/search') },
+  { id: 'search', label: t('全局搜索', 'Global search'), hint: t('导航', 'Navigation'), run: () => router.push('/search') },
-  { id: 'chat', label: '打开 AI 对话', hint: '导航', run: () => router.push('/chat') },
+  { id: 'chat', label: t('打开 AI 对话', 'Open AI chat'), hint: t('导航', 'Navigation'), run: () => router.push('/chat') },
-  { id: 'agent', label: '创建智能体运行', hint: '导航', run: () => router.push('/agent/runs') },
+  { id: 'agent', label: t('创建智能体运行', 'Create agent run'), hint: t('导航', 'Navigation'), run: () => router.push('/agent/runs') },
-  { id: 'settings', label: '打开设置', hint: '导航', run: () => router.push('/settings') },
+  { id: 'settings', label: t('打开设置', 'Open settings'), hint: t('导航', 'Navigation'), run: () => router.push('/settings') },
-  { id: 'mode', label: `切换为${editorStore.mode === 'source' ? '写作' : '源码'}模式`, hint: '编辑器', run: () => editorStore.toggleMode() },
+  { id: 'mode', label: editorStore.mode === 'source' ? t('切换为写作模式', 'Switch to writing mode') : t('切换为源码模式', 'Switch to source mode'), hint: t('编辑器', 'Editor'), run: () => editorStore.toggleMode() },
-  { id: 'save', label: '保存当前笔记', hint: '编辑器', run: () => editorStore.save() },
+  { id: 'save', label: t('保存当前笔记', 'Save current note'), hint: t('编辑器', 'Editor'), run: () => editorStore.save() },
-  { id: 'theme', label: `切换为${themeStore.isDark ? '浅色' : '深色'}主题`, hint: '外观', run: () => themeStore.toggleTheme() },
+  { id: 'theme', label: themeStore.isDark ? t('切换为浅色主题', 'Switch to light theme') : t('切换为深色主题', 'Switch to dark theme'), hint: t('外观', 'Appearance'), run: () => themeStore.toggleTheme() },
-  { id: 'new-note', label: '创建笔记', hint: '工作区', run: createNote },
+  { id: 'new-note', label: t('创建笔记', 'Create note'), hint: t('工作区', 'Workspace'), run: createNote },
 ])
 
 const commands = computed<Command[]>(() => [
@@ -78,12 +79,12 @@ async function execute(command: Command | undefined) {
   try {
     await command.run()
   } catch (error) {
-    commandNotice.value = error instanceof Error ? error.message : '命令执行失败'
+    commandNotice.value = error instanceof Error ? error.message : t('命令执行失败', 'Command failed')
   }
 }
 
 async function createNote() {
-  const rawName = window.prompt('笔记名称')?.trim()
+  const rawName = window.prompt(t('笔记名称', 'Note name'))?.trim()
   if (!rawName) return
   const name = rawName.endsWith('.md') ? rawName : `${rawName}.md`
   const file = await workspaceService.createFile('/', name, `# ${rawName}\n\n`)
@@ -97,7 +98,7 @@ async function loadPluginCommands() {
   try {
     pluginCommands.value = await pluginService.listPluginCommands('command_palette')
   } catch (error) {
-    commandError.value = error instanceof Error ? error.message : 'Plugin 命令加载失败'
+    commandError.value = error instanceof Error ? error.message : t('Plugin 命令加载失败', 'Failed to load plugin commands')
   }
 }
 
@@ -109,7 +110,7 @@ async function executePluginCommand(command: PluginCommand) {
   if (hasRequiredArguments(command)) {
     pluginStore.selectPlugin(command.plugin_id)
     await router.push('/extensions/plugins')
-    commandNotice.value = '请在 Plugin 详情页填写参数后执行“' + command.title + '”。'
+    commandNotice.value = `${t('请在 Plugin 详情页填写参数后执行', 'Enter parameters on the Plugin details page, then run')} “${command.title}”.`
     return
   }
   const result = await pluginService.executePluginCommand(command.command_id, {}, {
@@ -135,11 +136,11 @@ async function applyPluginEffect(effect: PluginCommandEffect) {
   if (effect.type === 'refresh') {
     if (effect.payload.scope === 'plugins') await pluginStore.loadPlugins()
     if (effect.payload.scope === 'commands') await loadPluginCommands()
-    commandNotice.value = '相关数据已刷新。'
+    commandNotice.value = t('相关数据已刷新。', 'Related data refreshed.')
     return
   }
-  if (effect.type === 'job') { commandNotice.value = '后台任务已创建：' + effect.payload.job_id; return }
+  if (effect.type === 'job') { commandNotice.value = t('后台任务已创建：', 'Background job created: ') + effect.payload.job_id; return }
-  commandNotice.value = 'Plugin 命令执行完成。'
+  commandNotice.value = t('Plugin 命令执行完成。', 'Plugin command completed.')
 }
 
 function handleKeydown(event: KeyboardEvent) {
@@ -157,20 +158,20 @@ onBeforeUnmount(() => window.removeEventListener('keydown', handleKeydown))
 
 <template>
   <div v-if="commandNotice" class="command-toast" role="status">
-    <span>{{ commandNotice }}</span><button aria-label="关闭通知" @click="commandNotice = ''">×</button>
+    <span>{{ commandNotice }}</span><button :aria-label="t('关闭通知', 'Close notification')" @click="commandNotice = ''">×</button>
   </div>
   <Teleport to="body">
     <div v-if="open" class="command-backdrop" @click.self="hide">
-      <section class="command-palette" role="dialog" aria-modal="true" aria-label="命令面板">
+      <section class="command-palette" role="dialog" aria-modal="true" :aria-label="t('命令面板', 'Command palette')">
-        <input ref="input" v-model="query" class="command-input" placeholder="输入命令…" @keydown.enter.prevent="execute(filteredCommands[0])" />
+        <input ref="input" v-model="query" class="command-input" :placeholder="t('输入命令…', 'Enter a command…')" @keydown.enter.prevent="execute(filteredCommands[0])" />
         <p v-if="commandError" class="command-error">{{ commandError }}</p>
         <div class="command-list">
           <button v-for="command in filteredCommands" :key="command.id" type="button" @click="execute(command)">
             <span>{{ command.label }}</span><small>{{ command.hint }}</small>
           </button>
-          <p v-if="!filteredCommands.length">没有匹配的命令</p>
+          <p v-if="!filteredCommands.length">{{ t('没有匹配的命令', 'No matching commands') }}</p>
         </div>
-        <footer><span>Enter 执行</span><span>Esc 关闭</span></footer>
+        <footer><span>Enter · {{ t('执行', 'Run') }}</span><span>Esc · {{ t('关闭', 'Close') }}</span></footer>
       </section>
     </div>
   </Teleport>

frontend/src/components/common/FilePicker.spec.ts

@@ -0,0 +1,36 @@
+// @vitest-environment happy-dom
+import { mount } from '@vue/test-utils'
+import { describe, expect, it } from 'vitest'
+import FilePicker from './FilePicker.vue'
+
+describe('FilePicker', () => {
+  it('keeps the native file input accessible and reports the selected file', async () => {
+    const wrapper = mount(FilePicker, {
+      props: { file: null, label: '选择文件', emptyLabel: '尚未选择文件', accept: '.json' },
+    })
+    const input = wrapper.get('input[type="file"]')
+    const file = new File(['{}'], 'rules.json', { type: 'application/json' })
+    Object.defineProperty(input.element, 'files', { value: [file], configurable: true })
+
+    await input.trigger('change')
+
+    expect(wrapper.emitted('select')).toEqual([[file]])
+    expect(wrapper.get('label').attributes('for')).toBe(input.attributes('id'))
+    expect(wrapper.text()).toContain('尚未选择文件')
+
+    await wrapper.setProps({ file })
+    expect(wrapper.text()).toContain('rules.json')
+  })
+
+  it('emits null when the native selection is cleared', async () => {
+    const wrapper = mount(FilePicker, {
+      props: { file: null, label: '选择文件', emptyLabel: '尚未选择文件' },
+    })
+    const input = wrapper.get('input[type="file"]')
+    Object.defineProperty(input.element, 'files', { value: [], configurable: true })
+
+    await input.trigger('change')
+
+    expect(wrapper.emitted('select')).toEqual([[null]])
+  })
+})

frontend/src/components/common/FilePicker.vue

@@ -0,0 +1,58 @@
+<script setup lang="ts">
+import { useId } from 'vue'
+import { Upload } from '@element-plus/icons-vue'
+
+defineProps<{
+  file: File | null
+  label: string
+  emptyLabel: string
+  accept?: string
+  disabled?: boolean
+}>()
+
+const emit = defineEmits<{ select: [file: File | null] }>()
+const inputId = useId()
+
+function selectFile(event: Event) {
+  emit('select', (event.target as HTMLInputElement).files?.[0] ?? null)
+}
+
+function allowReselect(event: MouseEvent) {
+  ;(event.currentTarget as HTMLInputElement).value = ''
+}
+</script>
+
+<template>
+  <div class="file-picker" :class="{ disabled }">
+    <input
+      :id="inputId"
+      class="file-picker-input"
+      type="file"
+      :accept="accept"
+      :disabled="disabled"
+      @click="allowReselect"
+      @change="selectFile"
+    />
+    <label class="file-picker-trigger" :for="inputId">
+      <Upload aria-hidden="true" />
+      <span>{{ label }}</span>
+    </label>
+    <span class="file-picker-name" :class="{ empty: !file }" :title="file?.name || emptyLabel">
+      {{ file?.name || emptyLabel }}
+    </span>
+  </div>
+</template>
+
+<style scoped>
+.file-picker { display: flex; min-width: 0; align-items: center; gap: var(--space-sm); }
+.file-picker-input { position: absolute; width: 1px; height: 1px; overflow: hidden; clip: rect(0 0 0 0); clip-path: inset(50%); white-space: nowrap; }
+.file-picker-trigger { display: inline-flex; min-height: 36px; flex: 0 0 auto; align-items: center; gap: var(--space-sm); padding: 0 var(--space-md); border: 1px solid var(--color-border-default); border-radius: var(--radius-md); background: var(--color-surface-primary); font-weight: 600; cursor: pointer; transition: border-color var(--motion-fast), background-color var(--motion-fast), color var(--motion-fast), box-shadow var(--motion-fast), transform var(--motion-fast); }
+.file-picker-trigger svg { width: 16px; height: 16px; }
+.file-picker-trigger:hover { border-color: var(--color-accent-secondary); background: var(--color-background-hover); color: var(--color-accent-primary); transform: translateY(-1px); }
+.file-picker-input:focus-visible + .file-picker-trigger { outline: 2px solid var(--color-border-focus); outline-offset: 2px; }
+.file-picker-name { min-width: 0; overflow: hidden; color: var(--color-text-secondary); text-overflow: ellipsis; white-space: nowrap; user-select: text; }
+.file-picker-name.empty { color: var(--color-text-tertiary); }
+.disabled { opacity: .55; }
+.disabled .file-picker-trigger { cursor: not-allowed; transform: none; }
+@media (max-width: 560px) { .file-picker { align-items: stretch; flex-direction: column; } .file-picker-trigger { justify-content: center; } }
+</style>

frontend/src/components/common/PrimarySidebar.vue

@@ -3,24 +3,25 @@ import { useRoute, useRouter } from 'vue-router'
 import { computed, ref } from 'vue'
 import { ArrowLeftBold, ArrowRightBold, Brush, ChatDotRound, CircleCheck, Connection, Cpu, FolderOpened, Lightning, Monitor, Search, Setting } from '@element-plus/icons-vue'
 import AppIcon from './AppIcon.vue'
+import { t } from '@/i18n'
 
 const route = useRoute()
 const router = useRouter()
 const expanded = ref(localStorage.getItem('primary-sidebar-expanded') === 'true')
 
-const navItems = [
+const navItems = computed(() => [
-  { name: 'workspace', icon: FolderOpened, label: '工作区' },
+  { name: 'workspace', icon: FolderOpened, label: t('工作区', 'Workspace') },
-  { name: 'search', icon: Search, label: '搜索' },
+  { name: 'search', icon: Search, label: t('搜索', 'Search') },
-  { name: 'chat', icon: ChatDotRound, label: 'AI 对话' },
+  { name: 'chat', icon: ChatDotRound, label: t('AI 对话', 'AI Chat') },
-  { name: 'agent', icon: Cpu, label: '智能体' },
+  { name: 'agent', icon: Cpu, label: t('智能体', 'Agent') },
-  { name: 'tasks', icon: CircleCheck, label: '任务' },
+  { name: 'tasks', icon: CircleCheck, label: t('任务', 'Tasks') },
-  { name: 'media', icon: Monitor, label: '音视频' },
+  { name: 'media', icon: Monitor, label: t('音视频', 'Media') },
   { name: 'skills', icon: Lightning, label: 'Skill' },
   { name: 'plugins', icon: Connection, label: 'Plugin' },
   { name: 'mcp-servers', icon: Monitor, label: 'MCP' },
-  { name: 'themes', icon: Brush, label: '主题' },
+  { name: 'themes', icon: Brush, label: t('主题', 'Themes') },
-  { name: 'settings', icon: Setting, label: '设置' },
+  { name: 'settings', icon: Setting, label: t('设置', 'Settings') },
-]
+])
 
 const currentName = computed(() => {
   return route.name as string
@@ -52,9 +53,9 @@ function toggleExpanded() {
       </div>
     </nav>
     <div class="sidebar-footer">
-      <button class="nav-item collapse-button" type="button" :title="expanded ? '收起导航' : '展开导航'" @click="toggleExpanded">
+      <button class="nav-item collapse-button" type="button" :title="expanded ? t('收起导航', 'Collapse navigation') : t('展开导航', 'Expand navigation')" @click="toggleExpanded">
         <AppIcon class="nav-icon" :icon="expanded ? ArrowLeftBold : ArrowRightBold" />
-        <span class="nav-label">{{ expanded ? '收起' : '展开' }}</span>
+        <span class="nav-label">{{ expanded ? t('收起', 'Collapse') : t('展开', 'Expand') }}</span>
       </button>
     </div>
   </aside>

frontend/src/components/common/SecondarySidebar.vue

@@ -7,6 +7,7 @@ import SearchFiltersPanel from '@/features/search/SearchFiltersPanel.vue'
 import TaskFiltersPanel from '@/features/tasks/TaskFiltersPanel.vue'
 import ExtensionListPanel from '@/components/common/ExtensionListPanel.vue'
 import { useRoute } from 'vue-router'
+import { t } from '@/i18n'
 
 const props = defineProps<{
   component: string | null
@@ -17,12 +18,12 @@ const routeName = computed(() => route.name as string)
 
 const sidebarTitle = computed(() => {
   const titles: Record<string, string> = {
-    'file-tree': '文件',
+    'file-tree': t('文件', 'Files'),
-    'conversation-list': '对话',
+    'conversation-list': t('对话', 'Conversations'),
-    'run-list': '智能体运行',
+    'run-list': t('智能体运行', 'Agent Runs'),
-    'search-filters': '搜索筛选',
+    'search-filters': t('搜索筛选', 'Search Filters'),
-    'task-filters': '任务筛选',
+    'task-filters': t('任务筛选', 'Task Filters'),
-    'extension-list': '扩展',
+    'extension-list': t('扩展', 'Extensions'),
   }
   return titles[props.component || ''] || ''
 })

frontend/src/components/common/StatusBar.vue

@@ -5,6 +5,7 @@ import { useSettingsStore } from '@/stores/settings'
 import { useProviderStore } from '@/stores/provider'
 import { useAgentStore } from '@/stores/agent'
 import { useRoute } from 'vue-router'
+import { t } from '@/i18n'
 
 const editorStore = useEditorStore()
 const settingsStore = useSettingsStore()
@@ -15,12 +16,12 @@ const route = useRoute()
 const saveStatusText = computed(() => {
   const map: Record<string, string> = {
     idle: '',
-    dirty: '未保存',
+    dirty: t('未保存', 'Unsaved'),
-    saving: '保存中...',
+    saving: t('保存中...', 'Saving...'),
-    saved: '已保存',
+    saved: t('已保存', 'Saved'),
-    save_failed: '保存失败',
+    save_failed: t('保存失败', 'Save failed'),
-    external_changed: '外部已更新',
+    external_changed: t('外部已更新', 'Changed externally'),
-    conflict: '存在冲突',
+    conflict: t('存在冲突', 'Conflict'),
   }
   return map[editorStore.saveStatus] || ''
 })
@@ -39,16 +40,16 @@ const saveStatusColor = computed(() => {
 
 const indexStatusText = computed(() => {
   const s = settingsStore.indexStatus.status
-  return s === 'unknown' ? '索引状态未获取' : s === 'idle' ? '索引就绪' : s === 'indexing' ? `索引中 (${settingsStore.indexStatus.pending_jobs})` : '索引错误'
+  return s === 'unknown' ? t('索引状态未获取', 'Index status unavailable') : s === 'idle' ? t('索引就绪', 'Index ready') : s === 'indexing' ? `${t('索引中', 'Indexing')} (${settingsStore.indexStatus.pending_jobs})` : t('索引错误', 'Index error')
 })
 
 const aiCoreStatusText = computed(() => {
   const map: Record<string, string> = {
-    unknown: 'AI Core 状态未获取',
+    unknown: t('AI Core 状态未获取', 'AI Core status unavailable'),
-    starting: 'AI Core 启动中',
+    starting: t('AI Core 启动中', 'AI Core starting'),
-    running: 'AI Core 运行中',
+    running: t('AI Core 运行中', 'AI Core running'),
-    stopped: 'AI Core 已停止',
+    stopped: t('AI Core 已停止', 'AI Core stopped'),
-    error: 'AI Core 错误',
+    error: t('AI Core 错误', 'AI Core error'),
   }
   return map[settingsStore.aiCoreStatus] || ''
 })
@@ -85,7 +86,7 @@ const showEditorInfo = computed(() => route.name === 'workspace')
       </span>
       <span v-if="agentStore.isRunning" class="status-item agent-status">
         <span class="spinner" />
-        智能体运行中
+        {{ t('智能体运行中', 'Agent running') }}
       </span>
     </div>
     <div class="statusbar-right">
@@ -93,10 +94,10 @@ const showEditorInfo = computed(() => route.name === 'workspace')
         {{ defaultProvider.name }} · {{ defaultProvider.default_model }}
       </span>
       <span v-if="showEditorInfo" class="status-item">
-        {{ editorStore.lineCount }} 行
+        {{ editorStore.lineCount }} {{ t('行', 'lines') }}
       </span>
       <span v-if="showEditorInfo" class="status-item">
-        {{ editorStore.wordCount }} 字
+        {{ editorStore.wordCount }} {{ t('字', 'words') }}
       </span>
     </div>
   </footer>

frontend/src/components/common/TitleBar.vue

@@ -6,6 +6,7 @@ import { useEditorStore } from '@/stores/editor'
 import { useThemeStore } from '@/stores/theme'
 import { Moon, Sunny } from '@element-plus/icons-vue'
 import AppIcon from './AppIcon.vue'
+import { t } from '@/i18n'
 
 const route = useRoute()
 const workspaceStore = useWorkspaceStore()
@@ -15,15 +16,15 @@ const themeStore = useThemeStore()
 const pageTitle = computed(() => {
   const name = route.name as string
   const titles: Record<string, string> = {
-    workspace: '工作区',
+    workspace: t('工作区', 'Workspace'),
-    search: '搜索',
+    search: t('搜索', 'Search'),
-    chat: 'AI 对话',
+    chat: t('AI 对话', 'AI Chat'),
-    agent: '智能体执行轨迹',
+    agent: t('智能体执行轨迹', 'Agent Trace'),
-    tasks: '任务',
+    tasks: t('任务', 'Tasks'),
-    skills: 'Skill 管理',
+    skills: t('Skill 管理', 'Skill Management'),
-    plugins: 'Plugin 与 MCP',
+    plugins: t('Plugin 与 MCP', 'Plugins and MCP'),
-    themes: '主题管理',
+    themes: t('主题管理', 'Theme Management'),
-    settings: '设置',
+    settings: t('设置', 'Settings'),
   }
   return titles[name] || 'NotesAgent'
 })
@@ -53,7 +54,7 @@ const isDirty = computed(() => editorStore.saveStatus === 'dirty' || editorStore
       <span class="app-name">NotesAgent</span>
     </div>
     <div class="titlebar-right">
-      <button class="icon-btn" @click="themeStore.toggleTheme()" :title="themeStore.isDark ? '切换浅色主题' : '切换深色主题'">
+      <button class="icon-btn" @click="themeStore.toggleTheme()" :title="themeStore.isDark ? t('切换浅色主题', 'Switch to light theme') : t('切换深色主题', 'Switch to dark theme')">
         <AppIcon :icon="themeStore.isDark ? Sunny : Moon" :size="16" />
       </button>
       <div class="window-controls">

frontend/src/features/agent/AgentView.vue

@@ -7,6 +7,7 @@ import { useSkillStore } from '@/stores/skill'
 import type { AgentEvent } from '@/contracts'
 import { eventLabel, localizeDetails, permissionLabel, runStatusLabel, toolLabel } from './labels'
 import ToolOption from './ToolOption.vue'
+import { localeTag, t } from '@/i18n'
 
 const route = useRoute()
 const router = useRouter()
@@ -27,19 +28,19 @@ onMounted(async () => {
   try {
     await Promise.all([providerStore.loadProviders(), skillStore.loadSkills(), agentStore.loadTools()])
     form.provider_id = providerStore.defaultProviderId
-  } catch (error) { pageError.value = error instanceof Error ? error.message : '智能体配置加载失败' }
+  } catch (error) { pageError.value = error instanceof Error ? error.message : t('智能体配置加载失败', 'Failed to load agent configuration') }
 })
 
 watch(() => route.params.runId, async (runId) => {
   if (typeof runId !== 'string') return
-  try { await agentStore.loadRun(runId) } catch (error) { pageError.value = error instanceof Error ? error.message : '运行记录加载失败' }
+  try { await agentStore.loadRun(runId) } catch (error) { pageError.value = error instanceof Error ? error.message : t('运行记录加载失败', 'Failed to load run') }
 }, { immediate: true })
 
 watch(() => form.provider_id, async (providerId) => {
   form.model = providerStore.providers.find(p => p.provider_id === providerId)?.default_model ?? ''
   if (!providerId) return
   try { await providerStore.loadModels(providerId) }
-  catch (error) { if (form.provider_id === providerId) pageError.value = error instanceof Error ? error.message : '模型列表加载失败，请手动填写模型 ID。' }
+  catch (error) { if (form.provider_id === providerId) pageError.value = error instanceof Error ? error.message : t('模型列表加载失败，请手动填写模型 ID。', 'Unable to load models. Enter a model ID manually.') }
 })
 
 function toggleTool(name: string) {
@@ -51,7 +52,7 @@ function toggleTool(name: string) {
 async function createRun() {
   pageError.value = ''
   try {
-    if (!form.provider_id || !form.model.trim()) throw new Error('请选择提供商并填写模型 ID。')
+    if (!form.provider_id || !form.model.trim()) throw new Error(t('请选择提供商并填写模型 ID。', 'Select a provider and enter a model ID.'))
     const run = await agentStore.createRun({
       input: form.input, provider_id: form.provider_id, model: form.model,
       skill_id: form.skill_id || undefined, allowed_tools: form.allowed_tools,
@@ -60,12 +61,12 @@ async function createRun() {
       allow_network: form.allow_network, max_concurrent_tools: form.max_concurrent_tools,
     })
     await router.replace({ name: 'agent', params: { runId: run.run_id } })
-  } catch (error) { pageError.value = error instanceof Error ? error.message : '运行创建失败' }
+  } catch (error) { pageError.value = error instanceof Error ? error.message : t('运行创建失败', 'Failed to create run') }
 }
 
 function eventText(event: AgentEvent) {
-  if (event.event === 'RunCompleted') return '任务已成功完成。'
+  if (event.event === 'RunCompleted') return t('任务已成功完成。', 'The task completed successfully.')
-  if (event.event === 'RunCancelled') return '任务已取消。'
+  if (event.event === 'RunCancelled') return t('任务已取消。', 'The task was cancelled.')
   const text = event.data.text ?? event.data.message ?? event.data.code
   if (text) return String(text)
   return ''
@@ -74,40 +75,40 @@ function eventText(event: AgentEvent) {
 
 <template>
   <section class="feature-page agent-page">
-    <header class="feature-header"><div><h1>{{ isNewRun ? '创建智能体运行' : '智能体执行轨迹' }}</h1><p>配置执行边界，并实时查看模型、工具和权限事件。</p></div>
+    <header class="feature-header"><div><h1>{{ isNewRun ? t('创建智能体运行', 'Create Agent Run') : t('智能体执行轨迹', 'Agent Trace') }}</h1><p>{{ t('配置执行边界，并实时查看模型、工具和权限事件。', 'Configure execution limits and inspect model, tool, and permission events in real time.') }}</p></div>
-      <button v-if="!isNewRun" class="button-secondary" @click="router.push({ name: 'agent' })">新建运行</button></header>
+      <button v-if="!isNewRun" class="button-secondary" @click="router.push({ name: 'agent' })">{{ t('新建运行', 'New run') }}</button></header>
     <div v-if="pageError || agentStore.error || providerStore.error" class="error-banner">{{ pageError || agentStore.error || providerStore.error }}</div>
     <form v-if="isNewRun" class="panel run-form" @submit.prevent="createRun">
-      <div class="field"><label>任务</label><textarea v-model="form.input" class="textarea" required placeholder="描述希望智能体完成的任务" /></div>
+      <div class="field"><label>{{ t('任务', 'Task') }}</label><textarea v-model="form.input" class="textarea" required :placeholder="t('描述希望智能体完成的任务', 'Describe the task for the agent')" /></div>
       <div class="form-grid">
-        <div class="field"><label>模型提供商</label><select v-model="form.provider_id" class="select"><option v-for="p in providerStore.enabledProviders" :key="p.provider_id" :value="p.provider_id">{{ p.name }}</option></select></div>
+        <div class="field"><label>{{ t('模型提供商', 'Model provider') }}</label><select v-model="form.provider_id" class="select"><option v-for="p in providerStore.enabledProviders" :key="p.provider_id" :value="p.provider_id">{{ p.name }}</option></select></div>
-        <div class="field"><label>模型</label><input v-model="form.model" class="input" list="agent-models" placeholder="填写模型 ID" required /><datalist id="agent-models"><option v-for="m in models" :key="m.model_id" :value="m.model_id">{{ m.name }}</option></datalist></div>
+        <div class="field"><label>{{ t('模型', 'Model') }}</label><input v-model="form.model" class="input" list="agent-models" :placeholder="t('填写模型 ID', 'Enter model ID')" required /><datalist id="agent-models"><option v-for="m in models" :key="m.model_id" :value="m.model_id">{{ m.name }}</option></datalist></div>
-        <div class="field"><label>技能</label><select v-model="form.skill_id" class="select"><option value="">不使用技能</option><option v-for="s in skillStore.readySkills" :key="s.skill_id" :value="s.skill_id">{{ s.name }}</option></select></div>
+        <div class="field"><label>{{ t('技能', 'Skill') }}</label><select v-model="form.skill_id" class="select"><option value="">{{ t('不使用技能', 'No skill') }}</option><option v-for="s in skillStore.readySkills" :key="s.skill_id" :value="s.skill_id">{{ s.name }}</option></select></div>
-        <div class="field"><label>最大步骤</label><input v-model.number="form.max_steps" class="input" type="number" min="1" max="100" /></div>
+        <div class="field"><label>{{ t('最大步骤', 'Maximum steps') }}</label><input v-model.number="form.max_steps" class="input" type="number" min="1" max="100" /></div>
-        <div class="field"><label>工具超时（秒）</label><input v-model.number="form.tool_timeout_seconds" class="input" type="number" min="1" /></div>
+        <div class="field"><label>{{ t('工具超时（秒）', 'Tool timeout (seconds)') }}</label><input v-model.number="form.tool_timeout_seconds" class="input" type="number" min="1" /></div>
-        <div class="field"><label>运行超时（秒）</label><input v-model.number="form.run_timeout_seconds" class="input" type="number" min="1" /></div>
+        <div class="field"><label>{{ t('运行超时（秒）', 'Run timeout (seconds)') }}</label><input v-model.number="form.run_timeout_seconds" class="input" type="number" min="1" /></div>
-        <div class="field"><label>令牌预算</label><input v-model.number="form.token_budget" class="input" type="number" min="1" /></div>
+        <div class="field"><label>{{ t('令牌预算', 'Token budget') }}</label><input v-model.number="form.token_budget" class="input" type="number" min="1" /></div>
-        <div class="field"><label>最大并发工具</label><input v-model.number="form.max_concurrent_tools" class="input" type="number" min="1" /></div>
+        <div class="field"><label>{{ t('最大并发工具', 'Maximum concurrent tools') }}</label><input v-model.number="form.max_concurrent_tools" class="input" type="number" min="1" /></div>
       </div>
-      <div class="field"><label>允许使用的工具</label><div class="tool-grid"><ToolOption v-for="tool in agentStore.tools" :key="tool.name" :name="tool.name" :description="tool.description" :selected="form.allowed_tools.includes(tool.name)" @toggle="toggleTool" /></div></div>
+      <div class="field"><label>{{ t('允许使用的工具', 'Allowed tools') }}</label><div class="tool-grid"><ToolOption v-for="tool in agentStore.tools" :key="tool.name" :name="tool.name" :description="tool.description" :selected="form.allowed_tools.includes(tool.name)" @toggle="toggleTool" /></div></div>
-      <label class="network"><input v-model="form.allow_network" type="checkbox" /> 允许本次运行调用网络工具</label>
+      <label class="network"><input v-model="form.allow_network" type="checkbox" /> {{ t('允许本次运行调用网络工具', 'Allow network tools for this run') }}</label>
-      <div class="inline-actions"><button class="button-primary" :disabled="agentStore.isCreating || !form.input.trim() || !form.provider_id || !form.model.trim()">{{ agentStore.isCreating ? '创建中…' : '创建并运行' }}</button></div>
+      <div class="inline-actions"><button class="button-primary" :disabled="agentStore.isCreating || !form.input.trim() || !form.provider_id || !form.model.trim()">{{ agentStore.isCreating ? t('创建中…', 'Creating…') : t('创建并运行', 'Create and run') }}</button></div>
     </form>
 
     <div v-else class="trace-layout">
-      <div class="panel run-summary"><div><span class="badge info">{{ runStatusLabel(agentStore.activeRun?.status) }}</span><h2>{{ agentStore.activeRunId }}</h2></div><div class="inline-actions"><span>步骤 {{ agentStore.currentStep }} / {{ agentStore.activeRun?.max_steps }}</span><button v-if="agentStore.isRunning" class="button-danger" @click="agentStore.cancelRun(agentStore.activeRunId!)">取消运行</button></div></div>
+      <div class="panel run-summary"><div><span class="badge info">{{ runStatusLabel(agentStore.activeRun?.status) }}</span><h2>{{ agentStore.activeRunId }}</h2></div><div class="inline-actions"><span>{{ t('步骤', 'Step') }} {{ agentStore.currentStep }} / {{ agentStore.activeRun?.max_steps }}</span><button v-if="agentStore.isRunning" class="button-danger" @click="agentStore.cancelRun(agentStore.activeRunId!)">{{ t('取消运行', 'Cancel run') }}</button></div></div>
       <div class="timeline">
         <article v-for="event in agentStore.events" :key="event.sequence" class="event-card item-card">
-          <div class="event-head"><span class="badge" :class="{ success: event.event === 'RunCompleted', error: event.event === 'RunFailed', warning: event.event === 'PermissionRequired' }">{{ eventLabel(event.event) }}</span><span>第 {{ event.sequence }} 条 · {{ new Date(event.timestamp).toLocaleTimeString() }}</span></div>
+          <div class="event-head"><span class="badge" :class="{ success: event.event === 'RunCompleted', error: event.event === 'RunFailed', warning: event.event === 'PermissionRequired' }">{{ eventLabel(event.event) }}</span><span>#{{ event.sequence }} · {{ new Date(event.timestamp).toLocaleTimeString(localeTag()) }}</span></div>
           <p v-if="eventText(event)" class="event-text">{{ eventText(event) }}</p>
           <pre v-if="['ToolCall', 'ToolResult', 'Citation', 'Usage'].includes(event.event)">{{ JSON.stringify(localizeDetails(event.data), null, 2) }}</pre>
         </article>
-        <div v-if="!agentStore.events.length" class="empty-state"><div><strong>等待执行轨迹</strong><p>事件连接建立后将在这里实时显示。</p></div></div>
+        <div v-if="!agentStore.events.length" class="empty-state"><div><strong>{{ t('等待执行轨迹', 'Waiting for trace events') }}</strong><p>{{ t('事件连接建立后将在这里实时显示。', 'Events will appear here after the connection is established.') }}</p></div></div>
       </div>
     </div>
 
     <div v-if="agentStore.permissionRequest" class="modal-backdrop">
-      <div class="modal"><span class="badge warning">权限确认</span><h2>{{ toolLabel(agentStore.permissionRequest.tool_name) }}</h2><p>{{ agentStore.permissionRequest.impact }}</p><p class="subtle">所需权限：{{ permissionLabel(agentStore.permissionRequest.permission) }}（{{ agentStore.permissionRequest.permission }}）</p><pre>{{ JSON.stringify(localizeDetails(agentStore.permissionRequest.parameters), null, 2) }}</pre><div class="inline-actions permission-actions"><button class="button-primary" @click="agentStore.respondPermission('allow', 'once')">仅本次允许</button><button class="button-secondary" @click="agentStore.respondPermission('allow', 'session')">本次会话允许</button><button class="button-danger" @click="agentStore.respondPermission('deny')">拒绝</button></div></div>
+      <div class="modal"><span class="badge warning">{{ t('权限确认', 'Permission Confirmation') }}</span><h2>{{ toolLabel(agentStore.permissionRequest.tool_name) }}</h2><p>{{ agentStore.permissionRequest.impact }}</p><p class="subtle">{{ t('所需权限：', 'Required permission: ') }}{{ permissionLabel(agentStore.permissionRequest.permission) }} ({{ agentStore.permissionRequest.permission }})</p><pre>{{ JSON.stringify(localizeDetails(agentStore.permissionRequest.parameters), null, 2) }}</pre><div class="inline-actions permission-actions"><button class="button-primary" @click="agentStore.respondPermission('allow', 'once')">{{ t('仅本次允许', 'Allow once') }}</button><button class="button-secondary" @click="agentStore.respondPermission('allow', 'session')">{{ t('本次会话允许', 'Allow for session') }}</button><button class="button-danger" @click="agentStore.respondPermission('deny')">{{ t('拒绝', 'Deny') }}</button></div></div>
     </div>
   </section>
 </template>

frontend/src/features/agent/RunListPanel.vue

@@ -2,6 +2,7 @@
 import { onMounted, ref } from 'vue'
 import { useRouter } from 'vue-router'
 import { useAgentStore } from '@/stores/agent'
+import { localeTag, t } from '@/i18n'
 import { runStatusLabel } from './labels'
 
 const agentStore = useAgentStore()
@@ -9,7 +10,7 @@ const router = useRouter()
 const error = ref('')
 
 onMounted(async () => {
-  try { await agentStore.loadRuns() } catch (reason) { error.value = reason instanceof Error ? reason.message : '运行记录加载失败' }
+  try { await agentStore.loadRuns() } catch (reason) { error.value = reason instanceof Error ? reason.message : t('运行记录加载失败', 'Failed to load runs') }
 })
 
 function selectRun(runId: string) { void router.push({ name: 'agent', params: { runId } }) }
@@ -17,13 +18,13 @@ function selectRun(runId: string) { void router.push({ name: 'agent', params: {
 
 <template>
   <div class="sidebar-panel">
-    <button class="button-primary new-button" @click="router.push({ name: 'agent' })">＋ 新建运行</button>
+    <button class="button-primary new-button" @click="router.push({ name: 'agent' })">＋ {{ t('新建运行', 'New run') }}</button>
     <p v-if="error" class="subtle error-text">{{ error }}</p>
     <div class="sidebar-list">
       <button v-for="run in agentStore.sortedRuns" :key="run.run_id" class="sidebar-list-item run-item"
         :class="{ active: agentStore.activeRunId === run.run_id }" @click="selectRun(run.run_id)">
         <span class="badge" :class="{ success: run.status === 'completed', error: run.status === 'failed', warning: run.status === 'waiting_permission' }">{{ runStatusLabel(run.status) }}</span>
-        <strong>{{ run.run_id.slice(0, 12) }}</strong><small>{{ run.started_at ? new Date(run.started_at).toLocaleString() : '等待开始' }}</small>
+        <strong>{{ run.run_id.slice(0, 12) }}</strong><small>{{ run.started_at ? new Date(run.started_at).toLocaleString(localeTag()) : t('等待开始', 'Waiting to start') }}</small>
       </button>
     </div>
   </div>

frontend/src/features/agent/ToolOption.vue

@@ -1,6 +1,7 @@
 <script setup lang="ts">
 import { computed } from 'vue'
 import { toolDescription, toolLabel } from './labels'
+import { t } from '@/i18n'
 
 const props = defineProps<{ name: string; description: string; selected: boolean }>()
 const emit = defineEmits<{ toggle: [name: string] }>()
@@ -19,7 +20,7 @@ const showOriginal = computed(() => props.description.length > 0)
       </span>
     </label>
     <details v-if="showOriginal" class="tool-original">
-      <summary>查看服务原文与参数</summary>
+      <summary>{{ t('查看服务原文与参数', 'View original service description and parameters') }}</summary>
       <p>{{ description }}</p>
     </details>
   </article>

frontend/src/features/agent/labels.ts

@@ -1,4 +1,5 @@
 import type { AgentEventType, AgentRunStatus } from '@/contracts'
+import { appLocale, t } from '@/i18n'
 
 const runStatusLabels: Record<AgentRunStatus, string> = {
   queued: '排队中',
@@ -27,6 +28,20 @@ const eventLabels: Record<AgentEventType, string> = {
   RunCancelled: '运行取消',
 }
 
+const runStatusLabelsEn: Record<AgentRunStatus, string> = {
+  queued: 'Queued', running: 'Running', waiting_permission: 'Waiting for permission',
+  completed: 'Completed', failed: 'Failed', cancelled: 'Cancelled',
+}
+
+const eventLabelsEn: Record<AgentEventType, string> = {
+  RunStarted: 'Run started', TextDelta: 'Response', ThinkingDelta: 'Reasoning',
+  ToolCall: 'Tool call', ToolResult: 'Tool result', PermissionRequired: 'Permission required',
+  Usage: 'Usage', Citation: 'Citation', ModelCallStarted: 'Model call started',
+  ModelCallCompleted: 'Model call completed', ModelCallFailed: 'Model call failed',
+  PermissionResolved: 'Permission resolved', RunCompleted: 'Run completed',
+  RunFailed: 'Run failed', RunCancelled: 'Run cancelled',
+}
+
 const toolLabels: Record<string, string> = {
   'system.echo': '回显测试',
   'math.add': '数值相加',
@@ -116,45 +131,50 @@ const detailLabels: Record<string, string> = {
 }
 
 export function runStatusLabel(status?: AgentRunStatus): string {
-  return status ? runStatusLabels[status] : '未知状态'
+  if (!status) return t('未知状态', 'Unknown status')
+  return appLocale.value === 'en' ? runStatusLabelsEn[status] : runStatusLabels[status]
 }
 
 export function eventLabel(event: AgentEventType): string {
-  return eventLabels[event]
+  return appLocale.value === 'en' ? eventLabelsEn[event] : eventLabels[event]
 }
 
 export function toolLabel(name: string): string {
   const remote = mcpName(name)
-  if (remote) return mcpTools[remote]?.label ?? `MCP 工具 · ${remote}`
+  if (remote) return appLocale.value === 'en' ? `MCP Tool · ${remote}` : (mcpTools[remote]?.label ?? `MCP 工具 · ${remote}`)
+  if (appLocale.value === 'en') return name.split('.').map(part => part[0]?.toUpperCase() + part.slice(1)).join(' ')
   return toolLabels[name] ?? name
 }
 
 export function toolDescription(name: string, fallback: string): string {
   const remote = mcpName(name)
   if (remote) {
+    if (appLocale.value === 'en') return fallback && !/\p{Script=Han}/u.test(fallback) ? fallback : `MCP tool ${remote}. See the original service description for full parameters.`
     if (/\p{Script=Han}/u.test(fallback)) return fallback
     return mcpTools[remote]?.description ?? '暂无中文说明，请展开查看服务原文。'
   }
+  if (appLocale.value === 'en') return fallback && !/\p{Script=Han}/u.test(fallback) ? fallback : `Built-in tool: ${name}`
   return toolDescriptions[name] ?? fallback
 }
 
 export function permissionLabel(permission: string): string {
+  if (appLocale.value === 'en') return permission.split('.').map(part => part[0]?.toUpperCase() + part.slice(1)).join(' ')
   return permissionLabels[permission] ?? permission
 }
 
 function localizeValue(value: unknown): unknown {
   if (Array.isArray(value)) return value.map(localizeValue)
   if (value && typeof value === 'object') return localizeDetails(value as Record<string, unknown>)
-  if (value === true) return '是'
+  if (value === true) return t('是', 'Yes')
-  if (value === false) return '否'
+  if (value === false) return t('否', 'No')
   if (typeof value === 'string' && value in runStatusLabels) {
-    return runStatusLabels[value as AgentRunStatus]
+    return runStatusLabel(value as AgentRunStatus)
   }
   return value
 }
 
 export function localizeDetails(data: Record<string, unknown>): Record<string, unknown> {
   return Object.fromEntries(
-    Object.entries(data).map(([key, value]) => [detailLabels[key] ?? key, localizeValue(value)])
+    Object.entries(data).map(([key, value]) => [appLocale.value === 'en' ? key.replaceAll('_', ' ') : (detailLabels[key] ?? key), localizeValue(value)])
   )
 }

frontend/src/features/chat/ChatView.spec.ts

@@ -11,6 +11,10 @@ vi.mock('vue-router', () => ({ useRouter: () => ({ push: vi.fn() }) }))
 vi.mock('@/stores/editor', () => ({ useEditorStore: () => ({}) }))
 vi.mock('@/stores/workspace', () => ({ useWorkspaceStore: () => ({}) }))
 vi.mock('@/components/common/MarkdownContent.vue', () => ({ default: { template: '<div />' } }))
+vi.mock('@/services/chatService', () => ({
+  listConversations: vi.fn().mockResolvedValue({ items: [], page: { total: 0, limit: 100, offset: 0 } }),
+  listConversationMessages: vi.fn(), createConversation: vi.fn(), removeConversation: vi.fn(), streamChat: vi.fn(),
+}))
 
 beforeEach(() => {
   setActivePinia(createPinia())

frontend/src/features/chat/ChatView.vue

@@ -8,6 +8,7 @@ import { useProviderStore } from '@/stores/provider'
 import { useSkillStore } from '@/stores/skill'
 import { useWorkspaceStore } from '@/stores/workspace'
 import MarkdownContent from '@/components/common/MarkdownContent.vue'
+import { t } from '@/i18n'
 
 const chatStore = useChatStore()
 const providerStore = useProviderStore()
@@ -23,7 +24,7 @@ const availableModels = computed(() => providerStore.modelsByProvider[chatStore.
 
 onMounted(async () => {
   try {
-    await Promise.all([providerStore.loadProviders(), skillStore.loadSkills()])
+    await Promise.all([providerStore.loadProviders(), skillStore.loadSkills(), chatStore.loadConversations()])
     if (disposed || providerStore.error) return
     const selected = providerStore.enabledProviders.find(p => p.provider_id === chatStore.selectedProviderId)
     if (!selected) {
@@ -33,7 +34,7 @@ onMounted(async () => {
     }
   } catch (error) {
     if (disposed) return
-    loadError.value = error instanceof Error ? error.message : '无法加载 AI 配置，请检查后端连接。'
+    loadError.value = error instanceof Error ? error.message : t('无法加载 AI 配置，请检查后端连接。', 'Unable to load AI configuration. Check the backend connection.')
   }
 })
 
@@ -41,7 +42,7 @@ async function refreshModels(providerId: string) {
   loadError.value = ''
   if (!providerId) return
   try { await providerStore.loadModels(providerId) }
-  catch (error) { if (!disposed && chatStore.selectedProviderId === providerId) loadError.value = error instanceof Error ? error.message : '模型列表加载失败，请手动填写模型 ID。' }
+  catch (error) { if (!disposed && chatStore.selectedProviderId === providerId) loadError.value = error instanceof Error ? error.message : t('模型列表加载失败，请手动填写模型 ID。', 'Unable to load models. Enter a model ID manually.') }
 }
 
 watch(() => chatStore.selectedProviderId, async (providerId) => {
@@ -65,19 +66,19 @@ async function openCitation(citation: Citation) {
       <div class="field compact"><label>Provider</label><select v-model="chatStore.selectedProviderId" class="select">
         <option v-for="provider in providerStore.enabledProviders" :key="provider.provider_id" :value="provider.provider_id">{{ provider.name }}</option>
       </select></div>
-      <div class="field compact"><label>模型 ID</label><input v-model="chatStore.selectedModel" class="input" list="chat-models" placeholder="填写模型 ID" /><datalist id="chat-models"><option v-for="model in availableModels" :key="model.model_id" :value="model.model_id">{{ model.name }}</option></datalist></div>
+      <div class="field compact"><label>{{ t('模型 ID', 'Model ID') }}</label><input v-model="chatStore.selectedModel" class="input" list="chat-models" :placeholder="t('填写模型 ID', 'Enter model ID')" /><datalist id="chat-models"><option v-for="model in availableModels" :key="model.model_id" :value="model.model_id">{{ model.name }}</option></datalist></div>
-      <label class="rag-toggle"><input v-model="chatStore.useRag" type="checkbox" :disabled="chatStore.isStreaming" />检索知识库</label>
+      <label class="rag-toggle"><input v-model="chatStore.useRag" type="checkbox" :disabled="chatStore.isStreaming" />{{ t('检索知识库', 'Search knowledge base') }}</label>
-      <span class="subtle">开启后，将相关笔记片段发送给所选模型，并显示来源。技能调用请使用智能体。</span>
+      <span class="subtle">{{ t('开启后，将相关笔记片段发送给所选模型，并显示来源。技能调用请使用智能体。', 'When enabled, relevant note excerpts are sent to the selected model and citations are shown. Use Agent for skills.') }}</span>
     </header>
-    <div v-if="loadError || providerStore.error" class="error-banner chat-error">{{ loadError || providerStore.error }}</div>
+    <div v-if="loadError || providerStore.error || chatStore.historyError" class="error-banner chat-error">{{ loadError || providerStore.error || chatStore.historyError }}</div>
     <main class="message-timeline">
-      <div v-if="!chatStore.messages.length" class="empty-state"><div><strong>开始一段知识对话</strong><p>请先配置模型提供商。聊天记录仅保留在本次页面会话中。</p></div></div>
+      <div v-if="!chatStore.messages.length" class="empty-state"><div><strong>{{ t('开始一段知识对话', 'Start a knowledge conversation') }}</strong><p>{{ t('请先配置模型提供商。聊天记录保存在本地数据库中。', 'Configure a model provider first. Messages are saved in the local database.') }}</p></div></div>
       <article v-for="message in chatStore.messages" :key="message.message_id" class="message" :class="message.role">
-        <div class="avatar">{{ message.role === 'user' ? '你' : 'AI' }}</div>
+        <div class="avatar">{{ message.role === 'user' ? t('你', 'You') : 'AI' }}</div>
         <div class="message-body">
-          <details v-if="message.thinking" class="thinking"><summary>思考过程</summary><p>{{ message.thinking }}</p></details>
+          <details v-if="message.thinking" class="thinking"><summary>{{ t('思考过程', 'Reasoning') }}</summary><p>{{ message.thinking }}</p></details>
           <MarkdownContent v-if="message.content" class="message-content" :source="message.content" />
-          <div v-else-if="chatStore.isStreaming" class="message-content">正在思考…</div>
+          <div v-else-if="chatStore.isStreaming" class="message-content">{{ t('正在思考…', 'Thinking…') }}</div>
           <div v-if="message.tool_calls?.length" class="tool-calls"><div v-for="call in message.tool_calls" :key="call.tool_call_id" class="item-card"><span class="badge info">{{ call.status }}</span><strong>{{ call.name }}</strong><pre>{{ JSON.stringify(call.parameters, null, 2) }}</pre></div></div>
           <div v-if="message.citations?.length" class="citations">
             <button v-for="(citation, index) in message.citations" :key="citation.block_id" class="citation-card" @click="openCitation(citation)">
@@ -85,16 +86,16 @@ async function openCitation(citation: Citation) {
             </button>
           </div>
           <time>{{ new Date(message.created_at).toLocaleTimeString() }}</time>
-          <small v-if="message.usage" class="usage">Token {{ message.usage.total_tokens }}<span v-if="message.usage.input_tokens !== undefined && message.usage.output_tokens !== undefined">（输入 {{ message.usage.input_tokens }} / 输出 {{ message.usage.output_tokens }}）</span></small>
+          <small v-if="message.usage" class="usage">Token {{ message.usage.total_tokens }}<span v-if="message.usage.input_tokens !== undefined && message.usage.output_tokens !== undefined"> ({{ t('输入', 'input') }} {{ message.usage.input_tokens }} / {{ t('输出', 'output') }} {{ message.usage.output_tokens }})</span></small>
         </div>
       </article>
     </main>
     <footer class="composer">
-      <textarea v-model="chatStore.inputText" class="textarea" placeholder="输入问题，Ctrl + Enter 发送"
+      <textarea v-model="chatStore.inputText" class="textarea" :placeholder="t('输入问题，Ctrl + Enter 发送', 'Enter a question; press Ctrl + Enter to send')"
         @keydown.ctrl.enter.prevent="send" />
-      <div class="composer-actions"><span class="subtle">回答可能包含错误，请核对 Citation。</span>
+      <div class="composer-actions"><span class="subtle">{{ t('回答可能包含错误，请核对 Citation。', 'Answers may contain errors. Verify the citations.') }}</span>
-        <button v-if="chatStore.isStreaming" class="button-danger" @click="chatStore.stopGeneration">停止</button>
+        <button v-if="chatStore.isStreaming || chatStore.isPreparing" class="button-danger" @click="chatStore.stopGeneration">{{ t('停止', 'Stop') }}</button>
-        <button v-else class="button-primary" :disabled="!chatStore.inputText.trim() || !chatStore.selectedProviderId || !chatStore.selectedModel.trim()" @click="send">发送</button>
+        <button v-else class="button-primary" :disabled="!chatStore.canSend || !chatStore.inputText.trim() || !chatStore.selectedProviderId || !chatStore.selectedModel.trim()" @click="send">{{ t('发送', 'Send') }}</button>
       </div>
     </footer>
   </section>

frontend/src/features/chat/ConversationListPanel.vue

@@ -1,18 +1,21 @@
 <script setup lang="ts">
+import { onMounted } from 'vue'
 import { useChatStore } from '@/stores/chat'
+import { t } from '@/i18n'
 
 const chatStore = useChatStore()
+onMounted(() => { void chatStore.loadConversations() })
 </script>
 
 <template>
   <div class="sidebar-panel">
-    <button class="button-primary new-button" @click="chatStore.createNewConversation">＋ 新对话</button>
+    <button class="button-primary new-button" @click="chatStore.createNewConversation">＋ {{ t('新对话', 'New conversation') }}</button>
     <div class="sidebar-list conversation-list">
       <div v-for="conversation in chatStore.sortedConversations" :key="conversation.conversation_id"
         class="sidebar-list-item conversation" :class="{ active: chatStore.activeConversationId === conversation.conversation_id }"
         @click="chatStore.setActiveConversation(conversation.conversation_id)">
-        <div><strong>{{ conversation.title }}</strong><p>{{ conversation.message_count }} 条消息</p></div>
+        <div><strong>{{ conversation.title }}</strong><p>{{ conversation.message_count }} {{ t('条消息', 'messages') }}</p></div>
-        <button class="delete" title="删除会话" @click.stop="chatStore.deleteConversation(conversation.conversation_id)">×</button>
+        <button class="delete" :title="t('删除会话', 'Delete conversation')" @click.stop="chatStore.deleteConversation(conversation.conversation_id)">×</button>
       </div>
     </div>
   </div>

frontend/src/features/editor/EditorHeader.vue

@@ -1,26 +1,28 @@
 <script setup lang="ts">
 import { useEditorStore } from '@/stores/editor'
 import { useWorkspaceStore } from '@/stores/workspace'
+import { computed } from 'vue'
+import { t } from '@/i18n'
 
 const editorStore = useEditorStore()
 const workspaceStore = useWorkspaceStore()
 
-const statusText: Record<string, string> = {
+const statusText = computed<Record<string, string>>(() => ({
-  idle: '空闲', dirty: '未保存', saving: '保存中…', saved: '已保存', save_failed: '保存失败',
+  idle: t('空闲', 'Idle'), dirty: t('未保存', 'Unsaved'), saving: t('保存中…', 'Saving…'), saved: t('已保存', 'Saved'), save_failed: t('保存失败', 'Save failed'),
-  external_changed: '外部文件已变化', conflict: '存在编辑冲突',
+  external_changed: t('外部文件已变化', 'File changed externally'), conflict: t('存在编辑冲突', 'Edit conflict'),
-}
+}))
 </script>
 
 <template>
   <header class="editor-header">
-    <div class="file-identity"><strong>{{ workspaceStore.activeFile?.name ?? '未命名笔记' }}</strong><small>{{ workspaceStore.activeFilePath }}</small></div>
+    <div class="file-identity"><strong>{{ workspaceStore.activeFile?.name ?? t('未命名笔记', 'Untitled note') }}</strong><small>{{ workspaceStore.activeFilePath }}</small></div>
     <div class="editor-actions">
       <span class="save-status" :class="editorStore.saveStatus">{{ statusText[editorStore.saveStatus] }}</span>
-      <div class="mode-switch" aria-label="编辑模式">
+      <div class="mode-switch" :aria-label="t('编辑模式', 'Editor mode')">
-        <button type="button" :class="{ active: editorStore.mode === 'wysiwyg' }" @click="editorStore.setMode('wysiwyg')">写作</button>
+        <button type="button" :class="{ active: editorStore.mode === 'wysiwyg' }" @click="editorStore.setMode('wysiwyg')">{{ t('写作', 'Writing') }}</button>
-        <button type="button" :class="{ active: editorStore.mode === 'source' }" @click="editorStore.setMode('source')">源码</button>
+        <button type="button" :class="{ active: editorStore.mode === 'source' }" @click="editorStore.setMode('source')">{{ t('源码', 'Source') }}</button>
       </div>
-      <button type="button" class="save-button" :disabled="editorStore.saveStatus === 'saving'" @click="editorStore.save">保存</button>
+      <button type="button" class="save-button" :disabled="editorStore.saveStatus === 'saving'" @click="editorStore.save">{{ t('保存', 'Save') }}</button>
     </div>
   </header>
 </template>

frontend/src/features/editor/EditorPane.spec.ts

@@ -53,4 +53,19 @@ describe('EditorPane file switching', () => {
     expect(store.currentFilePath).toBe('/数据结构/红黑树.md')
     expect(wrapper.text()).not.toContain('祝你写作愉快')
   })
+
+  it('applies the saved spell-check and language settings to source mode', async () => {
+    const editor = useEditorStore()
+    const settings = (await import('@/stores/settings')).useSettingsStore()
+    editor.setMode('source')
+    settings.spellCheck = true
+    settings.language = 'en'
+    wrapper = mount(EditorPane, { attachTo: document.body })
+    await nextTick()
+
+    const textarea = wrapper.get('textarea')
+    expect(textarea.attributes('spellcheck')).toBe('true')
+    expect(textarea.attributes('lang')).toBe('en')
+    expect(textarea.attributes('aria-label')).toBe('Markdown source editor')
+  })
 })

frontend/src/features/editor/EditorPane.vue

@@ -14,10 +14,10 @@ function updateContent(event: Event) {
 </script>
 
 <template>
-  <VisualMarkdownEditor v-if="editorStore.mode === 'wysiwyg'" :key="`${editorStore.currentFilePath ?? 'empty'}:${themeStore.resolvedCodeBlockTheme}`"
+  <VisualMarkdownEditor v-if="editorStore.mode === 'wysiwyg'" :key="`${editorStore.currentFilePath ?? 'empty'}:${themeStore.resolvedCodeBlockTheme}:${settingsStore.language}`"
     :initial-content="editorStore.content" />
-  <textarea v-else class="editor-pane source" :value="editorStore.content" :spellcheck="false"
+  <textarea v-else class="editor-pane source" :value="editorStore.content" :spellcheck="settingsStore.spellCheck"
-    aria-label="Markdown 源码编辑器" @input="updateContent" />
+    :lang="settingsStore.language" :aria-label="settingsStore.language === 'en' ? 'Markdown source editor' : 'Markdown 源码编辑器'" @input="updateContent" />
 </template>
 
 <style scoped>

frontend/src/features/editor/VisualMarkdownEditor.spec.ts

@@ -6,6 +6,7 @@ import { editorViewCtx, type Editor } from '@milkdown/kit/core'
 import { TextSelection } from '@milkdown/kit/prose/state'
 import { getMarkdown } from '@milkdown/kit/utils'
 import VisualMarkdownEditor from './VisualMarkdownEditor.vue'
+import { useSettingsStore } from '@/stores/settings'
 
 type EditorComponent = { getEditor: () => Editor | undefined }
 
@@ -90,4 +91,19 @@ describe('VisualMarkdownEditor formatting toolbars', () => {
 
     expect(editor.action(getMarkdown()).trim()).toBe('alpha')
   })
+
+  it('updates native spell checking on the ProseMirror editor', async () => {
+    const settings = useSettingsStore()
+    const wrapper = mount(VisualMarkdownEditor, { props: { initialContent: 'mispelled word' }, attachTo: document.body })
+    mounted.push(wrapper)
+    await waitForEditor(wrapper)
+
+    settings.spellCheck = true
+    settings.language = 'en'
+    await wrapper.vm.$nextTick()
+
+    const editable = wrapper.get('.ProseMirror')
+    expect(editable.attributes('spellcheck')).toBe('true')
+    expect(editable.attributes('lang')).toBe('en')
+  })
 })

frontend/src/features/editor/VisualMarkdownEditor.vue

@@ -1,5 +1,5 @@
 <script setup lang="ts">
-import { onBeforeUnmount, onMounted, ref } from 'vue'
+import { onBeforeUnmount, onMounted, ref, watch } from 'vue'
 import { Link } from '@element-plus/icons-vue'
 import { Crepe } from '@milkdown/crepe'
 import { oneDark } from '@codemirror/theme-one-dark'
@@ -22,6 +22,7 @@ import { useEditorStore } from '@/stores/editor'
 import { useSettingsStore } from '@/stores/settings'
 import { useThemeStore } from '@/stores/theme'
 import { applyMarkdownFontSize, fontSizeMarkdownPlugin } from './fontSizeMarkdown'
+import { t } from '@/i18n'
 import '@milkdown/crepe/theme/common/style.css'
 import '@milkdown/crepe/theme/frame.css'
 
@@ -34,6 +35,14 @@ const loading = ref(true)
 const fontSizeInput = ref(16)
 let crepe: Crepe | null = null
 
+function applyProofingPreferences() {
+  const editable = editorRoot.value?.querySelector<HTMLElement>('.ProseMirror')
+  if (!editable) return
+  editable.spellcheck = settingsStore.spellCheck
+  editable.setAttribute('spellcheck', String(settingsStore.spellCheck))
+  editable.lang = settingsStore.language
+}
+
 type ToolbarCommand = 'bold' | 'italic' | 'ordered-list' | 'bullet-list' | 'inline-code' | 'code-block' | 'inline-math' | 'math-block'
 
 function runCommand(command: ToolbarCommand) {
@@ -57,14 +66,14 @@ function runCommand(command: ToolbarCommand) {
 function applyLink() {
   if (!crepe) return
   // TODO(editor): 用受控 Element Plus 对话框替换 prompt，补充 URL 校验和键盘焦点管理。
-  const href = window.prompt('请输入链接地址', 'https://')?.trim()
+  const href = window.prompt(t('请输入链接地址', 'Enter link address'), 'https://')?.trim()
   if (!href) return
 
   crepe.editor.action((ctx) => {
     const view = ctx.get(editorViewCtx)
     const commands = ctx.get(commandsCtx)
     if (view.state.selection.empty) {
-      const label = window.prompt('请输入链接文字', href)?.trim() || href
+      const label = window.prompt(t('请输入链接文字', 'Enter link text'), href)?.trim() || href
       const from = view.state.selection.from
       const transaction = view.state.tr.insertText(label, from)
       transaction.setSelection(TextSelection.create(transaction.doc, from, from + label.length))
@@ -107,56 +116,56 @@ onMounted(async () => {
     defaultValue: props.initialContent,
     features: { [Crepe.Feature.TopBar]: false },
     featureConfigs: {
-      [Crepe.Feature.Placeholder]: { text: '开始记录你的想法…' },
+      [Crepe.Feature.Placeholder]: { text: t('开始记录你的想法…', 'Start writing your thoughts…') },
       [Crepe.Feature.CodeMirror]: {
         theme: themeStore.resolvedCodeBlockTheme === 'github-dark' ? oneDark : [],
         previewOnlyByDefault: false,
-        searchPlaceholder: '搜索语言',
+        searchPlaceholder: t('搜索语言', 'Search languages'),
-        noResultText: '没有匹配的语言',
+        noResultText: t('没有匹配的语言', 'No matching language'),
-        copyText: '复制',
+        copyText: t('复制', 'Copy'),
       },
       [Crepe.Feature.Latex]: {
-        inlineEditConfirm: '确认',
+        inlineEditConfirm: t('确认', 'Confirm'),
       },
       [Crepe.Feature.LinkTooltip]: {
-        editButton: '编辑',
+        editButton: t('编辑', 'Edit'),
-        removeButton: '移除',
+        removeButton: t('移除', 'Remove'),
-        confirmButton: '确认',
+        confirmButton: t('确认', 'Confirm'),
-        inputPlaceholder: '粘贴链接地址…',
+        inputPlaceholder: t('粘贴链接地址…', 'Paste link address…'),
       },
       [Crepe.Feature.Toolbar]: {
-        boldLabel: '加粗',
+        boldLabel: t('加粗', 'Bold'),
-        italicLabel: '斜体',
+        italicLabel: t('斜体', 'Italic'),
-        strikethroughLabel: '删除线',
+        strikethroughLabel: t('删除线', 'Strikethrough'),
-        codeLabel: '行内代码',
+        codeLabel: t('行内代码', 'Inline code'),
-        latexLabel: '行内公式',
+        latexLabel: t('行内公式', 'Inline formula'),
-        linkLabel: '链接',
+        linkLabel: t('链接', 'Link'),
       },
       [Crepe.Feature.BlockEdit]: {
         textGroup: {
-          label: '文本',
+          label: t('文本', 'Text'),
-          text: { label: '正文' },
+          text: { label: t('正文', 'Paragraph') },
-          h1: { label: '一级标题' },
+          h1: { label: t('一级标题', 'Heading 1') },
-          h2: { label: '二级标题' },
+          h2: { label: t('二级标题', 'Heading 2') },
-          h3: { label: '三级标题' },
+          h3: { label: t('三级标题', 'Heading 3') },
-          h4: { label: '四级标题' },
+          h4: { label: t('四级标题', 'Heading 4') },
-          h5: { label: '五级标题' },
+          h5: { label: t('五级标题', 'Heading 5') },
-          h6: { label: '六级标题' },
+          h6: { label: t('六级标题', 'Heading 6') },
-          quote: { label: '引用' },
+          quote: { label: t('引用', 'Quote') },
-          divider: { label: '分割线' },
+          divider: { label: t('分割线', 'Divider') },
         },
         listGroup: {
-          label: '列表',
+          label: t('列表', 'Lists'),
-          bulletList: { label: '无序列表' },
+          bulletList: { label: t('无序列表', 'Bullet list') },
-          orderedList: { label: '有序列表' },
+          orderedList: { label: t('有序列表', 'Ordered list') },
-          taskList: { label: '任务列表' },
+          taskList: { label: t('任务列表', 'Task list') },
         },
         advancedGroup: {
-          label: '插入',
+          label: t('插入', 'Insert'),
-          image: { label: '图片' },
+          image: { label: t('图片', 'Image') },
-          codeBlock: { label: '代码块' },
+          codeBlock: { label: t('代码块', 'Code block') },
-          table: { label: '表格' },
+          table: { label: t('表格', 'Table') },
-          math: { label: '公式块' },
+          math: { label: t('公式块', 'Formula block') },
         },
       },
     },
@@ -171,9 +180,12 @@ onMounted(async () => {
     })
   })
   await crepe.create()
+  applyProofingPreferences()
   loading.value = false
 })
 
+watch([() => settingsStore.spellCheck, () => settingsStore.language], applyProofingPreferences)
+
 onBeforeUnmount(() => { void crepe?.destroy() })
 
 defineExpose({ getEditor: () => crepe?.editor })
@@ -181,42 +193,42 @@ defineExpose({ getEditor: () => crepe?.editor })
 
 <template>
   <div class="visual-editor">
-    <div class="markdown-toolbar" role="toolbar" aria-label="Markdown 格式工具栏">
+    <div class="markdown-toolbar" role="toolbar" :aria-label="t('Markdown 格式工具栏', 'Markdown formatting toolbar')">
-      <label class="toolbar-select heading-select" title="设置标题级别">
+      <label class="toolbar-select heading-select" :title="t('设置标题级别', 'Set heading level')">
         <span class="format-glyph heading-glyph">H</span>
-        <select aria-label="标题级别" @change="applyHeading">
+        <select :aria-label="t('标题级别', 'Heading level')" @change="applyHeading">
-          <option value="" selected>标题</option>
+          <option value="" selected>{{ t('标题', 'Heading') }}</option>
-          <option value="paragraph">正文</option>
+          <option value="paragraph">{{ t('正文', 'Paragraph') }}</option>
           <option v-for="level in 6" :key="level" :value="level">H{{ level }}</option>
         </select>
       </label>
-      <button type="button" title="加粗 (Ctrl+B)" aria-label="加粗" @pointerdown.prevent="runCommand('bold')"><strong class="format-glyph">B</strong></button>
+      <button type="button" :title="t('加粗 (Ctrl+B)', 'Bold (Ctrl+B)')" :aria-label="t('加粗', 'Bold')" @pointerdown.prevent="runCommand('bold')"><strong class="format-glyph">B</strong></button>
-      <button type="button" title="斜体 (Ctrl+I)" aria-label="斜体" @pointerdown.prevent="runCommand('italic')"><em class="format-glyph">I</em></button>
+      <button type="button" :title="t('斜体 (Ctrl+I)', 'Italic (Ctrl+I)')" :aria-label="t('斜体', 'Italic')" @pointerdown.prevent="runCommand('italic')"><em class="format-glyph">I</em></button>
       <span class="toolbar-divider" />
-      <button type="button" class="list-glyph" title="有序列表" aria-label="有序列表" @pointerdown.prevent="runCommand('ordered-list')"><span class="list-marker">1</span><span class="list-lines">☰</span></button>
+      <button type="button" class="list-glyph" :title="t('有序列表', 'Ordered list')" :aria-label="t('有序列表', 'Ordered list')" @pointerdown.prevent="runCommand('ordered-list')"><span class="list-marker">1</span><span class="list-lines">☰</span></button>
-      <button type="button" class="list-glyph" title="无序列表" aria-label="无序列表" @pointerdown.prevent="runCommand('bullet-list')"><span class="list-marker">•</span><span class="list-lines">☰</span></button>
+      <button type="button" class="list-glyph" :title="t('无序列表', 'Bullet list')" :aria-label="t('无序列表', 'Bullet list')" @pointerdown.prevent="runCommand('bullet-list')"><span class="list-marker">•</span><span class="list-lines">☰</span></button>
       <span class="toolbar-divider" />
-      <label class="toolbar-select font-size-select" title="选择预设字号">
+      <label class="toolbar-select font-size-select" :title="t('选择预设字号', 'Choose a preset font size')">
         <span class="format-glyph font-size-glyph">A</span>
-        <select aria-label="文字字号" @change="applyFontSize">
+        <select :aria-label="t('文字字号', 'Font size')" @change="applyFontSize">
-          <option value="" selected>字号</option>
+          <option value="" selected>{{ t('字号', 'Size') }}</option>
           <option v-for="size in [12, 14, 16, 18, 20, 24, 28, 32]" :key="size" :value="size">{{ size }} px</option>
         </select>
       </label>
-      <div class="font-size-input" title="输入字号后按 Enter 或点击应用">
+      <div class="font-size-input" :title="t('输入字号后按 Enter 或点击应用', 'Enter a font size, then press Enter or Apply')">
-        <input v-model.number="fontSizeInput" type="number" min="8" max="96" step="1" aria-label="自定义字号"
+        <input v-model.number="fontSizeInput" type="number" min="8" max="96" step="1" :aria-label="t('自定义字号', 'Custom font size')"
           @keydown.enter.prevent="applyFontSizeValue" />
         <span>px</span>
-        <button type="button" aria-label="应用自定义字号" @pointerdown.prevent="applyFontSizeValue">应用</button>
+        <button type="button" :aria-label="t('应用自定义字号', 'Apply custom font size')" @pointerdown.prevent="applyFontSizeValue">{{ t('应用', 'Apply') }}</button>
       </div>
       <span class="toolbar-divider" />
-      <button type="button" title="行内代码" aria-label="行内代码" @pointerdown.prevent="runCommand('inline-code')"><code class="code-glyph">&lt;/&gt;</code></button>
+      <button type="button" :title="t('行内代码', 'Inline code')" :aria-label="t('行内代码', 'Inline code')" @pointerdown.prevent="runCommand('inline-code')"><code class="code-glyph">&lt;/&gt;</code></button>
-      <button type="button" title="代码块" aria-label="代码块" @pointerdown.prevent="runCommand('code-block')"><span class="block-glyph">{ }</span></button>
+      <button type="button" :title="t('代码块', 'Code block')" :aria-label="t('代码块', 'Code block')" @pointerdown.prevent="runCommand('code-block')"><span class="block-glyph">{ }</span></button>
-      <button type="button" title="行内公式" aria-label="行内公式" @pointerdown.prevent="runCommand('inline-math')"><span class="math-glyph">ƒx</span></button>
+      <button type="button" :title="t('行内公式', 'Inline formula')" :aria-label="t('行内公式', 'Inline formula')" @pointerdown.prevent="runCommand('inline-math')"><span class="math-glyph">ƒx</span></button>
-      <button type="button" title="公式块" aria-label="公式块" @pointerdown.prevent="runCommand('math-block')"><span class="math-glyph">∑</span></button>
+      <button type="button" :title="t('公式块', 'Formula block')" :aria-label="t('公式块', 'Formula block')" @pointerdown.prevent="runCommand('math-block')"><span class="math-glyph">∑</span></button>
-      <button type="button" title="插入链接" aria-label="插入链接" @pointerdown.prevent="applyLink"><AppIcon :icon="Link" :size="17" /></button>
+      <button type="button" :title="t('插入链接', 'Insert link')" :aria-label="t('插入链接', 'Insert link')" @pointerdown.prevent="applyLink"><AppIcon :icon="Link" :size="17" /></button>
     </div>
-    <div v-if="loading" class="editor-loading">正在加载编辑器…</div>
+    <div v-if="loading" class="editor-loading">{{ t('正在加载编辑器…', 'Loading editor…') }}</div>
     <div ref="editorRoot" class="milkdown-host" :class="{ loading }" />
   </div>
 </template>

frontend/src/features/mcp/McpServersView.vue

@@ -5,6 +5,7 @@ import AppIcon from '@/components/common/AppIcon.vue'
 import type { McpServer, McpServerInput, McpServerTransport } from '@/contracts'
 import * as service from '@/services/mcpServerService'
 import { emptyMcpConfig, mergeImportedSecrets, normalizeMcpConfig, parseMcpJson, type ImportedSecret, type SecretKind } from './configuration'
+import { t } from '@/i18n'
 
 const servers = ref<McpServer[]>([])
 const busy = ref('')
@@ -24,12 +25,12 @@ const secretDrafts = reactive<Record<string, string>>({})
 const form = reactive<McpServerInput>(emptyMcpConfig())
 const importedSecrets = ref<ImportedSecret[]>([])
 
-const dialogTitle = computed(() => editingId.value ? '编辑 MCP 服务器' : '新增 MCP 服务器')
+const dialogTitle = computed(() => editingId.value ? t('编辑 MCP 服务器', 'Edit MCP Server') : t('新增 MCP 服务器', 'Add MCP Server'))
 
 async function load() {
   error.value = ''
   try { servers.value = await service.listMcpServers() }
-  catch (cause) { error.value = message(cause, '读取 MCP 服务器失败') }
+  catch (cause) { error.value = message(cause, t('读取 MCP 服务器失败', 'Failed to load MCP servers')) }
 }
 
 function resetEditor(input: McpServerInput) {
@@ -84,8 +85,8 @@ function applyTemplate(transport: McpServerTransport) {
 
 function parseObject(value: string, label: string): Record<string, string> {
   let parsed: unknown
-  try { parsed = JSON.parse(value || '{}') } catch { throw new Error(`${label}必须是 JSON 对象`) }
+  try { parsed = JSON.parse(value || '{}') } catch { throw new Error(`${label}${t('必须是 JSON 对象', ' must be a JSON object')}`) }
-  if (!parsed || Array.isArray(parsed) || typeof parsed !== 'object' || Object.values(parsed).some(item => typeof item !== 'string')) throw new Error(`${label}必须是字符串键值 JSON 对象`)
+  if (!parsed || Array.isArray(parsed) || typeof parsed !== 'object' || Object.values(parsed).some(item => typeof item !== 'string')) throw new Error(`${label}${t('必须是字符串键值 JSON 对象', ' must be a JSON object with string keys and values')}`)
   return parsed as Record<string, string>
 }
 
@@ -97,8 +98,8 @@ function formPayload(): McpServerInput {
     command: stdio ? form.command?.trim() : null,
     args: stdio ? argsText.value.split('\n').map(value => value.trim()).filter(Boolean) : [],
     url: stdio ? null : form.url?.trim(),
-    headers: stdio ? {} : parseObject(headersText.value, '普通 Header'),
+    headers: stdio ? {} : parseObject(headersText.value, t('普通 Header', 'Headers')),
-    environment: stdio ? parseObject(environmentText.value, '普通环境变量') : {},
+    environment: stdio ? parseObject(environmentText.value, t('普通环境变量', 'Environment variables')) : {},
     secret_environment_keys: stdio ? splitKeys(secretKeysText.value) : [],
     secret_header_keys: stdio ? [] : splitKeys(secretHeaderKeysText.value),
     permissions: permissionsText.value.split(',').map(value => value.trim()).filter(Boolean),
@@ -131,7 +132,7 @@ function switchMode(mode: 'form' | 'json') {
     if (mode === 'json') rawConfig.value = JSON.stringify(payload(false), null, 2)
     else resetEditor(payload(false))
     editorMode.value = mode
-  } catch (cause) { error.value = message(cause, '配置转换失败') }
+  } catch (cause) { error.value = message(cause, t('配置转换失败', 'Configuration conversion failed')) }
 }
 
 async function save() {
@@ -140,8 +141,8 @@ async function save() {
   try {
     error.value = ''
     const input = payload()
-    if (!input.name || (input.transport === 'stdio' ? !input.command : !input.url)) throw new Error('请填写服务器名称和连接地址')
+    if (!input.name || (input.transport === 'stdio' ? !input.command : !input.url)) throw new Error(t('请填写服务器名称和连接地址', 'Enter a server name and connection address'))
-    if (editingOriginal.value && executionChanged(editingOriginal.value, input) && !confirm('连接命令、地址或认证配置已变化，保存后旧测试与授权会失效。是否保存？')) return
+    if (editingOriginal.value && executionChanged(editingOriginal.value, input) && !confirm(t('连接命令、地址或认证配置已变化，保存后旧测试与授权会失效。是否保存？', 'The command, address, or authentication settings changed. Previous tests and authorization will be invalidated. Save?'))) return
     busy.value = 'save'
     saved = editingId.value ? await service.updateMcpServer(editingId.value, input) : await service.createMcpServer(input)
     // Commit the returned ID/version before saving secrets so a partial failure can
@@ -157,7 +158,7 @@ async function save() {
     await load()
   } catch (cause) {
     if (saved) await load()
-    error.value = `${saved ? '服务器配置已保存，但密钥保存失败；可点击保存重试。' : ''}${message(cause, '保存失败')}`
+    error.value = `${saved ? t('服务器配置已保存，但密钥保存失败；可点击保存重试。', 'Server settings were saved, but saving secrets failed. Save again to retry.') : ''}${message(cause, t('保存失败', 'Save failed'))}`
   }
   finally { busy.value = '' }
 }
@@ -189,8 +190,8 @@ function executionChanged(server: McpServer, input: McpServerInput) {
 
 async function approve(server: McpServer): Promise<McpServer | null> {
   if (server.trusted) return server
-  const localWarning = server.transport === 'stdio' ? '\n\n本机进程尚无系统级沙箱，仅应运行可信服务器。' : '\n\n连接可能向该地址发送配置的 Header。'
+  const localWarning = server.transport === 'stdio' ? t('\n\n本机进程尚无系统级沙箱，仅应运行可信服务器。', '\n\nLocal processes have no system-level sandbox. Run trusted servers only.') : t('\n\n连接可能向该地址发送配置的 Header。', '\n\nThe connection may send configured headers to this address.')
-  if (!confirm(`请确认 MCP 连接：\n\n${server.command_summary}${localWarning}\n\n是否继续？`)) return null
+  if (!confirm(`${t('请确认 MCP 连接：', 'Confirm MCP connection:')}\n\n${server.command_summary}${localWarning}\n\n${t('是否继续？', 'Continue?')}`)) return null
   return service.trustMcpServer(server)
 }
 
@@ -199,14 +200,14 @@ async function toggle(server: McpServer) { await act(server, 'toggle', current =
 async function act(server: McpServer, action: string, operation: (server: McpServer) => Promise<McpServer>) {
   busy.value = `${action}:${server.server_id}`; error.value = ''
   try { const current = action === 'toggle' && server.enabled ? server : await approve(server); if (!current) return; await operation(current); await load() }
-  catch (cause) { error.value = message(cause, '操作失败') }
+  catch (cause) { error.value = message(cause, t('操作失败', 'Operation failed')) }
   finally { busy.value = '' }
 }
 
 async function remove(server: McpServer) {
-  if (!confirm(`删除“${server.name}”及其加密凭据？`)) return
+  if (!confirm(t(`删除“${server.name}”及其加密凭据？`, `Delete “${server.name}” and its encrypted credentials?`))) return
   try { busy.value = `delete:${server.server_id}`; await service.deleteMcpServer(server.server_id); await load() }
-  catch (cause) { error.value = message(cause, '删除失败') } finally { busy.value = '' }
+  catch (cause) { error.value = message(cause, t('删除失败', 'Delete failed')) } finally { busy.value = '' }
 }
 
 async function saveSecret(server: McpServer, key: string, kind: SecretKind) {
@@ -214,7 +215,7 @@ async function saveSecret(server: McpServer, key: string, kind: SecretKind) {
   const value = secretDrafts[draftKey]?.trim()
   if (!value) return
   try { busy.value = `secret:${draftKey}`; await service.putMcpServerSecret(server.server_id, key, value, kind); secretDrafts[draftKey] = ''; await load() }
-  catch (cause) { error.value = message(cause, '保存密钥失败') } finally { busy.value = '' }
+  catch (cause) { error.value = message(cause, t('保存密钥失败', 'Failed to save secret')) } finally { busy.value = '' }
 }
 
 function splitKeys(value: string) { return value.split(/[\n,]/).map(item => item.trim()).filter(Boolean) }
@@ -224,20 +225,20 @@ onMounted(load)
 
 <template>
   <section class="feature-page mcp-page">
-    <header class="feature-header"><div><h1>MCP 服务器</h1><p>管理独立 MCP Server 的连接、凭据与工具生命周期。</p></div><div class="inline-actions"><button class="button-secondary" :disabled="!!busy" @click="load"><AppIcon :icon="Refresh" /> 刷新</button><button class="button-primary" @click="openCreate"><AppIcon :icon="Plus" /> 新增服务器</button></div></header>
+    <header class="feature-header"><div><h1>{{ t('MCP 服务器', 'MCP Servers') }}</h1><p>{{ t('管理独立 MCP Server 的连接、凭据与工具生命周期。', 'Manage standalone MCP server connections, credentials, and tool lifecycles.') }}</p></div><div class="inline-actions"><button class="button-secondary" :disabled="!!busy" @click="load"><AppIcon :icon="Refresh" /> {{ t('刷新', 'Refresh') }}</button><button class="button-primary" @click="openCreate"><AppIcon :icon="Plus" /> {{ t('新增服务器', 'Add server') }}</button></div></header>
-    <div class="notice-banner">stdio 本机进程仅在开发环境开放；Streamable HTTP 为首选远程传输，SSE 仅用于兼容旧服务器。uvx 隔离依赖但不是安全沙箱。</div>
+    <div class="notice-banner">{{ t('stdio 本机进程仅在开发环境开放；Streamable HTTP 为首选远程传输，SSE 仅用于兼容旧服务器。uvx 隔离依赖但不是安全沙箱。', 'Local stdio processes are available only in development. Streamable HTTP is the preferred remote transport; SSE supports legacy servers. uvx isolates dependencies but is not a security sandbox.') }}</div>
     <div v-if="error" class="error-banner">{{ error }}</div>
-    <div v-if="!servers.length" class="panel empty"><AppIcon :icon="Connection" :size="34" /><h2>尚未配置 MCP 服务器</h2><p>添加 Server，测试连接成功后才能启用工具。</p><button class="button-primary" @click="openCreate">新增服务器</button></div>
+    <div v-if="!servers.length" class="panel empty"><AppIcon :icon="Connection" :size="34" /><h2>{{ t('尚未配置 MCP 服务器', 'No MCP servers configured') }}</h2><p>{{ t('添加 Server，测试连接成功后才能启用工具。', 'Add a server and test its connection before enabling its tools.') }}</p><button class="button-primary" @click="openCreate">{{ t('新增服务器', 'Add server') }}</button></div>
     <div v-else class="server-list">
       <article v-for="server in servers" :key="server.server_id" class="panel server-card">
         <div class="server-main"><div class="server-title"><AppIcon :icon="Connection" :size="24" /><div><h2>{{ server.name }}</h2><code>{{ server.command_summary }}</code></div></div><span class="badge" :class="{ success: server.status === 'ready', error: ['error','unhealthy'].includes(server.status) }">{{ server.status }}</span></div>
-        <div class="metadata"><span>{{ server.transport }}</span><span>v{{ server.version }}</span><span>{{ server.tools_count }} 个工具</span><span>{{ server.trusted ? '连接已确认' : '等待确认连接' }}</span><span v-if="server.last_test_succeeded">当前配置测试成功</span><span v-if="server.remote_server_name">{{ server.remote_server_name }} {{ server.remote_server_version }}</span></div>
+        <div class="metadata"><span>{{ server.transport }}</span><span>v{{ server.version }}</span><span>{{ server.tools_count }} {{ t('个工具', 'tools') }}</span><span>{{ server.trusted ? t('连接已确认', 'Connection confirmed') : t('等待确认连接', 'Awaiting confirmation') }}</span><span v-if="server.last_test_succeeded">{{ t('当前配置测试成功', 'Current configuration passed') }}</span><span v-if="server.remote_server_name">{{ server.remote_server_name }} {{ server.remote_server_version }}</span></div>
         <div v-if="server.error" class="error-banner compact">{{ server.error }}</div>
         <div v-if="Object.keys(server.secret_environment).length || Object.keys(server.secret_headers).length" class="secrets">
-          <label v-for="(configured, key) in server.secret_environment" :key="`env:${key}`"><span>环境变量 · {{ key }} <small>{{ configured ? '已加密保存' : '未配置' }}</small></span><span class="secret-input"><input v-model="secretDrafts[`${server.server_id}:environment:${key}`]" type="password" autocomplete="new-password" placeholder="输入后保存（不会回显）"><button class="button-secondary" @click="saveSecret(server, key, 'environment')">保存</button></span></label>
+          <label v-for="(configured, key) in server.secret_environment" :key="`env:${key}`"><span>{{ t('环境变量', 'Environment variable') }} · {{ key }} <small>{{ configured ? t('已加密保存', 'Encrypted and saved') : t('未配置', 'Not configured') }}</small></span><span class="secret-input"><input v-model="secretDrafts[`${server.server_id}:environment:${key}`]" type="password" autocomplete="new-password" :placeholder="t('输入后保存（不会回显）', 'Enter and save (never displayed)')"><button class="button-secondary" @click="saveSecret(server, key, 'environment')">{{ t('保存', 'Save') }}</button></span></label>
-          <label v-for="(configured, key) in server.secret_headers" :key="`header:${key}`"><span>HTTP Header · {{ key }} <small>{{ configured ? '已加密保存' : '未配置' }}</small></span><span class="secret-input"><input v-model="secretDrafts[`${server.server_id}:header:${key}`]" type="password" autocomplete="new-password" placeholder="输入后保存（不会回显）"><button class="button-secondary" @click="saveSecret(server, key, 'header')">保存</button></span></label>
+          <label v-for="(configured, key) in server.secret_headers" :key="`header:${key}`"><span>HTTP Header · {{ key }} <small>{{ configured ? t('已加密保存', 'Encrypted and saved') : t('未配置', 'Not configured') }}</small></span><span class="secret-input"><input v-model="secretDrafts[`${server.server_id}:header:${key}`]" type="password" autocomplete="new-password" :placeholder="t('输入后保存（不会回显）', 'Enter and save (never displayed)')"><button class="button-secondary" @click="saveSecret(server, key, 'header')">{{ t('保存', 'Save') }}</button></span></label>
         </div>
-        <footer class="card-actions"><button class="button-secondary" :disabled="!!busy || server.enabled" @click="test(server)"><AppIcon :icon="VideoPlay" /> 测试连接</button><button class="button-secondary" :disabled="!!busy" @click="openEdit(server)"><AppIcon :icon="EditPen" /> 编辑</button><button class="button-danger" :disabled="!!busy" @click="remove(server)"><AppIcon :icon="Delete" /> 删除</button><button class="button-primary" :disabled="!!busy || (!server.enabled && !server.last_test_succeeded)" :title="!server.enabled && !server.last_test_succeeded ? '请先测试当前配置' : ''" @click="toggle(server)">{{ server.enabled ? '停用' : '启用' }}</button></footer>
+        <footer class="card-actions"><button class="button-secondary" :disabled="!!busy || server.enabled" @click="test(server)"><AppIcon :icon="VideoPlay" /> {{ t('测试连接', 'Test connection') }}</button><button class="button-secondary" :disabled="!!busy" @click="openEdit(server)"><AppIcon :icon="EditPen" /> {{ t('编辑', 'Edit') }}</button><button class="button-danger" :disabled="!!busy" @click="remove(server)"><AppIcon :icon="Delete" /> {{ t('删除', 'Delete') }}</button><button class="button-primary" :disabled="!!busy || (!server.enabled && !server.last_test_succeeded)" :title="!server.enabled && !server.last_test_succeeded ? t('请先测试当前配置', 'Test the current configuration first') : ''" @click="toggle(server)">{{ server.enabled ? t('停用', 'Disable') : t('启用', 'Enable') }}</button></footer>
       </article>
     </div>
 
@@ -246,18 +247,18 @@ onMounted(load)
         <fieldset :disabled="!!busy" class="editor-fields">
         <header><h2><AppIcon :icon="Plus" /> {{ dialogTitle }}</h2><button type="button" class="close" @click="closeEditor">×</button></header>
         <div v-if="error" class="error-banner" role="alert">{{ error }}</div>
-        <div v-if="importedSecrets.length" class="notice-banner">已识别 {{ importedSecrets.length }} 项密钥，保存时将单独加密，不会写入普通服务器配置；取消将清除未保存密钥。</div>
+        <div v-if="importedSecrets.length" class="notice-banner">{{ t('已识别', 'Detected') }} {{ importedSecrets.length }} {{ t('项密钥，保存时将单独加密，不会写入普通服务器配置；取消将清除未保存密钥。', 'secrets. They will be encrypted separately and excluded from regular server settings. Canceling clears unsaved secrets.') }}</div>
-        <div class="mode-tabs"><button type="button" :class="{ active: editorMode === 'form' }" @click="switchMode('form')">表单配置</button><button type="button" :class="{ active: editorMode === 'json' }" @click="switchMode('json')">JSON 配置</button></div>
+        <div class="mode-tabs"><button type="button" :class="{ active: editorMode === 'form' }" @click="switchMode('form')">{{ t('表单配置', 'Form') }}</button><button type="button" :class="{ active: editorMode === 'json' }" @click="switchMode('json')">{{ t('JSON 配置', 'JSON') }}</button></div>
         <template v-if="editorMode === 'form'">
-          <label>服务器名称<input v-model="form.name" maxlength="80" placeholder="例如：文件系统工具"></label>
+          <label>{{ t('服务器名称', 'Server name') }}<input v-model="form.name" maxlength="80" :placeholder="t('例如：文件系统工具', 'For example: Filesystem tools')"></label>
-          <div class="template-row"><span>服务器配置</span><button type="button" class="template" :class="{ active: form.transport === 'stdio' }" @click="applyTemplate('stdio')">stdio 模板</button><button type="button" class="template" :class="{ active: form.transport === 'streamable_http' }" @click="applyTemplate('streamable_http')">Streamable HTTP</button><button type="button" class="template" :class="{ active: form.transport === 'sse' }" @click="applyTemplate('sse')">SSE（兼容）</button></div>
+          <div class="template-row"><span>{{ t('服务器配置', 'Server configuration') }}</span><button type="button" class="template" :class="{ active: form.transport === 'stdio' }" @click="applyTemplate('stdio')">stdio {{ t('模板', 'template') }}</button><button type="button" class="template" :class="{ active: form.transport === 'streamable_http' }" @click="applyTemplate('streamable_http')">Streamable HTTP</button><button type="button" class="template" :class="{ active: form.transport === 'sse' }" @click="applyTemplate('sse')">SSE {{ t('（兼容）', '(legacy)') }}</button></div>
-          <template v-if="form.transport === 'stdio'"><label>可执行命令<input v-model="form.command" placeholder="uvx、npx 或可信可执行文件路径"></label><label>参数（每行一项）<textarea v-model="argsText" rows="5"></textarea></label><div class="two-columns"><label>普通环境变量（JSON）<textarea v-model="environmentText" rows="5"></textarea></label><label>敏感环境变量名（每行一项）<textarea v-model="secretKeysText" rows="5" placeholder="API_KEY"></textarea></label></div></template>
+          <template v-if="form.transport === 'stdio'"><label>{{ t('可执行命令', 'Executable command') }}<input v-model="form.command" :placeholder="t('uvx、npx 或可信可执行文件路径', 'uvx, npx, or a trusted executable path')"></label><label>{{ t('参数（每行一项）', 'Arguments (one per line)') }}<textarea v-model="argsText" rows="5"></textarea></label><div class="two-columns"><label>{{ t('普通环境变量（JSON）', 'Environment variables (JSON)') }}<textarea v-model="environmentText" rows="5"></textarea></label><label>{{ t('敏感环境变量名（每行一项）', 'Secret environment names (one per line)') }}<textarea v-model="secretKeysText" rows="5" placeholder="API_KEY"></textarea></label></div></template>
-          <template v-else><label>MCP URL<input v-model="form.url" placeholder="https://example.com/mcp"></label><div class="two-columns"><label>普通 Header（JSON）<textarea v-model="headersText" rows="5" placeholder='{"X-Client":"NotesAgent"}'></textarea></label><label>敏感 Header 名（每行一项）<textarea v-model="secretHeaderKeysText" rows="5" placeholder="Authorization"></textarea></label></div></template>
+          <template v-else><label>MCP URL<input v-model="form.url" placeholder="https://example.com/mcp"></label><div class="two-columns"><label>{{ t('普通 Header（JSON）', 'Headers (JSON)') }}<textarea v-model="headersText" rows="5" placeholder='{"X-Client":"NotesAgent"}'></textarea></label><label>{{ t('敏感 Header 名（每行一项）', 'Secret header names (one per line)') }}<textarea v-model="secretHeaderKeysText" rows="5" placeholder="Authorization"></textarea></label></div></template>
-          <label>声明权限（逗号分隔，可选）<input v-model="permissionsText" placeholder="network.request, notes.read"></label>
+          <label>{{ t('声明权限（逗号分隔，可选）', 'Declared permissions (comma-separated, optional)') }}<input v-model="permissionsText" placeholder="network.request, notes.read"></label>
-          <div class="two-columns"><label>启动超时（秒）<input v-model.number="form.startup_timeout_seconds" type="number" min="1" max="120"></label><label>工具超时（秒）<input v-model.number="form.tool_timeout_seconds" type="number" min="1" max="300"></label></div>
+          <div class="two-columns"><label>{{ t('启动超时（秒）', 'Startup timeout (seconds)') }}<input v-model.number="form.startup_timeout_seconds" type="number" min="1" max="120"></label><label>{{ t('工具超时（秒）', 'Tool timeout (seconds)') }}<input v-model.number="form.tool_timeout_seconds" type="number" min="1" max="300"></label></div>
         </template>
-        <label v-else>服务器 JSON 配置<textarea v-model="rawConfig" class="json-editor" rows="22" spellcheck="false"></textarea><small>支持 NotesAgent 配置、command/args/env 和单服务器 mcpServers 配置。已声明的 Secret 及常见 API Key、Token、Authorization 会拆分后加密保存。其他敏感值请显式声明；不要把密钥放入命令或参数。</small><small>兼容导入 timeout 为启动超时，sse_read_timeout 为工具等待上限（不保留 SSE 读取超时语义）。</small></label>
+        <label v-else>{{ t('服务器 JSON 配置', 'Server JSON configuration') }}<textarea v-model="rawConfig" class="json-editor" rows="22" spellcheck="false"></textarea><small>{{ t('支持 NotesAgent 配置、command/args/env 和单服务器 mcpServers 配置。已声明的 Secret 及常见 API Key、Token、Authorization 会拆分后加密保存。其他敏感值请显式声明；不要把密钥放入命令或参数。', 'Supports NotesAgent, command/args/env, and single-server mcpServers configurations. Declared secrets and common API key, token, and authorization values are separated and encrypted. Declare other sensitive values explicitly; never place secrets in commands or arguments.') }}</small><small>{{ t('兼容导入 timeout 为启动超时，sse_read_timeout 为工具等待上限（不保留 SSE 读取超时语义）。', 'For compatible imports, timeout maps to startup timeout and sse_read_timeout maps to the tool wait limit.') }}</small></label>
-        <footer><button type="button" class="button-secondary" @click="closeEditor">取消</button><button class="button-primary" :disabled="busy === 'save'">保存</button></footer>
+        <footer><button type="button" class="button-secondary" @click="closeEditor">{{ t('取消', 'Cancel') }}</button><button class="button-primary" :disabled="busy === 'save'">{{ t('保存', 'Save') }}</button></footer>
         </fieldset>
       </form>
     </div>

frontend/src/features/mcp/configuration.ts

@@ -1,4 +1,5 @@
 import type { McpServerInput } from '@/contracts'
+import { t } from '@/i18n'
 
 export type SecretKind = 'environment' | 'header'
 export interface ImportedSecret { kind: SecretKind; key: string; value: string }
@@ -26,38 +27,38 @@ export function emptyMcpConfig(): McpServerInput {
 }
 
 function object(value: unknown, label: string): Record<string, unknown> {
-  if (!value || Array.isArray(value) || typeof value !== 'object') throw new Error(`${label}必须是 JSON 对象`)
+  if (!value || Array.isArray(value) || typeof value !== 'object') throw new Error(`${label}${t('必须是 JSON 对象', ' must be a JSON object')}`)
   return value as Record<string, unknown>
 }
 
 function strings(value: unknown, label: string): string[] {
   if (value === undefined) return []
-  if (!Array.isArray(value) || value.some(item => typeof item !== 'string')) throw new Error(`${label}必须是字符串数组`)
+  if (!Array.isArray(value) || value.some(item => typeof item !== 'string')) throw new Error(`${label}${t('必须是字符串数组', ' must be a string array')}`)
   return [...value]
 }
 
 function entries(value: unknown, label: string): Record<string, string> {
   if (value === undefined) return {}
   const result = object(value, label)
-  if (Object.values(result).some(item => typeof item !== 'string')) throw new Error(`${label}必须是字符串键值 JSON 对象`)
+  if (Object.values(result).some(item => typeof item !== 'string')) throw new Error(`${label}${t('必须是字符串键值 JSON 对象', ' must be a JSON object with string keys and values')}`)
   return { ...result } as Record<string, string>
 }
 
 function timeout(value: unknown, fallback: number, max: number, label: string): number {
   if (value === undefined) return fallback
-  if (typeof value !== 'number' || !Number.isFinite(value) || value < 1 || value > max) throw new Error(`${label}必须是 1–${max} 秒之间的数字`)
+  if (typeof value !== 'number' || !Number.isFinite(value) || value < 1 || value > max) throw new Error(`${label}${t(`必须是 1–${max} 秒之间的数字`, ` must be a number from 1 to ${max} seconds`)}`)
   return value
 }
 
 // Do not silently rewrite executable arguments or secret values copied from chat.
 function checkUrl(value: string, label: string) {
-  if (/^\[https?:\/\//i.test(value)) throw new Error(`${label}请填写纯 URL，不要粘贴 Markdown 链接`)
+  if (/^\[https?:\/\//i.test(value)) throw new Error(`${label}${t('请填写纯 URL，不要粘贴 Markdown 链接', ': enter a plain URL instead of a Markdown link')}`)
 }
 
 export function parseMcpJson(raw: string, fallbackName = '', requireConnection = true) {
   let parsed: unknown
   try { parsed = JSON.parse(raw) }
-  catch { throw new Error('服务器配置不是有效 JSON；请检查逗号、引号和无效的 \\_ 转义') }
+  catch { throw new Error(t('服务器配置不是有效 JSON；请检查逗号、引号和无效的 \\_ 转义', 'The server configuration is not valid JSON. Check commas, quotes, and invalid \\_ escapes.')) }
   return normalizeMcpConfig(parsed, fallbackName, requireConnection)
 }
 
@@ -65,54 +66,54 @@ export function parseMcpJson(raw: string, fallbackName = '', requireConnection =
  * Inline secrets leave the public config here and are sent only to the Secret API.
  */
 export function normalizeMcpConfig(parsed: unknown, fallbackName = '', requireConnection = true) {
-  let raw = object(parsed, '服务器配置')
+  let raw = object(parsed, t('服务器配置', 'Server configuration'))
   if ('mcpServers' in raw) {
     const servers = Object.entries(object(raw.mcpServers, 'mcpServers'))
-    if (servers.length !== 1) throw new Error('请一次导入一个 MCP 服务器')
+    if (servers.length !== 1) throw new Error(t('请一次导入一个 MCP 服务器', 'Import one MCP server at a time'))
     fallbackName = servers[0]![0]
-    raw = object(servers[0]![1], '服务器配置')
+    raw = object(servers[0]![1], t('服务器配置', 'Server configuration'))
   }
   const allowed = new Set([...Object.keys(emptyMcpConfig()), 'version', 'env', 'type', 'timeout', 'sse_read_timeout'])
   if (Object.keys(raw).some(key => !allowed.has(key))) {
     // Never echo arbitrary unknown keys: pasted secrets sometimes become JSON keys.
-    throw new Error('服务器配置含不支持的字段；API Key 请放在 env/environment 的对应变量中，不要放在顶层')
+    throw new Error(t('服务器配置含不支持的字段；API Key 请放在 env/environment 的对应变量中，不要放在顶层', 'The server configuration contains unsupported fields. Put API keys in the corresponding env/environment variables, not at the top level.'))
   }
-  if (raw.env !== undefined && raw.environment !== undefined) throw new Error('env 与 environment 请只保留一个，避免覆盖配置')
+  if (raw.env !== undefined && raw.environment !== undefined) throw new Error(t('env 与 environment 请只保留一个，避免覆盖配置', 'Keep either env or environment, not both'))
   const transport = raw.transport ?? raw.type ?? (raw.url ? 'streamable_http' : 'stdio')
-  if (!['stdio', 'streamable_http', 'sse'].includes(transport as string)) throw new Error('transport 必须是 stdio、streamable_http 或 sse')
+  if (!['stdio', 'streamable_http', 'sse'].includes(transport as string)) throw new Error(t('transport 必须是 stdio、streamable_http 或 sse', 'transport must be stdio, streamable_http, or sse'))
   const config = emptyMcpConfig()
   config.transport = transport as McpServerInput['transport']
-  const name = raw.name ?? (fallbackName || (typeof raw.command === 'string' ? raw.command : 'MCP 服务器'))
+  const name = raw.name ?? (fallbackName || (typeof raw.command === 'string' ? raw.command : t('MCP 服务器', 'MCP Server')))
-  if (typeof name !== 'string' || (requireConnection && !name.trim()) || name.trim().length > 80) throw new Error('服务器名称必须为 1–80 个字符')
+  if (typeof name !== 'string' || (requireConnection && !name.trim()) || name.trim().length > 80) throw new Error(t('服务器名称必须为 1–80 个字符', 'The server name must contain 1–80 characters'))
   config.name = name.trim()
   for (const key of ['command', 'url'] as const) {
     const value = raw[key]
-    if (value !== undefined && value !== null && typeof value !== 'string') throw new Error(`${key}必须是字符串`)
+    if (value !== undefined && value !== null && typeof value !== 'string') throw new Error(`${key}${t('必须是字符串', ' must be a string')}`)
     config[key] = typeof value === 'string' ? value.trim() : null
   }
   config.args = strings(raw.args, 'args')
-  if (config.args.length > 64) throw new Error('args 最多允许 64 项')
+  if (config.args.length > 64) throw new Error(t('args 最多允许 64 项', 'args allows at most 64 items'))
-  for (const value of config.args) checkUrl(value, 'args 中的地址')
+  for (const value of config.args) checkUrl(value, t('args 中的地址', 'URL in args'))
   config.environment = entries(raw.environment ?? raw.env, 'environment/env')
   config.headers = entries(raw.headers, 'headers')
   config.secret_environment_keys = [...new Set(strings(raw.secret_environment_keys, 'secret_environment_keys'))]
   config.secret_header_keys = [...new Set(strings(raw.secret_header_keys, 'secret_header_keys'))]
   config.permissions = strings(raw.permissions, 'permissions')
-  config.startup_timeout_seconds = timeout(raw.startup_timeout_seconds ?? raw.timeout, 15, 120, '启动超时')
+  config.startup_timeout_seconds = timeout(raw.startup_timeout_seconds ?? raw.timeout, 15, 120, t('启动超时', 'Startup timeout'))
   // Compatibility policy: legacy read timeout becomes the tool wait budget, not an SSE transport setting.
-  config.tool_timeout_seconds = timeout(raw.tool_timeout_seconds ?? raw.sse_read_timeout, 30, 300, '工具超时')
+  config.tool_timeout_seconds = timeout(raw.tool_timeout_seconds ?? raw.sse_read_timeout, 30, 300, t('工具超时', 'Tool timeout'))
   if (config.transport === 'stdio') {
-    if (requireConnection && !config.command) throw new Error('stdio 配置必须填写 command')
+    if (requireConnection && !config.command) throw new Error(t('stdio 配置必须填写 command', 'stdio configuration requires command'))
-    if (config.url || Object.keys(config.headers).length || config.secret_header_keys.length) throw new Error('stdio 配置不能包含 URL 或 HTTP Header')
+    if (config.url || Object.keys(config.headers).length || config.secret_header_keys.length) throw new Error(t('stdio 配置不能包含 URL 或 HTTP Header', 'stdio configuration cannot contain a URL or HTTP headers'))
   } else {
-    if (requireConnection && !config.url) throw new Error('HTTP/SSE 配置必须填写 url')
+    if (requireConnection && !config.url) throw new Error(t('HTTP/SSE 配置必须填写 url', 'HTTP/SSE configuration requires a URL'))
     if (config.url) {
       checkUrl(config.url, 'url')
       let url: URL
-      try { url = new URL(config.url) } catch { throw new Error('url 必须是有效的 HTTP(S) 地址') }
+      try { url = new URL(config.url) } catch { throw new Error(t('url 必须是有效的 HTTP(S) 地址', 'url must be a valid HTTP(S) address')) }
-      if (!['http:', 'https:'].includes(url.protocol) || url.username || url.password || url.hash) throw new Error('url 必须为不含账号密码或片段的 HTTP(S) 地址')
+      if (!['http:', 'https:'].includes(url.protocol) || url.username || url.password || url.hash) throw new Error(t('url 必须为不含账号密码或片段的 HTTP(S) 地址', 'url must be an HTTP(S) address without credentials or a fragment'))
     }
-    if (config.command || config.args.length || Object.keys(config.environment).length || config.secret_environment_keys.length) throw new Error('HTTP/SSE 配置不能包含 command、args 或环境变量')
+    if (config.command || config.args.length || Object.keys(config.environment).length || config.secret_environment_keys.length) throw new Error(t('HTTP/SSE 配置不能包含 command、args 或环境变量', 'HTTP/SSE configuration cannot contain command, args, or environment variables'))
   }
   const secrets: ImportedSecret[] = []
   for (const kind of ['environment', 'header'] as const) {
@@ -120,19 +121,19 @@ export function normalizeMcpConfig(parsed: unknown, fallbackName = '', requireCo
     const keys = kind === 'environment' ? config.secret_environment_keys : config.secret_header_keys
     const identity = (key: string) => kind === 'header' ? key.toLowerCase() : key
     const allKeys = [...Object.keys(values), ...keys]
-    if (kind === 'header' && (new Set(keys.map(identity)).size !== keys.length || new Set(Object.keys(values).map(identity)).size !== Object.keys(values).length)) throw new Error('HTTP Header 名称不能仅大小写不同而重复声明')
+    if (kind === 'header' && (new Set(keys.map(identity)).size !== keys.length || new Set(Object.keys(values).map(identity)).size !== Object.keys(values).length)) throw new Error(t('HTTP Header 名称不能仅大小写不同而重复声明', 'HTTP header names cannot be duplicated with case-only differences'))
     const validKey = kind === 'environment' ? /^[A-Za-z_][A-Za-z0-9_]{0,127}$/ : /^[!#$%&'*+.^_`|~0-9A-Za-z-]{1,128}$/
-    if (allKeys.some(key => !validKey.test(key))) throw new Error(`${kind === 'environment' ? '环境变量' : 'Header'}名称无效；敏感变量名只能填名称，不能填密钥值`)
+    if (allKeys.some(key => !validKey.test(key))) throw new Error(`${kind === 'environment' ? t('环境变量', 'Environment variable') : 'Header'}${t('名称无效；敏感变量名只能填名称，不能填密钥值', ' name is invalid; secret variable declarations accept names only, not secret values')}`)
     for (const [key, value] of Object.entries(values)) {
       const declared = keys.find(item => identity(item) === identity(key))
       const sensitive = /api[_-]?key|token|secret|password|authorization|cookie|credential/i.test(key)
       if (declared || sensitive) {
-        if (!value || value.length > 32768) throw new Error('密钥值必须为 1–32768 个字符')
+        if (!value || value.length > 32768) throw new Error(t('密钥值必须为 1–32768 个字符', 'Secret values must contain 1–32768 characters'))
         const secretKey = declared ?? key
         if (!declared) keys.push(key)
         secrets.push({ kind, key: secretKey, value })
         delete values[key]
-      } else if (/host|url|endpoint/i.test(key)) checkUrl(value, '环境变量或 Header 地址')
+      } else if (/host|url|endpoint/i.test(key)) checkUrl(value, t('环境变量或 Header 地址', 'Environment variable or Header URL'))
     }
   }
   return { config, secrets }

frontend/src/features/media/MediaView.vue

@@ -1,14 +1,19 @@
 <script setup lang="ts">
 import { computed, onMounted, onUnmounted, ref } from 'vue'
 import { useRoute } from 'vue-router'
-import { mediaService, type MediaJob } from '@/services/mediaService'
+import { mediaService, createMediaSubmission, type MediaJob } from '@/services/mediaService'
+import { localeTag, t } from '@/i18n'
+import FilePicker from '@/components/common/FilePicker.vue'
 
 const route = useRoute()
+const submission = createMediaSubmission()
+const updateExisting = ref(false)
 const jobs = ref<MediaJob[]>([])
 const selected = ref<MediaJob | null>(null)
 const file = ref<File | null>(null)
 const reference = ref<File | null>(null)
 const matchResult = ref('')
+const terminologyPlaceholder = computed(() => t('{"错误术语": "正确术语"}', '{"incorrect term": "correct term"}'))
 const localOnly = ref(false)
 const diarization = ref(true)
 const terminology = ref('')
@@ -16,17 +21,22 @@ const busy = ref(false)
 const error = ref('')
 const notice = ref('')
 const dirty = ref(false)
-const title = ref('课堂转写')
+const title = ref(t('课堂转写', 'Class transcript'))
 const player = ref<HTMLAudioElement | null>(null)
 const position = ref(0)
 const speed = ref(1)
 const history = ref<MediaJob[]>([])
 let timer: ReturnType<typeof setTimeout> | undefined
 let stopped = false
-const labels = {queued: '排队中', running: '转写中', processing: '处理中', completed: '已完成', failed: '失败', cancelled: '已取消'}
+const labels = computed(() => ({queued: t('排队中', 'Queued'), running: t('转写中', 'Transcribing'), processing: t('处理中', 'Processing'), completed: t('已完成', 'Completed'), failed: t('失败', 'Failed'), cancelled: t('已取消', 'Cancelled')}))
 const speakers = computed(() => [...new Set(selected.value?.segments.map(s => s.speaker).filter((s): s is string => !!s) || [])])
 const active = (job: MediaJob) => ['queued', 'running', 'processing'].includes(job.status)
 const stamp = (seconds: number) => `${Math.floor(seconds / 60).toString().padStart(2, '0')}:${Math.floor(seconds % 60).toString().padStart(2, '0')}`
+const warningLabel = (warning: string) => ({
+  DIARIZATION_UNAVAILABLE: t('当前无法分离说话人', 'Speaker identification is unavailable'),
+  WORD_TIMESTAMPS_UNAVAILABLE: t('未提供逐字时间戳', 'Word-level timestamps are unavailable'),
+  DIARIZATION_SEGMENT_LEVEL: t('说话人按音频段估计，同段多人或重叠发言需人工校对', 'Speakers are estimated per segment; multiple or overlapping speakers require manual correction'),
+} as Record<string, string>)[warning] || warning
 
 async function refresh() {
   try {
@@ -36,10 +46,11 @@ async function refresh() {
   if (!stopped) timer = setTimeout(refresh, 2000)
 }
 async function choose(job: MediaJob) {
-  if (dirty.value && !window.confirm('当前校对尚未保存，切换后放弃修改？')) return
+  if (dirty.value && !window.confirm(t('当前校对尚未保存，切换后放弃修改？', 'The current corrections are unsaved. Discard them and switch?'))) return
   selected.value = JSON.parse(JSON.stringify(job)); dirty.value = false; history.value = []
 }
 async function action(work: () => Promise<void>) {
+  if (busy.value) return
   busy.value = true; error.value = ''; notice.value = ''
   try { await work() } catch (e) { error.value = (e as Error).message } finally { busy.value = false }
 }
@@ -49,13 +60,12 @@ async function submit() {
     let terms = {}
     if (terminology.value.trim()) {
       terms = JSON.parse(terminology.value)
-      if (!terms || typeof terms !== 'object' || Array.isArray(terms) || Object.values(terms).some(v => typeof v !== 'string')) throw new Error('术语表需要 JSON 对象，值为替换后的文本。')
+      if (!terms || typeof terms !== 'object' || Array.isArray(terms) || Object.values(terms).some(v => typeof v !== 'string')) throw new Error(t('术语表需要 JSON 对象，值为替换后的文本。', 'The terminology map must be a JSON object whose values are replacement text.'))
     }
-    const uploaded = await mediaService.upload(file.value!)
+    selected.value = await submission.submit(file.value!, {local_only: localOnly.value,
-    selected.value = await mediaService.create({attachment_id: uploaded.attachment_id, local_only: localOnly.value,
+      diarization: diarization.value, terminology: terms})
-      diarization: diarization.value, idempotency_key: crypto.randomUUID(), terminology: terms})
     dirty.value = false
-    jobs.value.unshift(selected.value)
+    jobs.value = [selected.value, ...jobs.value.filter(job => job.job_id !== selected.value?.job_id)]
   })
 }
 function seek(seconds: number) { if (player.value) { player.value.currentTime = seconds; position.value = seconds } }
@@ -63,10 +73,10 @@ async function purge() {
   if (!selected.value) return
   await action(async () => {
     const impact = await mediaService.impact(selected.value!.attachment_id)
-    if (!window.confirm(`${impact.message}\n将保留 ${impact.retained_note_ids.length} 篇已保存笔记。确定清理？`)) return
+    if (!window.confirm(`${impact.message}\n${t('将保留', 'Will retain')} ${impact.retained_note_ids.length} ${t('篇已保存笔记。确定清理？', 'saved notes. Continue cleanup?')}`)) return
     await mediaService.purge(selected.value!.attachment_id)
     selected.value = await mediaService.get(selected.value!.job_id)
-    dirty.value = false; history.value = []; notice.value = '附件与转写内容已清理'
+    dirty.value = false; history.value = []; notice.value = t('附件与转写内容已清理', 'Attachment and transcript content were removed')
   })
 }
 async function compareSpeaker() {
@@ -77,10 +87,10 @@ async function compareSpeaker() {
       const sample = await mediaService.upload(file.value!); temporary.push(sample.attachment_id)
       const known = await mediaService.upload(reference.value!); temporary.push(known.attachment_id)
       const result = await mediaService.match(sample.attachment_id, known.attachment_id, localOnly.value)
-      matchResult.value = `相似度 ${result.score.toFixed(3)} · ${result.source === 'local' ? '本地模型' : 'API'}${result.fallback_reason ? ` · 回退：${result.fallback_reason}` : ''}`
+      matchResult.value = `${t('相似度', 'Similarity')} ${result.score.toFixed(3)} · ${result.source === 'local' ? t('本地模型', 'Local model') : 'API'}${result.fallback_reason ? ` · ${t('回退：', 'Fallback: ')}${result.fallback_reason}` : ''}`
     } finally {
       const cleanup = await Promise.allSettled(temporary.map(id => mediaService.purge(id)))
-      if (cleanup.some(result => result.status === 'rejected')) notice.value = '部分临时参考附件清理失败，请检查后端连接。'
+      if (cleanup.some(result => result.status === 'rejected')) notice.value = t('部分临时参考附件清理失败，请检查后端连接。', 'Some temporary reference files could not be removed. Check the backend connection.')
     }
   })
 }
@@ -96,56 +106,56 @@ onUnmounted(() => { stopped = true; clearTimeout(timer) })
 
 <template>
   <section class="media-page">
-    <header><h1>音视频转写</h1><p class="subtle">上传音频或视频音轨，转写、校对后保存到知识库。单个文件最多 25 MiB。</p></header>
+    <header><h1>{{ t('音视频转写', 'Media Transcription') }}</h1><p class="subtle">{{ t('上传音频或视频音轨，转写、校对后保存到知识库。单个文件最多 25 MiB。', 'Upload audio or a video soundtrack, transcribe and correct it, then save it to the knowledge base. Maximum file size: 25 MiB.') }}</p></header>
     <div v-if="error" class="error-banner" role="alert">{{ error }}</div><p v-if="notice" role="status">{{ notice }}</p>
     <form class="panel upload" @submit.prevent="submit">
-      <label>选择附件<input type="file" accept=".wav,.mp3,.flac,.ogg,.m4a,.mp4,.webm,.txt,.md" @change="file = ($event.target as HTMLInputElement).files?.[0] || null" /></label>
+      <FilePicker :file="file" :label="t('选择附件', 'Choose attachment')" :empty-label="t('尚未选择文件', 'No file selected')" accept=".wav,.mp3,.flac,.ogg,.m4a,.mp4,.webm,.txt,.md" @select="file = $event" />
-      <label><input v-model="localOnly" type="checkbox" />仅本地处理</label>
+      <div class="upload-options"><label><input v-model="localOnly" type="checkbox" />{{ t('仅本地处理', 'Process locally only') }}</label>
-      <label><input v-model="diarization" type="checkbox" />识别不同说话人</label>
+      <label><input v-model="diarization" type="checkbox" />{{ t('识别不同说话人', 'Identify different speakers') }}</label></div>
-      <p class="subtle">{{ localOnly ? '本次任务不调用远程模型 API，模型需预先下载。' : '若配置了转写 API，将上传所选附件；API 失败后回退到本地模型。' }}</p>
+      <p class="subtle">{{ localOnly ? t('本次任务不调用远程模型 API，模型需预先下载。', 'This job will not call a remote model API; models must already be downloaded.') : t('若配置了转写 API，将上传所选附件；API 失败后回退到本地模型。', 'When a transcription API is configured, the selected file is uploaded; failures fall back to the local model.') }}</p>
-      <details><summary>术语校对</summary><p class="subtle">在识别完成后替换文本，原始识别结果会保留。</p><textarea v-model="terminology" class="input" rows="3" placeholder='{"错误术语": "正确术语"}' /></details>
+      <details class="ui-disclosure"><summary>{{ t('术语校对', 'Terminology corrections') }}</summary><p class="subtle">{{ t('在识别完成后替换文本，原始识别结果会保留。', 'Replace text after recognition while retaining the original result.') }}</p><textarea v-model="terminology" class="input" rows="3" :placeholder="terminologyPlaceholder" /></details>
-      <button class="button-primary" :disabled="busy || !file">{{ busy ? '处理中…' : '上传并转写' }}</button>
+      <div class="inline-actions upload-actions"><button type="button" class="button-secondary" :disabled="busy" @click="submission.reset(); notice = t('下一次提交将作为新任务处理', 'The next submission will be processed as a new job')">{{ t('重新处理为新任务', 'Process as new job') }}</button><button class="button-primary" :disabled="busy || !file">{{ busy ? t('处理中…', 'Processing…') : t('上传并转写', 'Upload and transcribe') }}</button></div>
-      <details><summary>声纹参考比对</summary><p class="subtle">将所选附件与参考音频比对。至少各含 1 秒语音；分数是相似度，不是身份认证概率。临时参考文件在比对后清理。</p>
+      <details class="ui-disclosure"><summary>{{ t('声纹参考比对', 'Speaker reference comparison') }}</summary><p class="subtle">{{ t('将所选附件与参考音频比对。至少各含 1 秒语音；分数是相似度，不是身份认证概率。临时参考文件在比对后清理。', 'Compare the selected file with reference audio. Each must contain at least one second of speech. The score is similarity, not an identity probability. Temporary files are removed afterward.') }}</p>
-        <input type="file" accept=".wav,.mp3,.flac,.ogg,.m4a" aria-label="声纹参考音频" @change="reference = ($event.target as HTMLInputElement).files?.[0] || null" />
+        <FilePicker :file="reference" :label="t('选择参考音频', 'Choose reference audio')" :empty-label="t('尚未选择参考音频', 'No reference audio selected')" accept=".wav,.mp3,.flac,.ogg,.m4a" @select="reference = $event" />
-        <button type="button" class="button-secondary" :disabled="busy || !file || !reference" @click="compareSpeaker">比对声纹</button><p v-if="matchResult">{{ matchResult }}</p></details>
+        <button type="button" class="button-secondary" :disabled="busy || !file || !reference" @click="compareSpeaker">{{ t('比对声纹', 'Compare speakers') }}</button><p v-if="matchResult">{{ matchResult }}</p></details>
     </form>
     <div class="media-columns">
-      <aside class="panel"><h2>转写任务</h2><p v-if="!jobs.length" class="subtle">暂无转写任务</p>
+      <aside class="panel"><h2>{{ t('转写任务', 'Transcription Jobs') }}</h2><p v-if="!jobs.length" class="subtle">{{ t('暂无转写任务', 'No transcription jobs') }}</p>
         <button v-for="job in jobs" :key="job.job_id" class="job-row" :class="{ selected: selected?.job_id === job.job_id }" @click="choose(job)">
-          <strong>{{ labels[job.status] }}</strong><span>{{ new Date(job.created_at).toLocaleString() }}</span><small>{{ job.attachment_id }}</small>
+          <strong>{{ labels[job.status] }}</strong><span>{{ new Date(job.created_at).toLocaleString(localeTag()) }}</span><small>{{ job.attachment_id }}</small>
         </button>
       </aside>
       <article v-if="selected" class="panel transcript">
-        <header><h2>{{ labels[selected.status] }}</h2><span class="badge">修订 {{ selected.revision }}</span></header>
+        <header><h2>{{ labels[selected.status] }}</h2><span class="badge">{{ t('修订', 'Revision') }} {{ selected.revision }}</span></header>
-        <progress v-if="active(selected) && selected.progress !== null" :value="selected.progress" :max="1" aria-label="转写进度" />
+        <progress v-if="active(selected) && selected.progress !== null" :value="selected.progress" :max="1" :aria-label="t('转写进度', 'Transcription progress')" />
         <audio ref="player" controls :src="mediaService.audio(selected.attachment_id)" @loadedmetadata="loaded" @timeupdate="position = player?.currentTime || 0" />
-        <label>播放速度<select v-model.number="speed" class="select" @change="player && (player.playbackRate = speed)"><option v-for="value in [0.5, 0.75, 1, 1.25, 1.5, 2]" :key="value" :value="value">{{ value }}×</option></select></label>
+        <label>{{ t('播放速度', 'Playback speed') }}<select v-model.number="speed" class="select" @change="player && (player.playbackRate = speed)"><option v-for="value in [0.5, 0.75, 1, 1.25, 1.5, 2]" :key="value" :value="value">{{ value }}×</option></select></label>
         <p v-if="selected.error_message" class="error-banner">{{ selected.error_message }} · {{ selected.error_code }}</p>
-        <p v-if="selected.fallback_reason" class="subtle">已回退：{{ selected.fallback_reason }}</p>
+        <p v-if="selected.fallback_reason" class="subtle">{{ t('已回退：', 'Fallback: ') }}{{ selected.fallback_reason }}</p>
-        <p v-for="warning in selected.warnings" :key="warning" class="subtle">{{ ({DIARIZATION_UNAVAILABLE: '当前无法分离说话人', WORD_TIMESTAMPS_UNAVAILABLE: '未提供逐字时间戳', DIARIZATION_SEGMENT_LEVEL: '说话人按音频段估计，同段多人或重叠发言需人工校对'} as Record<string,string>)[warning] || warning }}</p>
+        <p v-for="warning in selected.warnings" :key="warning" class="subtle">{{ warningLabel(warning) }}</p>
-        <button v-if="active(selected)" class="button-secondary" :disabled="busy" @click="action(async () => { selected = await mediaService.cancel(selected!.job_id) })">取消任务</button>
+        <button v-if="active(selected)" class="button-secondary" :disabled="busy" @click="action(async () => { selected = await mediaService.cancel(selected!.job_id) })">{{ t('取消任务', 'Cancel job') }}</button>
-        <button v-if="['failed', 'cancelled'].includes(selected.status) && selected.error_code !== 'MEDIA_PURGED'" class="button-secondary" :disabled="busy" @click="action(async () => { selected = await mediaService.retry(selected!.job_id) })">重新处理</button>
+        <button v-if="['failed', 'cancelled'].includes(selected.status) && selected.error_code !== 'MEDIA_PURGED'" class="button-secondary" :disabled="busy" @click="action(async () => { selected = await mediaService.retry(selected!.job_id) })">{{ t('重新处理', 'Process again') }}</button>
-        <button v-if="!active(selected) && selected.error_code !== 'MEDIA_PURGED'" class="button-danger" :disabled="busy" @click="purge">清理原附件与转写</button>
+        <button v-if="!active(selected) && selected.error_code !== 'MEDIA_PURGED'" class="button-danger" :disabled="busy" @click="purge">{{ t('清理原附件与转写', 'Remove attachment and transcript') }}</button>
         <template v-if="selected.status === 'completed'">
-          <div class="speaker-names"><label v-for="speaker in speakers" :key="speaker">{{ speaker }}<input v-model="selected.speaker_names[speaker]" class="input" placeholder="说话人显示名" @input="dirty = true" /></label></div>
+          <div class="speaker-names"><label v-for="speaker in speakers" :key="speaker">{{ speaker }}<input v-model="selected.speaker_names[speaker]" class="input" :placeholder="t('说话人显示名', 'Speaker display name')" @input="dirty = true" /></label></div>
-          <p v-if="selected.segments.length" class="subtle">时间戳对应音频分段边界，可点击定位播放。</p>
+          <p v-if="selected.segments.length" class="subtle">{{ t('时间戳对应音频分段边界，可点击定位播放。', 'Timestamps mark segment boundaries; click one to seek playback.') }}</p>
           <div v-for="segment in selected.segments" :key="segment.segment_id" class="segment" :class="{ current: position >= segment.start_time && position < segment.end_time }">
             <button class="button-secondary" @click="seek(segment.start_time)">{{ stamp(segment.start_time) }}</button><small>{{ selected.speaker_names[segment.speaker || ''] || segment.speaker }}</small>
             <textarea v-model="segment.text" class="input" rows="2" @input="dirty = true; selected.text = selected.segments.map(s => s.text).join('\n')" />
           </div>
           <textarea v-if="!selected.segments.length" v-model="selected.text" class="input" rows="12" @input="dirty = true" />
-          <div class="inline-actions"><button class="button-primary" :disabled="busy || !dirty" @click="action(async () => { selected = await mediaService.save(selected!); dirty = false; notice = '校对已保存' })">保存校对</button>
+          <div class="inline-actions"><button class="button-primary" :disabled="busy || !dirty" @click="action(async () => { selected = await mediaService.save(selected!); dirty = false; notice = t('校对已保存', 'Corrections saved') })">{{ t('保存校对', 'Save corrections') }}</button>
-            <button class="button-secondary" @click="action(async () => { history = (await mediaService.revisions(selected!.job_id)).items })">修订历史</button></div>
+            <button class="button-secondary" @click="action(async () => { history = (await mediaService.revisions(selected!.job_id)).items })">{{ t('修订历史', 'Revision history') }}</button></div>
-          <details><summary>原始识别文本</summary><pre>{{ selected.original_text }}</pre></details>
+          <details class="ui-disclosure"><summary>{{ t('原始识别文本', 'Original recognition text') }}</summary><pre>{{ selected.original_text }}</pre></details>
-          <details v-for="revision in history" :key="revision.revision"><summary>修订 {{ revision.revision }}</summary><pre>{{ revision.text }}</pre></details>
+          <details v-for="revision in history" :key="revision.revision" class="ui-disclosure"><summary>{{ t('修订', 'Revision') }} {{ revision.revision }}</summary><pre>{{ revision.text }}</pre></details>
-          <div class="inline-actions"><input v-model="title" class="input" aria-label="笔记标题" /><button class="button-primary" :disabled="busy || dirty || !title.trim()" @click="action(async () => { const note = await mediaService.note(selected!.job_id, title); notice = `已保存笔记：${note.title}` })">保存为笔记</button></div>
+          <div class="inline-actions"><label><input v-model="updateExisting" type="checkbox" />{{ t('更新上次导出的笔记（已手动修改则拒绝）', 'Update the previously exported note (refuse if manually edited)') }}</label><input v-model="title" class="input" :aria-label="t('笔记标题', 'Note title')" /><button class="button-primary" :disabled="busy || dirty || !title.trim()" @click="action(async () => { const note = await mediaService.note(selected!.job_id, title, updateExisting); notice = `${t('已保存笔记：', 'Saved note: ')}${note.title}` })">{{ t('保存为笔记', 'Save as note') }}</button></div>
         </template>
       </article>
-      <div v-else class="panel subtle">选择任务查看转写结果。</div>
+      <div v-else class="panel subtle">{{ t('选择任务查看转写结果。', 'Select a job to view its transcript.') }}</div>
     </div>
   </section>
 </template>
 
 <style scoped>
-.media-page{padding:28px;overflow:auto;height:100%;display:flex;flex-direction:column;gap:20px}.upload{display:grid;gap:12px;padding:20px}.media-columns{display:grid;grid-template-columns:260px minmax(0,1fr);gap:20px}.panel{padding:20px}.job-row{display:flex;flex-direction:column;gap:6px;width:100%;text-align:left;padding:12px;background:transparent;border:1px solid var(--color-border-default);border-radius:10px;margin-bottom:8px;cursor:pointer;color:inherit}.job-row small{overflow:hidden;text-overflow:ellipsis;max-width:100%}.selected,.current{background:var(--color-background-hover);outline:1px solid var(--color-accent-primary)}.transcript{display:flex;flex-direction:column;gap:16px}.transcript header,.segment{display:flex;gap:12px;align-items:center}.transcript>.button-danger{align-self:flex-start}.transcript>label{white-space:nowrap}.transcript>label select{width:160px}.segment textarea{flex:1}.speaker-names{display:flex;flex-wrap:wrap;gap:10px}audio{width:100%}pre{white-space:pre-wrap;word-break:break-word}label{display:flex;gap:8px;align-items:center}@media(max-width:850px){.media-columns{grid-template-columns:1fr}.segment{flex-wrap:wrap}}
+.media-page{padding:28px;overflow:auto;height:100%;display:flex;flex-direction:column;gap:20px}.upload{display:grid;gap:12px;padding:20px}.upload-options{display:flex;flex-wrap:wrap;gap:16px}.upload-actions{justify-content:flex-end}.media-columns{display:grid;grid-template-columns:260px minmax(0,1fr);gap:20px}.panel{padding:20px}.job-row{display:flex;flex-direction:column;gap:6px;width:100%;text-align:left;padding:12px;background:transparent;border:1px solid var(--color-border-default);border-radius:10px;margin-bottom:8px;cursor:pointer;color:inherit}.job-row small{overflow:hidden;text-overflow:ellipsis;max-width:100%}.selected,.current{background:var(--color-background-hover);outline:1px solid var(--color-accent-primary)}.transcript{display:flex;flex-direction:column;gap:16px}.transcript header,.segment{display:flex;gap:12px;align-items:center}.transcript>.button-danger{align-self:flex-start}.transcript>label{white-space:nowrap}.transcript>label select{width:160px}.segment textarea{flex:1}.speaker-names{display:flex;flex-wrap:wrap;gap:10px}audio{width:100%;border-radius:var(--radius-md);accent-color:var(--color-accent-primary)}pre{white-space:pre-wrap;word-break:break-word}label{display:flex;gap:8px;align-items:center}@media(max-width:850px){.media-columns{grid-template-columns:1fr}.segment{flex-wrap:wrap}}@media(max-width:560px){.upload-actions>*{flex:1}.upload-options{flex-direction:column}}
 </style>

frontend/src/features/plugins/PluginMcpPanel.vue

@@ -8,6 +8,7 @@ import * as pluginService from '@/services/pluginService'
 import { useEditorStore } from '@/stores/editor'
 import { usePluginStore } from '@/stores/plugin'
 import { useWorkspaceStore } from '@/stores/workspace'
+import { t, localeTag } from '@/i18n'
 
 const props = defineProps<{ plugin: Plugin }>()
 const pluginStore = usePluginStore()
@@ -31,8 +32,8 @@ let loadVersion = 0
 const hasSettings = computed(() => props.plugin.contributions.some((item) => item.type === 'settings_section'))
 const tabs = computed(() => [
   ...(props.plugin.backend_type === 'mcp' ? [{ id: 'host' as const, label: 'MCP Host' }] : []),
-  ...(hasSettings.value ? [{ id: 'settings' as const, label: '设置与密钥' }] : []),
+  ...(hasSettings.value ? [{ id: 'settings' as const, label: t('设置与密钥', 'Settings and secrets') }] : []),
-  { id: 'commands' as const, label: '插件命令' },
+  { id: 'commands' as const, label: t('插件命令', 'Plugin commands') },
 ])
 
 watch(() => props.plugin.plugin_id, () => {
@@ -48,7 +49,7 @@ watch(() => props.plugin.plugin_id, () => {
 
 function feedback(message = '') { error.value = message; notice.value = '' }
 function message(reason: unknown, fallback: string) { return reason instanceof Error ? reason.message : fallback }
-function formatTime(value?: string | null) { return value ? new Date(value).toLocaleString() : '—' }
+function formatTime(value?: string | null) { return value ? new Date(value).toLocaleString(localeTag()) : '—' }
 
 async function selectTab(tab: typeof activeTab.value) {
   activeTab.value = tab
@@ -80,7 +81,7 @@ async function loadActive() {
       }
     }
   } catch (reason) {
-    if (version === loadVersion) feedback(message(reason, 'MCP 数据加载失败'))
+    if (version === loadVersion) feedback(message(reason, t('MCP 数据加载失败', 'Failed to load MCP data')))
   } finally {
     if (version === loadVersion) loading.value = false
   }
@@ -92,8 +93,8 @@ async function restartHost() {
     await pluginService.restartPluginHost(props.plugin.plugin_id)
     host.value = await pluginService.getPluginHostStatus(props.plugin.plugin_id)
     await pluginStore.loadPlugins()
-    notice.value = 'MCP Host 已重启。'
+    notice.value = t('MCP Host 已重启。', 'MCP Host restarted.')
-  } catch (reason) { feedback(message(reason, 'MCP Host 重启失败')) } finally { busy.value = '' }
+  } catch (reason) { feedback(message(reason, t('MCP Host 重启失败', 'Failed to restart MCP Host'))) } finally { busy.value = '' }
 }
 function updateValue(field: PluginSettingField, raw: string | boolean) {
   values.value[field.key] = field.type === 'number' && typeof raw === 'string' ? (raw === '' ? null : Number(raw)) : raw
@@ -105,31 +106,31 @@ async function saveSettings() {
   try {
     schema.value = await pluginService.updatePluginSettings(props.plugin.plugin_id, schema.value.schema_version, values.value)
     values.value = { ...schema.value.values }
-    notice.value = '普通设置已保存。'
+    notice.value = t('普通设置已保存。', 'Settings saved.')
-  } catch (reason) { feedback(message(reason, '设置保存失败')) } finally { busy.value = '' }
+  } catch (reason) { feedback(message(reason, t('设置保存失败', 'Failed to save settings'))) } finally { busy.value = '' }
 }
 async function saveSecret(field: PluginSettingField) {
   const secret = secrets.value[field.key]?.trim()
-  if (!secret) { feedback('请输入' + field.label); return }
+  if (!secret) { feedback(t('请输入', 'Enter ') + field.label); return }
   busy.value = 'secret:' + field.key
   feedback()
   try {
     const state = await pluginService.putPluginSecret(props.plugin.plugin_id, field.key, secret)
     if (schema.value) schema.value.secrets[field.key] = { configured: state.configured }
     secrets.value[field.key] = ''
-    notice.value = field.label + '已加密保存。'
+    notice.value = field.label + t('已加密保存。', ' encrypted and saved.')
-  } catch (reason) { feedback(message(reason, '密钥保存失败')) } finally { busy.value = '' }
+  } catch (reason) { feedback(message(reason, t('密钥保存失败', 'Failed to save secret'))) } finally { busy.value = '' }
 }
 async function deleteSecret(field: PluginSettingField) {
-  if (!confirm('删除已保存的' + field.label + '？')) return
+  if (!confirm(t('删除已保存的', 'Delete saved ') + field.label + '？')) return
   busy.value = 'secret:' + field.key
   feedback()
   try {
     const state = await pluginService.deletePluginSecret(props.plugin.plugin_id, field.key)
     if (schema.value) schema.value.secrets[field.key] = { configured: state.configured }
     secrets.value[field.key] = ''
-    notice.value = field.label + '已删除。'
+    notice.value = field.label + t('已删除。', ' deleted.')
-  } catch (reason) { feedback(message(reason, '密钥删除失败')) } finally { busy.value = '' }
+  } catch (reason) { feedback(message(reason, t('密钥删除失败', 'Failed to delete secret'))) } finally { busy.value = '' }
 }
 function properties(command: PluginCommand): Record<string, Record<string, unknown>> {
   const result = command.parameters.properties
@@ -165,7 +166,7 @@ async function execute(command: PluginCommand) {
       selection: null,
     })
     if (result.effect.type === 'notification') notice.value = result.effect.payload.message
-    else if (result.effect.type === 'job') notice.value = '后台任务已创建：' + result.effect.payload.job_id
+    else if (result.effect.type === 'job') notice.value = t('后台任务已创建：', 'Background job created: ') + result.effect.payload.job_id
     else if (result.effect.type === 'navigate') {
       const routes: Record<string, string> = {
         'vault-entry': '/', workspace: '/workspace', search: '/search', chat: '/chat',
@@ -175,64 +176,64 @@ async function execute(command: PluginCommand) {
       await router.push(routes[result.effect.payload.route])
     } else if (result.effect.type === 'refresh') {
       await loadActive()
-      notice.value = '相关数据已刷新。'
+      notice.value = t('相关数据已刷新。', 'Related data refreshed.')
-    } else notice.value = '命令执行完成。'
+    } else notice.value = t('命令执行完成。', 'Command completed.')
-  } catch (reason) { feedback(message(reason, '命令执行失败')) } finally { busy.value = '' }
+  } catch (reason) { feedback(message(reason, t('命令执行失败', 'Command failed'))) } finally { busy.value = '' }
 }
 </script>
 
 <template>
   <section class="mcp-panel">
-    <nav class="mcp-tabs" aria-label="MCP 与 Plugin 配置">
+    <nav class="mcp-tabs" :aria-label="t('MCP 与 Plugin 配置', 'MCP and Plugin settings')">
       <button v-for="tab in tabs" :key="tab.id" :class="{ active: activeTab === tab.id }" @click="selectTab(tab.id)">{{ tab.label }}</button>
     </nav>
     <div v-if="error" class="error-banner">{{ error }}</div>
     <div v-if="notice" class="notice-banner">{{ notice }}</div>
 
     <div v-if="activeTab === 'host'" class="mcp-section">
-      <div class="section-head"><div><h3>MCP Host 状态</h3><p>查看协议协商、运行状态与 Host 错误。</p></div><div class="inline-actions"><button class="button-secondary" :disabled="loading" @click="loadActive"><AppIcon :icon="Refresh" :size="15" />刷新</button><button class="button-primary" :disabled="busy === 'host' || !plugin.enabled" @click="restartHost">{{ busy === 'host' ? '重启中…' : '重启 Host' }}</button></div></div>
+      <div class="section-head"><div><h3>{{ t('MCP Host 状态', 'MCP Host status') }}</h3><p>{{ t('查看协议协商、运行状态与 Host 错误。', 'Inspect protocol negotiation, runtime status, and Host errors.') }}</p></div><div class="inline-actions"><button class="button-secondary" :disabled="loading" @click="loadActive"><AppIcon :icon="Refresh" :size="15" />{{ t('刷新', 'Refresh') }}</button><button class="button-primary" :disabled="busy === 'host' || !plugin.enabled" @click="restartHost">{{ busy === 'host' ? t('重启中…', 'Restarting…') : t('重启 Host', 'Restart Host') }}</button></div></div>
       <div v-if="host" class="status-grid">
-        <div><span>状态</span><strong><i class="status-dot" :class="host.status"></i>{{ host.status }}</strong></div>
+        <div><span>{{ t('状态', 'Status') }}</span><strong><i class="status-dot" :class="host.status"></i>{{ host.status }}</strong></div>
-        <div><span>服务</span><strong>{{ host.server_name || '—' }} {{ host.server_version || '' }}</strong></div>
+        <div><span>{{ t('服务', 'Server') }}</span><strong>{{ host.server_name || '—' }} {{ host.server_version || '' }}</strong></div>
-        <div><span>协议版本</span><strong>{{ host.protocol_version || '—' }}</strong></div>
+        <div><span>{{ t('协议版本', 'Protocol version') }}</span><strong>{{ host.protocol_version || '—' }}</strong></div>
-        <div><span>工具数量</span><strong>{{ host.tools_count }}</strong></div>
+        <div><span>{{ t('工具数量', 'Tools') }}</span><strong>{{ host.tools_count }}</strong></div>
-        <div><span>启动时间</span><strong>{{ formatTime(host.started_at) }}</strong></div>
+        <div><span>{{ t('启动时间', 'Started') }}</span><strong>{{ formatTime(host.started_at) }}</strong></div>
-        <div><span>最后心跳</span><strong>{{ formatTime(host.last_seen_at) }}</strong></div>
+        <div><span>{{ t('最后心跳', 'Last heartbeat') }}</span><strong>{{ formatTime(host.last_seen_at) }}</strong></div>
       </div>
-      <div v-else-if="loading" class="empty-state">正在读取 Host 状态…</div>
+      <div v-else-if="loading" class="empty-state">{{ t('正在读取 Host 状态…', 'Loading Host status…') }}</div>
       <div v-if="host?.error" class="error-banner host-error">{{ host.error }}</div>
-      <p class="security-hint">当前仅运行插件清单声明的 stdio MCP Server，不开放任意 Shell 命令和环境变量编辑。</p>
+      <p class="security-hint">{{ t('当前仅运行插件清单声明的 stdio MCP Server，不开放任意 Shell 命令和环境变量编辑。', 'Only stdio MCP servers declared by the plugin manifest can run. Arbitrary shell commands and environment variable editing are unavailable.') }}</p>
     </div>
 
     <div v-else-if="activeTab === 'settings'" class="mcp-section">
-      <div class="section-head"><div><h3>设置与密钥</h3><p>表单由后端 Schema 生成；密钥不会被读取或回显。</p></div><button class="button-primary" :disabled="!schema || busy === 'settings'" @click="saveSettings">{{ busy === 'settings' ? '保存中…' : '保存普通设置' }}</button></div>
+      <div class="section-head"><div><h3>{{ t('设置与密钥', 'Settings and secrets') }}</h3><p>{{ t('表单由后端 Schema 生成；密钥不会被读取或回显。', 'The backend schema generates this form. Secrets are never read back or displayed.') }}</p></div><button class="button-primary" :disabled="!schema || busy === 'settings'" @click="saveSettings">{{ busy === 'settings' ? t('保存中…', 'Saving…') : t('保存普通设置', 'Save settings') }}</button></div>
       <div v-if="schema" class="settings-list">
         <div v-for="field in schema.fields" :key="field.key" class="setting-row">
-          <div class="field-copy"><label :for="'plugin-setting-' + field.key"><AppIcon v-if="field.type === 'secret'" :icon="Key" :size="15" />{{ field.label }}<em v-if="field.required">必填</em></label><p>{{ field.description || (field.type === 'secret' ? '加密保存，不在页面回显。' : '') }}</p></div>
+          <div class="field-copy"><label :for="'plugin-setting-' + field.key"><AppIcon v-if="field.type === 'secret'" :icon="Key" :size="15" />{{ field.label }}<em v-if="field.required">{{ t('必填', 'Required') }}</em></label><p>{{ field.description || (field.type === 'secret' ? t('加密保存，不在页面回显。', 'Encrypted and never displayed.') : '') }}</p></div>
           <template v-if="field.type === 'secret'">
-            <div class="secret-control"><input :id="'plugin-setting-' + field.key" :value="secrets[field.key] || ''" class="input" type="password" autocomplete="new-password" :placeholder="schema.secrets[field.key]?.configured ? '已配置；输入新值可替换' : '输入密钥'" @input="secrets[field.key] = ($event.target as HTMLInputElement).value"><button class="button-secondary" :disabled="!secrets[field.key]?.trim() || busy === 'secret:' + field.key" @click="saveSecret(field)">安全保存</button><button v-if="schema.secrets[field.key]?.configured" class="button-danger" @click="deleteSecret(field)">删除</button></div>
+            <div class="secret-control"><input :id="'plugin-setting-' + field.key" :value="secrets[field.key] || ''" class="input" type="password" autocomplete="new-password" :placeholder="schema.secrets[field.key]?.configured ? t('已配置；输入新值可替换', 'Configured; enter a new value to replace') : t('输入密钥', 'Enter secret')" @input="secrets[field.key] = ($event.target as HTMLInputElement).value"><button class="button-secondary" :disabled="!secrets[field.key]?.trim() || busy === 'secret:' + field.key" @click="saveSecret(field)">{{ t('安全保存', 'Save securely') }}</button><button v-if="schema.secrets[field.key]?.configured" class="button-danger" @click="deleteSecret(field)">{{ t('删除', 'Delete') }}</button></div>
-            <span class="secret-state" :class="{ configured: schema.secrets[field.key]?.configured }">{{ schema.secrets[field.key]?.configured ? '已配置' : '未配置' }}</span>
+            <span class="secret-state" :class="{ configured: schema.secrets[field.key]?.configured }">{{ schema.secrets[field.key]?.configured ? t('已配置', 'Configured') : t('未配置', 'Not configured') }}</span>
           </template>
-          <template v-else-if="field.type === 'boolean'"><label class="check-control"><input :id="'plugin-setting-' + field.key" type="checkbox" :checked="Boolean(values[field.key])" @change="updateValue(field, ($event.target as HTMLInputElement).checked)">{{ values[field.key] ? '开启' : '关闭' }}</label></template>
+          <template v-else-if="field.type === 'boolean'"><label class="check-control"><input :id="'plugin-setting-' + field.key" type="checkbox" :checked="Boolean(values[field.key])" @change="updateValue(field, ($event.target as HTMLInputElement).checked)">{{ values[field.key] ? t('开启', 'On') : t('关闭', 'Off') }}</label></template>
           <template v-else-if="field.type === 'select'"><select :id="'plugin-setting-' + field.key" class="select" :value="values[field.key]" @change="updateValue(field, ($event.target as HTMLSelectElement).value)"><option v-for="option in field.options" :key="option" :value="option">{{ option }}</option></select></template>
           <template v-else><input :id="'plugin-setting-' + field.key" class="input" :type="field.type === 'number' ? 'number' : 'text'" :min="field.minimum ?? undefined" :max="field.maximum ?? undefined" :required="field.required" :value="values[field.key] ?? ''" @input="updateValue(field, ($event.target as HTMLInputElement).value)"></template>
         </div>
       </div>
-      <div v-else-if="loading" class="empty-state">正在读取 Plugin 设置…</div>
+      <div v-else-if="loading" class="empty-state">{{ t('正在读取 Plugin 设置…', 'Loading Plugin settings…') }}</div>
     </div>
 
     <div v-else class="mcp-section">
-      <div class="section-head"><div><h3>Plugin 命令</h3><p>执行该 Plugin 注册的受控 Command Contribution。</p></div><button class="button-secondary" :disabled="loading" @click="loadActive"><AppIcon :icon="Refresh" :size="15" />刷新</button></div>
+      <div class="section-head"><div><h3>{{ t('Plugin 命令', 'Plugin commands') }}</h3><p>{{ t('执行该 Plugin 注册的受控 Command Contribution。', 'Run controlled command contributions registered by this Plugin.') }}</p></div><button class="button-secondary" :disabled="loading" @click="loadActive"><AppIcon :icon="Refresh" :size="15" />{{ t('刷新', 'Refresh') }}</button></div>
       <div v-if="commands.length" class="command-list">
         <article v-for="command in commands" :key="command.command_id" class="item-card command-card">
-          <div class="command-head"><div><strong>{{ command.title }}</strong><p>{{ command.description || command.command_id }}</p></div><span class="badge" :class="{ success: commandAvailable(command), warning: command.enabled && !commandAvailable(command) }">{{ commandAvailable(command) ? '可执行' : command.enabled ? '缺少上下文' : '不可用' }}</span></div>
+          <div class="command-head"><div><strong>{{ command.title }}</strong><p>{{ command.description || command.command_id }}</p></div><span class="badge" :class="{ success: commandAvailable(command), warning: command.enabled && !commandAvailable(command) }">{{ commandAvailable(command) ? t('可执行', 'Available') : command.enabled ? t('缺少上下文', 'Missing context') : t('不可用', 'Unavailable') }}</span></div>
           <div v-if="Object.keys(properties(command)).length" class="command-fields">
-            <label v-for="(definition, key) in properties(command)" :key="key" class="field"><span>{{ String(definition.title || key) }}<em v-if="required(command, key)">必填</em></span><select v-if="Array.isArray(definition.enum)" class="select" @change="updateArgument(command.command_id, key, ($event.target as HTMLSelectElement).value, definition)"><option value="">请选择</option><option v-for="option in definition.enum" :key="String(option)" :value="String(option)">{{ option }}</option></select><select v-else-if="definition.type === 'boolean'" class="select" @change="updateArgument(command.command_id, key, ($event.target as HTMLSelectElement).value, definition)"><option value="false">否</option><option value="true">是</option></select><input v-else class="input" :type="definition.type === 'number' || definition.type === 'integer' ? 'number' : 'text'" @input="updateArgument(command.command_id, key, ($event.target as HTMLInputElement).value, definition)"></label>
+            <label v-for="(definition, key) in properties(command)" :key="key" class="field"><span>{{ String(definition.title || key) }}<em v-if="required(command, key)">{{ t('必填', 'Required') }}</em></span><select v-if="Array.isArray(definition.enum)" class="select" @change="updateArgument(command.command_id, key, ($event.target as HTMLSelectElement).value, definition)"><option value="">{{ t('请选择', 'Select') }}</option><option v-for="option in definition.enum" :key="String(option)" :value="String(option)">{{ option }}</option></select><select v-else-if="definition.type === 'boolean'" class="select" @change="updateArgument(command.command_id, key, ($event.target as HTMLSelectElement).value, definition)"><option value="false">{{ t('否', 'No') }}</option><option value="true">{{ t('是', 'Yes') }}</option></select><input v-else class="input" :type="definition.type === 'number' || definition.type === 'integer' ? 'number' : 'text'" @input="updateArgument(command.command_id, key, ($event.target as HTMLInputElement).value, definition)"></label>
           </div>
-          <button class="button-primary command-run" :disabled="!commandAvailable(command) || busy === command.command_id" @click="execute(command)"><AppIcon :icon="VideoPlay" :size="15" />{{ busy === command.command_id ? '执行中…' : '执行命令' }}</button>
+          <button class="button-primary command-run" :disabled="!commandAvailable(command) || busy === command.command_id" @click="execute(command)"><AppIcon :icon="VideoPlay" :size="15" />{{ busy === command.command_id ? t('执行中…', 'Running…') : t('执行命令', 'Run command') }}</button>
         </article>
       </div>
-      <div v-else-if="!loading" class="empty-state"><div><strong>没有可用命令</strong><p>启用 Plugin 后，已注册的命令会出现在这里。</p></div></div>
+      <div v-else-if="!loading" class="empty-state"><div><strong>{{ t('没有可用命令', 'No available commands') }}</strong><p>{{ t('启用 Plugin 后，已注册的命令会出现在这里。', 'Registered commands appear here after the Plugin is enabled.') }}</p></div></div>
     </div>
   </section>
 </template>

frontend/src/features/plugins/PluginsView.vue

@@ -4,31 +4,32 @@ import AppIcon from '@/components/common/AppIcon.vue'
 import PluginMcpPanel from './PluginMcpPanel.vue'
 import { onMounted, ref } from 'vue'
 import { usePluginStore } from '@/stores/plugin'
+import { t } from '@/i18n'
 
 const pluginStore = usePluginStore()
 const actionError = ref('')
 onMounted(() => { void pluginStore.loadPlugins() })
 
-async function install() { const path = prompt('请输入 Plugin Package 路径')?.trim(); if (!path) return; try { await pluginStore.installPlugin(path) } catch (error) { actionError.value = error instanceof Error ? error.message : '安装失败' } }
+async function install() { const path = prompt(t('请输入 Plugin Package 路径', 'Enter the Plugin Package path'))?.trim(); if (!path) return; try { await pluginStore.installPlugin(path) } catch (error) { actionError.value = error instanceof Error ? error.message : t('安装失败', 'Installation failed') } }
-async function toggle(id: string, enabled: boolean) { try { enabled ? await pluginStore.disablePlugin(id) : await pluginStore.enablePlugin(id) } catch (error) { actionError.value = error instanceof Error ? error.message : '状态更新失败' } }
+async function toggle(id: string, enabled: boolean) { try { enabled ? await pluginStore.disablePlugin(id) : await pluginStore.enablePlugin(id) } catch (error) { actionError.value = error instanceof Error ? error.message : t('状态更新失败', 'Status update failed') } }
-async function grant(id: string, permissions: string[]) { if (!confirm(`将授权：${permissions.join('、')}。是否继续？`)) return; try { await pluginStore.grantPermissions(id, permissions) } catch (error) { actionError.value = error instanceof Error ? error.message : '授权失败' } }
+async function grant(id: string, permissions: string[]) { if (!confirm(`${t('将授权：', 'Grant permissions: ')}${permissions.join(', ')}。${t('是否继续？', 'Continue?')}`)) return; try { await pluginStore.grantPermissions(id, permissions) } catch (error) { actionError.value = error instanceof Error ? error.message : t('授权失败', 'Authorization failed') } }
-async function uninstall(id: string, name: string) { if (!confirm(`卸载“${name}”将移除其全部 Contribution，是否继续？`)) return; try { await pluginStore.uninstallPlugin(id) } catch (error) { actionError.value = error instanceof Error ? error.message : '卸载失败' } }
+async function uninstall(id: string, name: string) { if (!confirm(t(`卸载“${name}”将移除其全部 Contribution，是否继续？`, `Uninstalling “${name}” removes all its contributions. Continue?`))) return; try { await pluginStore.uninstallPlugin(id) } catch (error) { actionError.value = error instanceof Error ? error.message : t('卸载失败', 'Uninstall failed') } }
 </script>
 
 <template>
   <section class="feature-page">
-    <header class="feature-header"><div><h1>Plugin 与 MCP</h1><p>管理插件生命周期、MCP Host、权限和受控 Contribution。</p></div><button class="button-primary" @click="install">安装 Plugin</button></header>
+    <header class="feature-header"><div><h1>{{ t('Plugin 与 MCP', 'Plugins and MCP') }}</h1><p>{{ t('管理插件生命周期、MCP Host、权限和受控 Contribution。', 'Manage plugin lifecycles, MCP hosts, permissions, and controlled contributions.') }}</p></div><button class="button-primary" @click="install">{{ t('安装 Plugin', 'Install Plugin') }}</button></header>
     <div v-if="pluginStore.error || actionError" class="error-banner">{{ pluginStore.error || actionError }}</div>
     <div v-if="pluginStore.selectedPlugin" class="panel">
-      <div class="detail-head"><div><span class="badge" :class="{ success: pluginStore.selectedPlugin.status === 'ready', error: pluginStore.selectedPlugin.status === 'error', warning: pluginStore.selectedPlugin.status === 'permission_required' }">{{ pluginStore.selectedPlugin.status }}</span><h2>{{ pluginStore.selectedPlugin.icon }} {{ pluginStore.selectedPlugin.name }}</h2><p class="muted">v{{ pluginStore.selectedPlugin.version }} · {{ pluginStore.selectedPlugin.backend_type || 'none' }}/{{ pluginStore.selectedPlugin.transport || 'none' }}</p></div><div class="inline-actions"><button v-if="pluginStore.selectedPlugin.status === 'permission_required'" class="button-primary" @click="grant(pluginStore.selectedPlugin.plugin_id, pluginStore.selectedPlugin.permissions)">授权权限</button><button class="button-secondary" @click="toggle(pluginStore.selectedPlugin.plugin_id, pluginStore.selectedPlugin.enabled)">{{ pluginStore.selectedPlugin.enabled ? '停用' : '启用' }}</button><button class="button-danger" @click="uninstall(pluginStore.selectedPlugin.plugin_id, pluginStore.selectedPlugin.name)">卸载</button></div></div>
+      <div class="detail-head"><div><span class="badge" :class="{ success: pluginStore.selectedPlugin.status === 'ready', error: pluginStore.selectedPlugin.status === 'error', warning: pluginStore.selectedPlugin.status === 'permission_required' }">{{ pluginStore.selectedPlugin.status }}</span><h2>{{ pluginStore.selectedPlugin.icon }} {{ pluginStore.selectedPlugin.name }}</h2><p class="muted">v{{ pluginStore.selectedPlugin.version }} · {{ pluginStore.selectedPlugin.backend_type || 'none' }}/{{ pluginStore.selectedPlugin.transport || 'none' }}</p></div><div class="inline-actions"><button v-if="pluginStore.selectedPlugin.status === 'permission_required'" class="button-primary" @click="grant(pluginStore.selectedPlugin.plugin_id, pluginStore.selectedPlugin.permissions)">{{ t('授权权限', 'Grant permissions') }}</button><button class="button-secondary" @click="toggle(pluginStore.selectedPlugin.plugin_id, pluginStore.selectedPlugin.enabled)">{{ pluginStore.selectedPlugin.enabled ? t('停用', 'Disable') : t('启用', 'Enable') }}</button><button class="button-danger" @click="uninstall(pluginStore.selectedPlugin.plugin_id, pluginStore.selectedPlugin.name)">{{ t('卸载', 'Uninstall') }}</button></div></div>
       <p class="description">{{ pluginStore.selectedPlugin.description }}</p>
-      <div class="detail-grid"><div><h3>权限</h3><div class="tag-list"><span v-for="permission in pluginStore.selectedPlugin.permissions" :key="permission" class="badge warning">{{ permission }}</span></div></div><div><h3>Contribution</h3><div class="contribution-list"><div v-for="item in pluginStore.selectedPlugin.contributions" :key="item.id" class="item-card"><span class="badge info">{{ item.type }}</span><strong>{{ item.name }}</strong><p class="subtle">{{ item.description || item.id }}</p></div></div></div></div>
+      <div class="detail-grid"><div><h3>{{ t('权限', 'Permissions') }}</h3><div class="tag-list"><span v-for="permission in pluginStore.selectedPlugin.permissions" :key="permission" class="badge warning">{{ permission }}</span></div></div><div><h3>Contribution</h3><div class="contribution-list"><div v-for="item in pluginStore.selectedPlugin.contributions" :key="item.id" class="item-card"><span class="badge info">{{ item.type }}</span><strong>{{ item.name }}</strong><p class="subtle">{{ item.description || item.id }}</p></div></div></div></div>
       <div v-if="pluginStore.selectedPlugin.last_error" class="error-banner last-error">{{ pluginStore.selectedPlugin.last_error }}</div>
-      <div v-if="pluginStore.selectedPlugin.dependent_skills?.length" class="notice-banner last-error">依赖此插件的 Skill：{{ pluginStore.selectedPlugin.dependent_skills.join('、') }}</div>
+      <div v-if="pluginStore.selectedPlugin.dependent_skills?.length" class="notice-banner last-error">{{ t('依赖此插件的 Skill：', 'Skills that depend on this plugin: ') }}{{ pluginStore.selectedPlugin.dependent_skills.join(', ') }}</div>
       <PluginMcpPanel :plugin="pluginStore.selectedPlugin" />
     </div>
-    <div v-else-if="!pluginStore.plugins.length" class="empty-state"><div><strong>{{ pluginStore.isLoading ? '正在加载…' : pluginStore.error ? '加载失败' : '尚未安装' }}</strong><button class="button-secondary" @click="pluginStore.loadPlugins">重新加载</button></div></div>
+    <div v-else-if="!pluginStore.plugins.length" class="empty-state"><div><strong>{{ pluginStore.isLoading ? t('正在加载…', 'Loading…') : pluginStore.error ? t('加载失败', 'Load failed') : t('尚未安装', 'No plugins installed') }}</strong><button class="button-secondary" @click="pluginStore.loadPlugins">{{ t('重新加载', 'Reload') }}</button></div></div>
-    <div v-else class="feature-grid"><article v-for="plugin in pluginStore.plugins" :key="plugin.plugin_id" class="item-card extension-card" @click="pluginStore.selectPlugin(plugin.plugin_id)"><div class="extension-title"><AppIcon :icon="Connection" :size="22" /><div><strong>{{ plugin.name }}</strong><p>v{{ plugin.version }}</p></div><span class="badge" :class="{ success: plugin.status === 'ready', error: plugin.status === 'error', warning: plugin.status === 'permission_required' }">{{ plugin.status }}</span></div><p class="muted">{{ plugin.description }}</p><p class="subtle">{{ plugin.permissions.length }} 项权限 · {{ plugin.contributions.length }} 项 Contribution</p></article></div>
+    <div v-else class="feature-grid"><article v-for="plugin in pluginStore.plugins" :key="plugin.plugin_id" class="item-card extension-card" @click="pluginStore.selectPlugin(plugin.plugin_id)"><div class="extension-title"><AppIcon :icon="Connection" :size="22" /><div><strong>{{ plugin.name }}</strong><p>v{{ plugin.version }}</p></div><span class="badge" :class="{ success: plugin.status === 'ready', error: plugin.status === 'error', warning: plugin.status === 'permission_required' }">{{ plugin.status }}</span></div><p class="muted">{{ plugin.description }}</p><p class="subtle">{{ plugin.permissions.length }} {{ t('项权限', 'permissions') }} · {{ plugin.contributions.length }} {{ t('项 Contribution', 'contributions') }}</p></article></div>
   </section>
 </template>
 

frontend/src/features/search/SearchFiltersPanel.vue

@@ -1,24 +1,26 @@
 <script setup lang="ts">
 import { useSearchStore } from '@/stores/search'
+import { computed } from 'vue'
+import { t } from '@/i18n'
 
 const searchStore = useSearchStore()
-const modes = [
+const modes = computed(() => [
-  { value: 'hybrid', label: '混合检索' },
+  { value: 'hybrid' as const, label: t('混合检索', 'Hybrid search') },
-  { value: 'fts', label: '全文检索' },
+  { value: 'fts' as const, label: t('全文检索', 'Full-text search') },
-  { value: 'vector', label: '向量检索' },
+  { value: 'vector' as const, label: t('向量检索', 'Vector search') },
-] as const
+])
 </script>
 
 <template>
   <div class="sidebar-panel">
-    <p class="subtle">检索模式</p>
+    <p class="subtle">{{ t('检索模式', 'Search mode') }}</p>
     <div class="sidebar-list mode-list">
       <button v-for="item in modes" :key="item.value" class="sidebar-list-item"
         :class="{ active: searchStore.mode === item.value }" @click="searchStore.setMode(item.value)">
         {{ item.label }}
       </button>
     </div>
-    <p class="subtle section-title">最近搜索</p>
+    <p class="subtle section-title">{{ t('最近搜索', 'Recent searches') }}</p>
     <div class="sidebar-list">
       <button v-for="query in searchStore.recentQueries" :key="query" class="sidebar-list-item recent"
         @click="searchStore.doSearch({ query, mode: searchStore.mode })">{{ query }}</button>

frontend/src/features/search/SearchView.vue

@@ -5,6 +5,7 @@ import type { SearchResult } from '@/contracts'
 import { useEditorStore } from '@/stores/editor'
 import { useSearchStore } from '@/stores/search'
 import { useWorkspaceStore } from '@/stores/workspace'
+import { t } from '@/i18n'
 
 const searchStore = useSearchStore()
 onMounted(() => { void searchStore.loadHistory() })
@@ -34,28 +35,28 @@ async function openResult(result: SearchResult) {
 <template>
   <section class="feature-page search-page">
     <header class="feature-header">
-      <div><h1>搜索知识库</h1><p>在当前 Vault 中进行全文、向量或混合检索。</p></div>
+      <div><h1>{{ t('搜索知识库', 'Search Knowledge Base') }}</h1><p>{{ t('在当前 Vault 中进行全文、向量或混合检索。', 'Run full-text, vector, or hybrid search in the current Vault.') }}</p></div>
     </header>
     <form class="search-form panel" @submit.prevent="submitSearch">
-      <input v-model="searchStore.query" class="input search-input" placeholder="搜索笔记内容、标题或标签" autofocus />
+      <input v-model="searchStore.query" class="input search-input" :placeholder="t('搜索笔记内容、标题或标签', 'Search note content, titles, or tags')" autofocus />
       <button class="button-primary" :disabled="!searchStore.query.trim() || searchStore.isSearching">
-        {{ searchStore.isSearching ? '搜索中…' : '搜索' }}
+        {{ searchStore.isSearching ? t('搜索中…', 'Searching…') : t('搜索', 'Search') }}
       </button>
       <div class="form-grid advanced">
-        <div class="field"><label>文件夹范围</label><input v-model="folder" class="input" placeholder="例如 /数据结构" /></div>
+        <div class="field"><label>{{ t('文件夹范围', 'Folder scope') }}</label><input v-model="folder" class="input" :placeholder="t('例如 /数据结构', 'For example /Data Structures')" /></div>
-        <div class="field"><label>标签</label><input v-model="tag" class="input" placeholder="例如 算法" /></div>
+        <div class="field"><label>{{ t('标签', 'Tag') }}</label><input v-model="tag" class="input" :placeholder="t('例如 算法', 'For example algorithms')" /></div>
       </div>
     </form>
     <div v-if="searchStore.error" class="error-banner">{{ searchStore.error }}</div>
     <div v-if="searchStore.historyError" class="notice-banner">{{ searchStore.historyError }}</div>
     <div v-if="searchStore.recentQueries.length" class="search-history">
-      <span class="subtle">最近搜索（保存在应用数据中）</span>
+      <span class="subtle">{{ t('最近搜索（保存在应用数据中）', 'Recent searches (stored in application data)') }}</span>
       <button v-for="item in searchStore.recentQueries" :key="item" class="button-secondary" @click="searchStore.query = item; submitSearch()">{{ item }}</button>
-      <button class="button-secondary" @click="searchStore.clearHistory">清空记录</button>
+      <button class="button-secondary" @click="searchStore.clearHistory">{{ t('清空记录', 'Clear history') }}</button>
     </div>
-    <div v-if="searchStore.vectorUnavailable" class="notice-banner">向量索引不可用，已保留全文检索能力。</div>
+    <div v-if="searchStore.vectorUnavailable" class="notice-banner">{{ t('向量索引不可用，已保留全文检索能力。', 'Vector search is unavailable; full-text search remains active.') }}</div>
     <div v-if="searchStore.results.length" class="results-header">
-      <span>找到 {{ searchStore.total }} 条结果</span><span class="badge info">{{ searchStore.mode }}</span>
+      <span>{{ t('找到', 'Found') }} {{ searchStore.total }} {{ t('条结果', 'results') }}</span><span class="badge info">{{ searchStore.mode }}</span>
     </div>
     <div v-if="searchStore.results.length" class="result-list">
       <article v-for="result in searchStore.results" :key="`${result.note_id}:${result.block_id}`"
@@ -63,11 +64,11 @@ async function openResult(result: SearchResult) {
         <div class="result-title"><strong>{{ result.note_title }}</strong><span class="badge">{{ result.match_type }}</span></div>
         <p class="subtle">{{ result.file_path }} · {{ result.heading_path }}</p>
         <p class="snippet">{{ result.snippet }}</p>
-        <div class="result-meta"><span>相关度 {{ Math.round(result.score * 100) }}%</span><span>点击定位原文 →</span></div>
+        <div class="result-meta"><span>{{ t('相关度', 'Relevance') }} {{ Math.round(result.score * 100) }}%</span><span>{{ t('点击定位原文 →', 'Open source →') }}</span></div>
       </article>
     </div>
     <div v-else-if="!searchStore.isSearching" class="empty-state">
-      <div><strong>{{ searchStore.query ? '没有找到匹配内容' : '从你的知识库开始搜索' }}</strong><p>可切换检索模式或缩小文件夹、标签范围。</p></div>
+      <div><strong>{{ searchStore.query ? t('没有找到匹配内容', 'No matching content') : t('从你的知识库开始搜索', 'Start searching your knowledge base') }}</strong><p>{{ t('可切换检索模式或缩小文件夹、标签范围。', 'Try another search mode or narrow the folder and tag scope.') }}</p></div>
     </div>
   </section>
 </template>

frontend/src/features/settings/LocalModelSettings.spec.ts

@@ -0,0 +1,26 @@
+// @vitest-environment happy-dom
+import { mount, flushPromises } from '@vue/test-utils'
+import { expect, it, vi } from 'vitest'
+import { apiClient } from '@/services/apiClient'
+import LocalModelSettings from './LocalModelSettings.vue'
+
+vi.mock('@/services/apiClient', () => ({apiClient:{get:vi.fn(),post:vi.fn()}}))
+it('shows an optional CUDA installer and live installation stage', async () => {
+  vi.mocked(apiClient.get).mockImplementation(async (url) => url.includes('runtime-components')
+    ? {status:'not_installed', stage:'尚未安装', supported:true, cuda_available:null,custom_interpreter:false}
+    : {items:[],config:null,runtime_installed:true,last_inference:null})
+  vi.mocked(apiClient.post).mockResolvedValue({status:'installing',stage:'下载并安装 PyTorch CUDA（约 3 GB）',supported:true})
+  const wrapper = mount(LocalModelSettings)
+  try {
+    await flushPromises()
+    const button = wrapper.findAll('button').find(b => b.text() === '下载并安装 CUDA 组件')!
+    expect(button.exists()).toBe(true)
+    expect(apiClient.post).not.toHaveBeenCalled()
+    await button.trigger('click')
+    await flushPromises()
+    expect(apiClient.post).toHaveBeenCalledWith('/api/local-models/runtime-components/cuda')
+    expect(wrapper.text()).toContain('下载并安装 PyTorch CUDA')
+    expect(wrapper.get('progress').attributes('value')).toBeUndefined()
+    expect(button.attributes('disabled')).toBeDefined()
+  } finally {wrapper.unmount()}
+})

frontend/src/features/settings/LocalModelSettings.vue

@@ -1,20 +1,32 @@
 <script setup lang="ts">
-import { onMounted, onUnmounted, ref } from 'vue'
+import { computed, onMounted, onUnmounted, ref } from 'vue'
 import { apiClient } from '@/services/apiClient'
+import { t } from '@/i18n'
 interface Config {device: 'cpu'|'cuda'; cpu_threads: number; memory_limit_mb: number; gpu_memory_limit_mb: number; timeout_seconds: number; embedding_model: string; version: number}
-interface Model {key: string; name: string; capability: string; revision: string; license: string; status: string; downloaded_bytes: number; total_bytes: number|null; error_code?: string}
+interface Model {key: string; name: string; capability: string; revision: string; license: string; status: string; disk_bytes: number|null; downloaded_bytes: number; total_bytes: number|null; error_code?: string}
 const items = ref<Model[]>([])
 const config = ref<Config | null>(null)
 const installed = ref(false)
-const lastInference = ref<{actual_device:string;requested_device:string;inference_seconds:number}|null>(null)
+interface CudaComponent {status:string;stage:string;cuda_available:boolean|null;supported:boolean;custom_interpreter:boolean;error?:string;torch?:string}
+const cuda = ref<CudaComponent|null>(null)
+const cudaError = ref('')
+async function loadCuda() {
+  try { cuda.value = await apiClient.get<CudaComponent>('/api/local-models/runtime-components/cuda'); cudaError.value = '' }
+  catch(e) { cudaError.value = (e as Error).message }
+}
+async function installCuda() {
+  await act(async () => { cuda.value = await apiClient.post<CudaComponent>('/api/local-models/runtime-components/cuda') })
+}
+const lastInference = ref<{actual_device:string;requested_device:string;inference_seconds?:number;elapsed_seconds?:number;status?:string;error_code?:string}|null>(null)
 const error = ref('')
 const dirty = ref(false)
 const busy = ref(false)
 let timer: ReturnType<typeof setTimeout> | undefined
 let stopped = false
-const size = (bytes: number | null) => bytes === null ? '未知' : `${(bytes / 1024 / 1024).toFixed(1)} MiB`
+const size = (bytes: number | null) => bytes === null ? t('未知', 'Unknown') : `${(bytes / 1024 / 1024).toFixed(1)} MiB`
-const labels: Record<string,string> = {not_installed:'未下载',downloading:'下载中',installed:'已下载并校验',failed:'下载失败',interrupted:'已中断，可续传'}
+const labels = computed<Record<string,string>>(() => ({not_installed:t('未下载','Not downloaded'),downloading:t('下载中','Downloading'),installed:t('已下载并校验','Downloaded and verified'),failed:t('下载失败','Download failed'),interrupted:t('已中断，可续传','Interrupted; resumable')}))
 async function load() {
+  await loadCuda()
   try {
     const data = await apiClient.get<{items:Model[];config:Config;runtime_installed:boolean;last_inference:typeof lastInference.value}>('/api/local-models')
     items.value = data.items; installed.value = data.runtime_installed
@@ -41,25 +53,39 @@ onUnmounted(() => { stopped = true; clearTimeout(timer) })
 </script>
 <template>
   <section class="local-models">
-    <h3>本地模型</h3><p class="subtle">默认 CPU。下载需要联网；推理只读取本地权重。文件校验通过不代表当前设备已完成推理验证。</p>
+    <h3>{{ t('本地模型', 'Local Models') }}</h3><p class="subtle">{{ t('默认 CPU。下载需要联网；推理只读取本地权重。文件校验通过不代表当前设备已完成推理验证。', 'CPU is the default. Downloads require network access; inference reads local weights only. File verification does not mean the current device passed inference validation.') }}</p>
     <p v-if="error" class="error-banner" role="alert">{{ error }}</p>
-    <p v-if="lastInference" class="subtle">最近实际运行：{{ lastInference.actual_device }} · 请求设备 {{ lastInference.requested_device }} · 推理 {{ lastInference.inference_seconds.toFixed(2) }} 秒</p>
+    <p v-if="lastInference" class="subtle">{{ t('最近实际运行：', 'Last actual run: ') }}{{ lastInference.actual_device || t('未开始推理', 'No inference yet') }} · {{ t('请求设备', 'requested device') }} {{ lastInference.requested_device }} · {{ t('推理', 'inference') }} {{ (lastInference.inference_seconds ?? lastInference.elapsed_seconds ?? 0).toFixed(2) }} {{ t('秒', 'sec') }} · {{ lastInference.status }} {{ lastInference.error_code || '' }}</p>
-    <p v-if="!installed" class="subtle">尚未安装模型运行环境。在项目根目录执行 <code>./backend/scripts/install-model-runtime.ps1</code>；CUDA 选装追加 <code>-Device cuda</code>。</p>
+    <p v-if="!installed" class="subtle">{{ t('尚未安装模型运行环境。在项目根目录执行', 'The model runtime is not installed. Run this from the project root:') }} <code>./backend/scripts/install-model-runtime.ps1</code>; {{ t('CUDA 选装追加', 'for optional CUDA, append') }} <code>-Device cuda</code>.</p>
+    <article class="item-card cuda-components" :aria-label="t('CUDA 运行组件', 'CUDA runtime components')">
+      <h4>{{ t('CUDA 运行组件（可选）', 'CUDA Runtime Components (Optional)') }}</h4>
+      <p class="subtle">{{ t('默认使用 CPU。需要 NVIDIA GPU 加速时下载此组件，约 3 GB，安装时还需要额外磁盘空间；不包含显卡驱动和模型权重。', 'CPU is used by default. Download this component for NVIDIA GPU acceleration. It is about 3 GB and needs extra installation space; drivers and model weights are not included.') }}</p>
+      <p v-if="cudaError" class="error-text" role="alert">{{ cudaError }} <button class="button-secondary" @click="loadCuda">{{ t('重新检查', 'Check again') }}</button></p>
+      <template v-if="cuda">
+        <p role="status">{{ cuda.stage }} {{ cuda.torch || '' }}</p>
+        <progress v-if="['checking','installing'].includes(cuda.status)" :aria-label="t('CUDA 组件安装进度', 'CUDA component installation progress')" />
+        <p v-if="cuda.error" class="error-text">{{ cuda.error }}</p>
+        <p v-if="!cuda.supported" class="subtle">{{ t('当前平台暂不支持页面安装，请使用对应平台的模型运行环境。', 'This platform does not support in-app installation. Use the model runtime for your platform.') }}</p>
+        <button v-else-if="cuda.status !== 'installed'" class="button-primary" :disabled="busy || ['checking','installing'].includes(cuda.status)" @click="installCuda">{{ cuda.status === 'installing' ? t('正在下载并安装…', 'Downloading and installing…') : ['failed','interrupted'].includes(cuda.status) ? t('重试安装 CUDA 组件', 'Retry CUDA installation') : t('下载并安装 CUDA 组件', 'Download and install CUDA components') }}</button>
+        <p v-if="cuda.status === 'installed'" class="subtle">{{ cuda.cuda_available ? t('组件已就绪。在下方选择 CUDA 并保存即可启用。', 'Components are ready. Select CUDA below and save to enable it.') : t('组件已安装，但当前未检测到可用 CUDA 设备，将回退 CPU。', 'Components are installed, but no CUDA device is available; CPU fallback will be used.') }}</p>
+        <p v-if="cuda.custom_interpreter" class="subtle">{{ t('当前后端设置了 APP_MODEL_PYTHON，优先使用指定环境；要使用页面安装的组件，请移除该覆盖并重启后端。', 'APP_MODEL_PYTHON is set and takes priority. Remove the override and restart the backend to use components installed from this page.') }}</p>
+      </template>
+    </article>
     <form v-if="config" @submit.prevent="save" @input="dirty = true" @change="dirty = true">
-      <div class="runtime-grid"><label>请求设备<select v-model="config.device" class="select"><option value="cpu">CPU（默认）</option><option value="cuda">CUDA（不可用则 CPU）</option></select></label>
+      <div class="runtime-grid"><label>{{ t('请求设备', 'Requested device') }}<select v-model="config.device" class="select"><option value="cpu">{{ t('CPU（默认）', 'CPU (default)') }}</option><option value="cuda">{{ t('CUDA（不可用则 CPU）', 'CUDA (CPU fallback)') }}</option></select></label>
-      <label>Embedding<select v-model="config.embedding_model" class="select"><option value="bekko">Bekko A8M</option><option value="granite">Granite 97M 多语言</option></select></label>
+      <label>Embedding<select v-model="config.embedding_model" class="select"><option value="bekko">Bekko A8M</option><option value="granite">Granite 97M {{ t('多语言', 'Multilingual') }}</option></select></label>
-      <label>CPU 线程<input v-model.number="config.cpu_threads" class="input" type="number" min="1" max="32" /></label>
+      <label>{{ t('CPU 线程', 'CPU threads') }}<input v-model.number="config.cpu_threads" class="input" type="number" min="1" max="32" /></label>
-      <label>内存预算 MiB<input v-model.number="config.memory_limit_mb" class="input" type="number" min="1024" max="131072" /></label>
+      <label>{{ t('内存预算 MiB', 'Memory budget MiB') }}<input v-model.number="config.memory_limit_mb" class="input" type="number" min="1024" max="131072" /></label>
-      <label>显存预算 MiB<input v-model.number="config.gpu_memory_limit_mb" class="input" type="number" min="512" max="65536" /></label></div>
+      <label>{{ t('显存预算 MiB', 'GPU memory budget MiB') }}<input v-model.number="config.gpu_memory_limit_mb" class="input" type="number" min="512" max="65536" /></label></div>
-      <p class="subtle">修改 Embedding 后需要重建索引。任务按预算串行运行，模型在任务结束后释放。</p><button class="button-primary" :disabled="busy || !dirty">保存运行设置</button>
+      <p class="subtle">{{ t('修改 Embedding 后需要重建索引。任务按预算串行运行，模型在任务结束后释放。', 'Changing the embedding model requires rebuilding the index. Jobs run serially within the resource budget, and models are released when each job finishes.') }}</p><button class="button-primary" :disabled="busy || !dirty">{{ t('保存运行设置', 'Save runtime settings') }}</button>
     </form>
-    <div class="model-grid"><article v-for="model in items" :key="model.key" class="item-card"><h4>{{ model.name }}</h4><p>{{ model.license }} · {{ labels[model.status] || model.status }}</p><small :title="model.revision">版本 {{ model.revision.slice(0,12) }}</small>
+    <div class="model-grid"><article v-for="model in items" :key="model.key" class="item-card"><h4>{{ model.name }}</h4><p>{{ model.license }} · {{ labels[model.status] || model.status }}</p><small :title="model.revision">{{ t('版本', 'Revision') }} {{ model.revision.slice(0,12) }}</small>
-      <p>{{ size(model.downloaded_bytes) }} / {{ size(model.total_bytes) }}</p><progress v-if="model.status === 'downloading' && model.total_bytes" :value="model.downloaded_bytes" :max="model.total_bytes" />
+      <p>{{ t('实际磁盘占用', 'Disk usage') }} {{ size(model.disk_bytes) }}</p><p>{{ size(model.downloaded_bytes) }} / {{ size(model.total_bytes) }}</p><progress v-if="model.status === 'downloading' && model.total_bytes" :value="model.downloaded_bytes" :max="model.total_bytes" />
       <p v-if="model.error_code" class="error-text">{{ model.error_code }}</p><div class="inline-actions">
-      <button v-if="model.status !== 'installed' && model.status !== 'downloading'" class="button-secondary" :disabled="busy" @click="act(() => apiClient.post(`/api/local-models/${model.key}/download`))">{{ model.status === 'not_installed' ? '下载模型' : '重试 / 续传' }}</button>
+      <button v-if="model.status !== 'installed' && model.status !== 'downloading'" class="button-secondary" :disabled="busy" @click="act(() => apiClient.post(`/api/local-models/${model.key}/download`))">{{ model.status === 'not_installed' ? t('下载模型', 'Download model') : t('重试 / 续传', 'Retry / Resume') }}</button>
-      <button v-if="model.status === 'downloading'" class="button-secondary" :disabled="busy" @click="act(() => apiClient.post(`/api/local-models/${model.key}/cancel`))">暂停</button>
+      <button v-if="model.status === 'downloading'" class="button-secondary" :disabled="busy" @click="act(() => apiClient.post(`/api/local-models/${model.key}/cancel`))">{{ t('暂停', 'Pause') }}</button>
-      <button v-if="model.status !== 'not_installed'" class="button-danger" :disabled="busy" @click="act(() => apiClient.delete(`/api/local-models/${model.key}`))">删除权重</button></div>
+      <button v-if="model.status !== 'not_installed'" class="button-danger" :disabled="busy" @click="act(() => apiClient.delete(`/api/local-models/${model.key}`))">{{ t('删除权重', 'Delete weights') }}</button></div>
-    </article></div><button class="button-secondary" @click="diagnostics">导出本次运行诊断</button><p class="subtle">诊断仅包含模型、设备、耗时和资源信息，不包含正文、音频和密钥。</p>
+    </article></div><button class="button-secondary" @click="diagnostics">{{ t('导出最近运行诊断', 'Export recent runtime diagnostics') }}</button><p class="subtle">{{ t('诊断仅包含模型、设备、耗时和资源信息，不包含正文、音频和密钥。', 'Diagnostics include only model, device, timing, and resource data. Note content, audio, and secrets are excluded.') }}</p>
   </section>
 </template>
 <style scoped>.local-models{display:grid;gap:16px}.runtime-grid,.model-grid{display:grid;grid-template-columns:repeat(auto-fit,minmax(220px,1fr));gap:12px}label{display:grid;gap:6px}.item-card{padding:16px}progress{width:100%}</style>

frontend/src/features/settings/ModelRoutingSettings.vue

@@ -4,14 +4,15 @@ import type { ModelBinding, ModelRoutingConfig, ModelRoutingResponse, ProviderCo
 import { getModelRouting, saveModelRouting } from '@/services/modelRoutingService'
 import { listProviders } from '@/services/providerService'
 import { ApiErrorClass } from '@/services/apiClient'
+import { t } from '@/i18n'
 
-const capabilities: Array<{ id: RoutingCapability; name: string; endpoint: string; placeholder: string; local: string }> = [
+const capabilities = computed<Array<{ id: RoutingCapability; name: string; endpoint: string; placeholder: string; local: string }>>(() => [
-  { id: 'embedding', name: '向量嵌入 · Embedding', endpoint: '/embeddings', placeholder: '例如 text-embedding-3-small', local: '本地支持 Bekko / Granite，安装权重后可离线运行。' },
+  { id: 'embedding', name: t('向量嵌入 · Embedding', 'Embedding'), endpoint: '/embeddings', placeholder: t('例如 text-embedding-3-small', 'For example, text-embedding-3-small'), local: t('本地支持 Bekko / Granite，安装权重后可离线运行。', 'Local Bekko / Granite can run offline after weights are installed.') },
-  { id: 'transcription', name: '语音转写 · Transcription', endpoint: '/audio/transcriptions', placeholder: '输入转写模型 ID', local: '本地采用 Qwen3-ASR 0.6B，默认 CPU。' },
+  { id: 'transcription', name: t('语音转写 · Transcription', 'Transcription'), endpoint: '/audio/transcriptions', placeholder: t('输入转写模型 ID', 'Enter a transcription model ID'), local: t('本地采用 Qwen3-ASR 0.6B，默认 CPU。', 'Local Qwen3-ASR 0.6B uses CPU by default.') },
-  { id: 'speaker_matching', name: '说话人匹配 · Speaker matching', endpoint: '/audio/speaker-matches', placeholder: '输入说话人匹配模型 ID', local: '本地采用 ERes2NetV2，比对结果是相似度。' },
+  { id: 'speaker_matching', name: t('说话人匹配 · Speaker matching', 'Speaker matching'), endpoint: '/audio/speaker-matches', placeholder: t('输入说话人匹配模型 ID', 'Enter a speaker matching model ID'), local: t('本地采用 ERes2NetV2，比对结果是相似度。', 'Local ERes2NetV2 returns a similarity score.') },
-]
+])
 type Draft = { provider_id: string; model: string; endpoint: string; dimensions: string | number }
-const drafts = reactive(Object.fromEntries(capabilities.map(item => [item.id, { provider_id: '', model: '', endpoint: item.endpoint, dimensions: '' }])) as Record<RoutingCapability, Draft>)
+const drafts = reactive(Object.fromEntries(capabilities.value.map(item => [item.id, { provider_id: '', model: '', endpoint: item.endpoint, dimensions: '' }])) as Record<RoutingCapability, Draft>)
 const providers = ref<ProviderConfig[]>([])
 const response = ref<ModelRoutingResponse | null>(null)
 const loading = ref(false)
@@ -26,7 +27,7 @@ const unavailable = computed(() => providers.value.filter(provider => !eligible(
 const localBackend = (capability: RoutingCapability) => response.value?.local_backends.find(item => item.capability === capability)
 const localLabel = (capability: RoutingCapability) => {
   const status = localBackend(capability)?.status
-  return status === 'ready' ? '已安装' : status === 'placeholder' ? '测试占位实现' : '未安装'
+  return status === 'ready' ? t('已安装', 'Installed') : status === 'placeholder' ? t('测试占位实现', 'Test placeholder') : t('未安装', 'Not installed')
 }
 const protocols = [
   { id: 'openai_chat', label: 'OpenAI Chat' }, { id: 'openai_compatible', label: 'OpenAI Compatible' },
@@ -35,7 +36,7 @@ const protocols = [
 
 function applyResponse(result: ModelRoutingResponse) {
   response.value = result
-  for (const item of capabilities) {
+  for (const item of capabilities.value) {
     const binding = result.config[item.id]
     Object.assign(drafts[item.id], { provider_id: binding?.provider_id ?? '', model: binding?.model ?? '', endpoint: binding?.endpoint ?? item.endpoint, dimensions: binding?.dimensions?.toString() ?? '' })
   }
@@ -53,7 +54,7 @@ async function load() {
     applyResponse(routing)
     conflict.value = false
   } catch (reason) {
-    if (active) error.value = `加载失败：${reason instanceof Error ? reason.message : '无法读取模型路由或提供商'}`
+    if (active) error.value = `${t('加载失败：', 'Load failed: ')}${reason instanceof Error ? reason.message : t('无法读取模型路由或提供商', 'Could not read model routes or providers')}`
   } finally { loading.value = false }
 }
 
@@ -64,20 +65,20 @@ function changeProvider(capability: RoutingCapability) {
   const draft = drafts[capability]
   draft.model = ''
   draft.dimensions = ''
-  draft.endpoint = capabilities.find(item => item.id === capability)!.endpoint
+  draft.endpoint = capabilities.value.find(item => item.id === capability)!.endpoint
   saved.value = false
 }
 
 function bindingFor(capability: RoutingCapability): ModelBinding | null {
   const draft = drafts[capability]
   if (!draft.provider_id) return null
-  if (!available.value.some(provider => provider.provider_id === draft.provider_id)) throw new Error('请选择已启用且协议可用的提供商，或切换到本地。')
+  if (!available.value.some(provider => provider.provider_id === draft.provider_id)) throw new Error(t('请选择已启用且协议可用的提供商，或切换到本地。', 'Select an enabled provider with a supported protocol, or switch to local.'))
-  if (!draft.model.trim()) throw new Error('请填写所选 API 的模型 ID。')
+  if (!draft.model.trim()) throw new Error(t('请填写所选 API 的模型 ID。', 'Enter the model ID for the selected API.'))
-  if (!/^\/[A-Za-z0-9_/-]+$/.test(draft.endpoint) || draft.endpoint.startsWith('//')) throw new Error('Endpoint 必须是以 / 开头的相对路径，只能包含字母、数字、下划线、连字符和 /。')
+  if (!/^\/[A-Za-z0-9_/-]+$/.test(draft.endpoint) || draft.endpoint.startsWith('//')) throw new Error(t('Endpoint 必须是以 / 开头的相对路径，只能包含字母、数字、下划线、连字符和 /。', 'Endpoint must be a relative path beginning with / and containing only letters, numbers, underscores, hyphens, and /.'))
   const binding: ModelBinding = { provider_id: draft.provider_id, model: draft.model.trim(), endpoint: draft.endpoint }
   if (capability === 'embedding') {
     const dimension = String(draft.dimensions).trim()
-    if (dimension && (!/^\d+$/.test(dimension) || !Number.isSafeInteger(Number(dimension)) || Number(dimension) < 1 || Number(dimension) > 16384)) throw new Error('嵌入维度必须为 1–16384 的整数，或留空使用 API 默认值。')
+    if (dimension && (!/^\d+$/.test(dimension) || !Number.isSafeInteger(Number(dimension)) || Number(dimension) < 1 || Number(dimension) > 16384)) throw new Error(t('嵌入维度必须为 1–16384 的整数，或留空使用 API 默认值。', 'Embedding dimensions must be an integer from 1 to 16384, or blank to use the API default.'))
     binding.dimensions = dimension ? Number(dimension) : null
   }
   return binding
@@ -99,47 +100,47 @@ async function save() {
     if (!active) return
     conflict.value = reason instanceof ApiErrorClass && /CONFLICT|VERSION|HTTP_409/i.test(reason.code)
     error.value = conflict.value
-      ? '配置版本冲突：其他窗口已修改路由。当前输入尚未保存，请重新加载最新配置后再编辑。'
+      ? t('配置版本冲突：其他窗口已修改路由。当前输入尚未保存，请重新加载最新配置后再编辑。', 'Configuration conflict: another window changed these routes. Your input is unsaved; reload the latest settings before editing.')
-      : `保存失败：${reason instanceof Error ? reason.message : '请重试'}`
+      : `${t('保存失败：', 'Save failed: ')}${reason instanceof Error ? reason.message : t('请重试', 'please retry')}`
   } finally { saving.value = false }
 }
 </script>
 
 <template>
   <section class="routing-settings" aria-labelledby="routing-title" :aria-busy="loading || saving">
-    <div><h2 id="routing-title">能力模型路由</h2><p class="subtle">向量嵌入、语音转写和说话人匹配分别选择提供商与模型，独立于默认聊天模型。API Key 在「模型提供商」中管理。</p></div>
+    <div><h2 id="routing-title">{{ t('能力模型路由', 'Capability model routing') }}</h2><p class="subtle">{{ t('向量嵌入、语音转写和说话人匹配分别选择提供商与模型，独立于默认聊天模型。API Key 在「模型提供商」中管理。', 'Choose providers and models separately for embeddings, transcription, and speaker matching. API keys are managed under Model Providers.') }}</p></div>
-    <p class="subtle">未选择提供商即使用本地模型。API 请求失败、配置不可用或响应无效时回退到本地；使用前请下载对应权重并安装运行环境。</p>
+    <p class="subtle">{{ t('未选择提供商即使用本地模型。API 请求失败、配置不可用或响应无效时回退到本地；使用前请下载对应权重并安装运行环境。', 'With no provider selected, the local model is used. Failed API requests, invalid settings, or invalid responses fall back to local. Download the required weights and runtime first.') }}</p>
-    <p v-if="loading" role="status">正在加载模型路由…</p>
+    <p v-if="loading" role="status">{{ t('正在加载模型路由…', 'Loading model routes…') }}</p>
     <div v-if="error" class="error-banner" role="alert">{{ error }}</div>
-    <div class="inline-actions"><button type="button" class="button-secondary" :disabled="loading || saving" @click="load">{{ conflict ? '放弃当前输入并加载最新配置' : response ? '重新加载（放弃未保存更改）' : '重试加载' }}</button><span v-if="response" class="subtle">配置版本 {{ response.config.version }}</span></div>
+    <div class="inline-actions"><button type="button" class="button-secondary" :disabled="loading || saving" @click="load">{{ conflict ? t('放弃当前输入并加载最新配置', 'Discard input and load latest settings') : response ? t('重新加载（放弃未保存更改）', 'Reload (discard unsaved changes)') : t('重试加载', 'Retry loading') }}</button><span v-if="response" class="subtle">{{ t('配置版本', 'Configuration version') }} {{ response.config.version }}</span></div>
     <form v-if="response" @submit.prevent="save" @input="saved = false" @change="saved = false">
       <fieldset :disabled="loading || saving || conflict">
         <article v-for="capability in capabilities" :key="capability.id" class="routing-card" :data-capability="capability.id">
           <h3>{{ capability.name }}</h3>
-          <p v-if="capability.id === 'embedding'" class="embedding-notice">保存配置或更换模型、接口后，请重建全部索引。配置成功不代表已有笔记的向量索引已更新；重建完成前可使用全文检索，混合检索会回退到全文检索。</p>
+          <p v-if="capability.id === 'embedding'" class="embedding-notice">{{ t('保存配置或更换模型、接口后，请重建全部索引。配置成功不代表已有笔记的向量索引已更新；重建完成前可使用全文检索，混合检索会回退到全文检索。', 'Rebuild all indexes after saving or changing the model or endpoint. Saving settings does not update existing note vectors. Full-text search remains available, and hybrid search falls back to it until rebuilding completes.') }}</p>
-          <div class="protocols" aria-label="协议可用性">
+          <div class="protocols" :aria-label="t('协议可用性', 'Protocol availability')">
-            <span v-for="protocol in protocols" :key="protocol.id" class="badge" :class="{ 'protocol-unavailable': !['openai_chat', 'openai_compatible'].includes(protocol.id) }">{{ protocol.label }}{{ ['openai_chat', 'openai_compatible'].includes(protocol.id) ? ' · 可用' : ' · 不可用' }}</span>
+            <span v-for="protocol in protocols" :key="protocol.id" class="badge" :class="{ 'protocol-unavailable': !['openai_chat', 'openai_compatible'].includes(protocol.id) }">{{ protocol.label }}{{ ['openai_chat', 'openai_compatible'].includes(protocol.id) ? t(' · 可用', ' · Available') : t(' · 不可用', ' · Unavailable') }}</span>
           </div>
-          <label class="field"><span>处理方式 / 提供商</span><select v-model="drafts[capability.id].provider_id" class="select" data-field="provider" @change="changeProvider(capability.id)">
+          <label class="field"><span>{{ t('处理方式 / 提供商', 'Processing / Provider') }}</span><select v-model="drafts[capability.id].provider_id" class="select" data-field="provider" @change="changeProvider(capability.id)">
-            <option value="">本地 · {{ localLabel(capability.id) }}</option>
+            <option value="">{{ t('本地', 'Local') }} · {{ localLabel(capability.id) }}</option>
             <option v-for="provider in available" :key="provider.provider_id" :value="provider.provider_id">{{ provider.name }} · {{ provider.provider_type }}</option>
-            <option v-for="provider in unavailable" :key="provider.provider_id" :value="provider.provider_id" disabled>{{ provider.name }} · {{ provider.enabled ? '协议不可用' : '未启用' }}</option>
+            <option v-for="provider in unavailable" :key="provider.provider_id" :value="provider.provider_id" disabled>{{ provider.name }} · {{ provider.enabled ? t('协议不可用', 'Protocol unavailable') : t('未启用', 'Disabled') }}</option>
-            <option v-if="drafts[capability.id].provider_id && !providers.some(provider => provider.provider_id === drafts[capability.id].provider_id)" :value="drafts[capability.id].provider_id" disabled>原提供商已不可用 · {{ drafts[capability.id].provider_id }}</option>
+            <option v-if="drafts[capability.id].provider_id && !providers.some(provider => provider.provider_id === drafts[capability.id].provider_id)" :value="drafts[capability.id].provider_id" disabled>{{ t('原提供商已不可用', 'Previous provider is unavailable') }} · {{ drafts[capability.id].provider_id }}</option>
           </select></label>
           <div v-if="drafts[capability.id].provider_id" class="routing-fields">
-            <label class="field"><span>模型 ID</span><input v-model="drafts[capability.id].model" class="input" data-field="model" :placeholder="capability.placeholder" maxlength="256" required /></label>
+            <label class="field"><span>{{ t('模型 ID', 'Model ID') }}</span><input v-model="drafts[capability.id].model" class="input" data-field="model" :placeholder="capability.placeholder" maxlength="256" required /></label>
-            <label class="field"><span>Endpoint（相对 Base URL）</span><input v-model="drafts[capability.id].endpoint" class="input" data-field="endpoint" :placeholder="capability.endpoint" maxlength="256" required /></label>
+            <label class="field"><span>{{ t('Endpoint（相对 Base URL）', 'Endpoint (relative to Base URL)') }}</span><input v-model="drafts[capability.id].endpoint" class="input" data-field="endpoint" :placeholder="capability.endpoint" maxlength="256" required /></label>
-            <label v-if="capability.id === 'embedding'" class="field"><span>向量维度（可选）</span><input v-model="drafts.embedding.dimensions" class="input" data-field="dimensions" type="number" min="1" max="16384" step="1" placeholder="留空使用 API 默认维度" /><small class="subtle">填写模型支持的 1–16384 整数维度，或留空使用 API 默认值。</small></label>
+            <label v-if="capability.id === 'embedding'" class="field"><span>{{ t('向量维度（可选）', 'Vector dimensions (optional)') }}</span><input v-model="drafts.embedding.dimensions" class="input" data-field="dimensions" type="number" min="1" max="16384" step="1" :placeholder="t('留空使用 API 默认维度', 'Blank uses the API default')" /><small class="subtle">{{ t('填写模型支持的 1–16384 整数维度，或留空使用 API 默认值。', 'Enter an integer from 1 to 16384 supported by the model, or leave blank for the API default.') }}</small></label>
           </div>
-          <p v-if="capability.id === 'speaker_matching'" class="subtle">说话人匹配使用本应用自定义 HTTP multipart 契约。该端点不是 OpenAI 标准接口；服务需实现对应的说话人匹配请求和响应。</p>
+          <p v-if="capability.id === 'speaker_matching'" class="subtle">{{ t('说话人匹配使用本应用自定义 HTTP multipart 契约。该端点不是 OpenAI 标准接口；服务需实现对应的说话人匹配请求和响应。', 'Speaker matching uses this app’s custom HTTP multipart contract. It is not an OpenAI-standard endpoint; the service must implement the corresponding request and response.') }}</p>
           <div class="local-status" :class="{ selected: !drafts[capability.id].provider_id }">
-            <strong>{{ drafts[capability.id].provider_id ? '本地回退状态' : '当前本地状态' }}</strong>
+            <strong>{{ drafts[capability.id].provider_id ? t('本地回退状态', 'Local fallback status') : t('当前本地状态', 'Current local status') }}</strong>
-            <p>{{ localBackend(capability.id)?.status === 'ready' ? '本地后端已就绪。' : capability.local }}</p>
+            <p>{{ localBackend(capability.id)?.status === 'ready' ? t('本地后端已就绪。', 'The local backend is ready.') : capability.local }}</p>
-            <p v-for="backend in response.local_backends.filter(item => item.capability === capability.id)" :key="backend.capability" class="subtle"><span class="badge">{{ backend.status === 'ready' ? '已就绪' : backend.status === 'placeholder' ? '占位实现' : '未安装 / 未接入' }}</span> {{ backend.message }}</p>
+            <p v-for="backend in response.local_backends.filter(item => item.capability === capability.id)" :key="backend.capability" class="subtle"><span class="badge">{{ backend.status === 'ready' ? t('已就绪', 'Ready') : backend.status === 'placeholder' ? t('占位实现', 'Placeholder') : t('未安装 / 未接入', 'Not installed / connected') }}</span> {{ backend.message }}</p>
           </div>
         </article>
       </fieldset>
-      <div class="inline-actions"><button type="submit" class="button-primary" :disabled="loading || saving || conflict">{{ saving ? '保存中…' : '保存模型路由' }}</button><span v-if="saved" role="status">模型路由已保存。</span></div>
+      <div class="inline-actions"><button type="submit" class="button-primary" :disabled="loading || saving || conflict">{{ saving ? t('保存中…', 'Saving…') : t('保存模型路由', 'Save model routes') }}</button><span v-if="saved" role="status">{{ t('模型路由已保存。', 'Model routes saved.') }}</span></div>
     </form>
   </section>
 </template>

frontend/src/features/settings/ProviderForm.spec.ts

@@ -5,8 +5,11 @@ import type { ProviderConfig, ProviderPreset } from '@/contracts'
 import * as service from '@/services/providerService'
 import ProviderForm from './ProviderForm.vue'
 import ProviderPresetSelector from './ProviderPresetSelector.vue'
+import RequestJsonEditor from './RequestJsonEditor.vue'
+import { apiClient } from '@/services/apiClient'
 
 vi.mock('@/services/providerService', () => ({ listProviderPresets: vi.fn(), getCredentialStatus: vi.fn(), putCredential: vi.fn(), createProvider: vi.fn(), updateProvider: vi.fn() }))
+vi.mock('@/services/apiClient', () => ({ apiClient: { post: vi.fn() } }))
 const presets: ProviderPreset[] = [
   { preset_id: 'deepseek', name: 'DeepSeek', provider_type: 'openai_compatible', base_url: 'https://deepseek.example.test', default_credential_id: 'shared-deepseek', requires_credential: true, logo_id: 'deepseek' },
   { preset_id: 'qwen', name: '通义千问', provider_type: 'openai_compatible', base_url: 'https://qwen.example.test', default_credential_id: 'shared-qwen', requires_credential: true, logo_id: 'qwen' },
@@ -30,6 +33,21 @@ beforeEach(() => {
 afterEach(() => { wrappers.splice(0).forEach(wrapper => wrapper.unmount()) })
 
 describe('ProviderForm', () => {
+  it('invalidates a pending inference result when JSON becomes invalid', async () => {
+    const wrapper = await render(existing)
+    let finish!: (value: {message: string}) => void
+    vi.mocked(apiClient.post).mockReturnValue(new Promise(resolve => { finish = resolve }))
+    const probe = wrapper.findAll('button').find(button => button.text() === '发送测试推理请求')!
+    await probe.trigger('click')
+    expect(apiClient.post).toHaveBeenCalledWith('/api/providers/request-probe', expect.objectContaining({stream:true}))
+    wrapper.getComponent(RequestJsonEditor).vm.$emit('valid', false)
+    await flushPromises()
+    finish({message:'旧配置验证通过'})
+    await flushPromises()
+    expect(wrapper.text()).not.toContain('旧配置验证通过')
+    expect(probe.attributes('disabled')).toBeDefined()
+  })
+
   it('filters compact preset chips and resolves bundled logos', async () => {
     const wrapper = await render()
     await wrapper.get('#provider-search').setValue('通义')

frontend/src/features/settings/ProviderForm.vue

@@ -1,10 +1,11 @@
 <script setup lang="ts">
-import { computed, nextTick, onBeforeUnmount, onMounted, reactive, ref } from 'vue'
+import { computed, watch, nextTick, onBeforeUnmount, onMounted, reactive, ref } from 'vue'
 import type { ModelInfo, ProviderConfig, ProviderPreset, ProviderType, RequestOverride } from '@/contracts'
 import * as service from '@/services/providerService'
 import ProviderPresetSelector from './ProviderPresetSelector.vue'
 import RequestJsonEditor from './RequestJsonEditor.vue'
 import { apiClient } from '@/services/apiClient'
+import { t } from '@/i18n'
 
 const props = defineProps<{ provider?: ProviderConfig; models?: ModelInfo[] }>()
 const emit = defineEmits<{ close: []; saved: [provider: ProviderConfig] }>()
@@ -27,16 +28,39 @@ const error = ref('')
 const requestOverrides = ref<RequestOverride[]>(JSON.parse(JSON.stringify(props.provider?.request_overrides || [])))
 const requestJsonValid = ref(true)
 const requestPreview = ref('')
+const probeResult = ref('')
+const probing = ref(false)
+const previewCapability = ref('chat')
+const previewStream = ref(true)
+let draftGeneration = 0
+watch([form, requestOverrides, requestJsonValid, apiKey, previewStream, previewCapability], () => { draftGeneration++; requestPreview.value = ''; probeResult.value = '' }, {deep:true, flush:'sync'})
 async function previewRequest() {
+  const generation = draftGeneration
   error.value = ''
   try {
-    if (!requestJsonValid.value) throw new Error('请先修正 JSON。')
+    if (!requestJsonValid.value) throw new Error(t('请先修正 JSON。', 'Fix the JSON first.'))
     const response = await apiClient.post<{body:Record<string,unknown>}>('/api/providers/request-preview', {
-      provider: {provider_type:form.provider_type,name:form.name || '预览',base_url:form.base_url || null,
+      provider: {provider_type:form.provider_type,name:form.name || t('预览', 'Preview'),base_url:form.base_url || null,
-        default_model:form.default_model || null,request_overrides:requestOverrides.value}, stream:true,
+        default_model:form.default_model || null,request_overrides:requestOverrides.value}, stream:previewStream.value, capability:previewCapability.value,
     })
-    requestPreview.value = JSON.stringify(response.body, null, 2)
+    if (active && generation === draftGeneration) requestPreview.value = JSON.stringify(response.body, null, 2)
-  } catch(e) { error.value = (e as Error).message }
+  } catch(e) { if (active && generation === draftGeneration) error.value = (e as Error).message }
+}
+async function probeRequest() {
+  if (probing.value) return
+  error.value = ''; probeResult.value = ''; probing.value = true
+  const generation = draftGeneration
+  try {
+    if (!requestJsonValid.value) throw new Error(t('请先修正 JSON。', 'Fix the JSON first.'))
+    if (apiKey.value.trim()) throw new Error(t('请先保存新的 API Key，再进行推理验证。', 'Save the new API key before testing inference.'))
+    const result = await apiClient.post<{message:string}>('/api/providers/request-probe', {
+      provider: {provider_type:form.provider_type,name:form.name || t('推理验证', 'Inference test'),base_url:form.base_url || null,
+        default_model:form.default_model || null,request_overrides:JSON.parse(JSON.stringify(requestOverrides.value)),
+        credential_id:configured.value ? credentialId.value : null}, stream:previewStream.value,
+    })
+    if (active && generation === draftGeneration) probeResult.value = result.message
+  } catch(e) { if (active && generation === draftGeneration) error.value = (e as Error).message }
+  finally { probing.value = false }
 }
 const contextChanged = ref(false)
 const dialog = ref<HTMLElement>()
@@ -52,7 +76,7 @@ async function loadPresets() {
   try {
     presets.value = await service.listProviderPresets()
     if (!contextChanged.value) form.preset_id = presets.value.find(preset => preset.provider_type === props.provider?.provider_type && preset.base_url === props.provider?.base_url)?.preset_id ?? ''
-  } catch { presetsError.value = '预设加载失败，请重试，或填写自定义服务。' }
+  } catch { presetsError.value = t('预设加载失败，请重试，或填写自定义服务。', 'Preset loading failed. Retry or enter a custom service.') }
   finally { presetsLoading.value = false }
 }
 
@@ -65,7 +89,7 @@ onMounted(async () => {
       const result = await service.getCredentialStatus(credentialId.value)
       if (active && generation === credentialGeneration) configured.value = result
     } catch {
-      if (active && generation === credentialGeneration) credentialError.value = '无法检查已保存的凭据。可输入新密钥，或关闭后重试。'
+      if (active && generation === credentialGeneration) credentialError.value = t('无法检查已保存的凭据。可输入新密钥，或关闭后重试。', 'Could not check the saved credential. Enter a new key or close and retry.')
     } finally {
       if (generation === credentialGeneration) credentialLoading.value = false
     }
@@ -125,9 +149,9 @@ async function save() {
   error.value = ''
   saving.value = true
   try {
-    if (!form.name.trim() || !form.base_url.trim()) throw new Error('请填写名称和 Base URL。')
+    if (!form.name.trim() || !form.base_url.trim()) throw new Error(t('请填写名称和 Base URL。', 'Enter a name and Base URL.'))
-    if (!requestJsonValid.value) throw new Error('请先修正自定义请求 JSON。')
+    if (!requestJsonValid.value) throw new Error(t('请先修正自定义请求 JSON。', 'Fix the custom request JSON first.'))
-    if (selectedPreset.value?.requires_credential && !apiKey.value.trim() && !configured.value) throw new Error('请输入 API Key。密钥将由后端加密保存。')
+    if (selectedPreset.value?.requires_credential && !apiKey.value.trim() && !configured.value) throw new Error(t('请输入 API Key。密钥将由后端加密保存。', 'Enter an API key. It will be encrypted by the backend.'))
     // Snapshot before awaiting: closing/unmounting must never create a provider with a changed draft.
     const data = { provider_type: form.provider_type, name: form.name.trim(), base_url: form.base_url.trim() || undefined, default_model: form.default_model.trim(), enabled: form.enabled, capabilities: {}, has_credential: false, request_overrides: requestOverrides.value }
     if (apiKey.value.trim()) {
@@ -148,7 +172,7 @@ async function save() {
       : await service.createProvider({ ...data, credential_id: reference })
     if (active) { emit('saved', saved); close() }
   } catch (reason) {
-    if (active) error.value = reason instanceof Error ? reason.message : 'Provider 保存失败，请重试。'
+    if (active) error.value = reason instanceof Error ? reason.message : t('Provider 保存失败，请重试。', 'Provider save failed. Please retry.')
   } finally { apiKey.value = ''; saving.value = false }
 }
 </script>
@@ -156,29 +180,32 @@ async function save() {
 <template>
   <div class="modal-backdrop provider-backdrop" @click.self="close" @keydown="handleKeydown">
     <div ref="dialog" class="modal provider-modal" role="dialog" aria-modal="true" aria-labelledby="provider-form-title" :aria-busy="saving">
-      <div class="form-heading"><h2 id="provider-form-title">{{ provider ? '编辑 Provider' : '新增 Provider' }}</h2><button type="button" class="button-secondary" aria-label="关闭提供商表单" @click="close">关闭</button></div>
+      <div class="form-heading"><h2 id="provider-form-title">{{ provider ? t('编辑 Provider', 'Edit Provider') : t('新增 Provider', 'Add Provider') }}</h2><button type="button" class="button-secondary" :aria-label="t('关闭提供商表单', 'Close provider form')" @click="close">{{ t('关闭', 'Close') }}</button></div>
-      <p v-if="presetsLoading" class="subtle" role="status">正在加载提供商预设…</p>
+      <p v-if="presetsLoading" class="subtle" role="status">{{ t('正在加载提供商预设…', 'Loading provider presets…') }}</p>
-      <div v-if="presetsError" class="error-banner" role="alert">{{ presetsError }} <button type="button" class="button-secondary" :disabled="presetsLoading || saving" @click="loadPresets">重试</button></div>
+      <div v-if="presetsError" class="error-banner" role="alert">{{ presetsError }} <button type="button" class="button-secondary" :disabled="presetsLoading || saving" @click="loadPresets">{{ t('重试', 'Retry') }}</button></div>
       <form @submit.prevent="save" @input="requestPreview = ''" @change="requestPreview = ''">
         <fieldset :disabled="saving">
           <ProviderPresetSelector :presets="presets" :model-value="form.preset_id" @update:model-value="applyPreset" />
           <p v-if="selectedPreset?.description" class="subtle">{{ selectedPreset.description }}</p>
           <div class="form-grid">
-            <label class="field"><span>接入协议</span><select v-model="form.provider_type" class="select" data-field="protocol" @change="changeConnection"><option value="openai_compatible">OpenAI Compatible</option><option value="openai_chat">OpenAI Chat</option><option value="openai_responses">OpenAI Responses</option><option value="anthropic_messages">Anthropic Messages</option><option value="ollama">Ollama</option></select></label>
+            <label class="field"><span>{{ t('接入协议', 'Protocol') }}</span><select v-model="form.provider_type" class="select" data-field="protocol" @change="changeConnection"><option value="openai_compatible">OpenAI Compatible</option><option value="openai_chat">OpenAI Chat</option><option value="openai_responses">OpenAI Responses</option><option value="anthropic_messages">Anthropic Messages</option><option value="ollama">Ollama</option></select></label>
-            <label class="field"><span>名称</span><input v-model="form.name" class="input" data-field="name" required /></label>
+            <label class="field"><span>{{ t('名称', 'Name') }}</span><input v-model="form.name" class="input" data-field="name" required /></label>
             <label class="field wide"><span>Base URL</span><input v-model="form.base_url" class="input" data-field="base-url" placeholder="https://api.example.com/v1" required @change="changeConnection" /></label>
-            <label class="field wide"><span>API Key</span><input v-model="apiKey" class="input" type="password" autocomplete="new-password" spellcheck="false" :placeholder="configured ? '已配置，留空表示不修改' : '请输入 API Key（无鉴权服务可留空）'" /><small class="subtle">密钥由本地 AI Core 加密保存；提供商配置仅保存独立的凭据引用。</small></label>
+            <label class="field wide"><span>API Key</span><input v-model="apiKey" class="input" type="password" autocomplete="new-password" spellcheck="false" :placeholder="configured ? t('已配置，留空表示不修改', 'Configured; leave blank to keep it') : t('请输入 API Key（无鉴权服务可留空）', 'Enter an API key (optional for unauthenticated services)')" /><small class="subtle">{{ t('密钥由本地 AI Core 加密保存；提供商配置仅保存独立的凭据引用。', 'The local AI Core encrypts the key; provider settings store only its credential reference.') }}</small></label>
-            <p v-if="credentialLoading" class="subtle wide" role="status">正在检查凭据状态…</p>
+            <p v-if="credentialLoading" class="subtle wide" role="status">{{ t('正在检查凭据状态…', 'Checking credential status…') }}</p>
             <p v-if="credentialError" class="error-text wide" role="alert">{{ credentialError }}</p>
-            <label class="field wide"><span>默认聊天模型</span><input v-model="form.default_model" class="input" data-field="model" list="provider-model-options" placeholder="输入模型 ID，或保存后获取模型列表" /><datalist id="provider-model-options"><option v-for="model in modelOptions" :key="model.model_id" :value="model.model_id">{{ model.name }}</option></datalist></label>
+            <label class="field wide"><span>{{ t('默认聊天模型', 'Default chat model') }}</span><input v-model="form.default_model" class="input" data-field="model" list="provider-model-options" :placeholder="t('输入模型 ID，或保存后获取模型列表', 'Enter a model ID, or save to fetch the model list')" /><datalist id="provider-model-options"><option v-for="model in modelOptions" :key="model.model_id" :value="model.model_id">{{ model.name }}</option></datalist></label>
           </div>
-          <label class="inline-actions"><input v-model="form.enabled" type="checkbox" /> 启用</label>
+          <label class="inline-actions"><input v-model="form.enabled" type="checkbox" /> {{ t('启用', 'Enabled') }}</label>
           <RequestJsonEditor v-model="requestOverrides" @valid="requestJsonValid = $event" />
-          <button type="button" class="button-secondary" @click="previewRequest">预览最终流式请求（隐藏正文）</button>
+          <div class="inline-actions"><label>{{ t('预览能力', 'Preview capability') }}<select v-model="previewCapability" class="select"><option value="chat">{{ t('聊天', 'Chat') }}</option><option value="embedding">Embedding</option><option value="transcription">{{ t('转写', 'Transcription') }}</option><option value="speaker_matching">{{ t('声纹', 'Speaker') }}</option></select></label><label><input v-model="previewStream" type="checkbox" />{{ t('流式聊天', 'Streaming chat') }}</label></div>
+          <button type="button" class="button-secondary" @click="previewRequest">{{ t('预览最终请求（隐藏正文）', 'Preview final request (content hidden)') }}</button>
+          <button v-if="previewCapability === 'chat'" type="button" class="button-secondary" :disabled="probing || credentialLoading || !requestJsonValid" @click="probeRequest">{{ probing ? t('推理验证中…', 'Testing inference…') : t('发送测试推理请求', 'Send test inference request') }}</button>
+          <p class="subtle">{{ t('推理验证会向当前模型发送固定短消息，并计入实际用量。媒体参数请通过真实转写或声纹操作验证。', 'The inference test sends a fixed short message to the current model and counts toward usage. Validate media parameters through an actual transcription or speaker operation.') }}</p><p v-if="probeResult" role="status">{{ probeResult }}</p>
           <pre v-if="requestPreview" class="request-preview">{{ requestPreview }}</pre>
         </fieldset>
         <div v-if="error" class="error-banner" role="alert">{{ error }}</div>
-        <div class="inline-actions form-footer"><button class="button-primary" type="submit" :disabled="saving || credentialLoading">{{ saving ? '保存中…' : '保存提供商' }}</button><button type="button" class="button-secondary" @click="close">取消</button></div>
+        <div class="inline-actions form-footer"><button class="button-primary" type="submit" :disabled="saving || credentialLoading">{{ saving ? t('保存中…', 'Saving…') : t('保存提供商', 'Save provider') }}</button><button type="button" class="button-secondary" @click="close">{{ t('取消', 'Cancel') }}</button></div>
       </form>
     </div>
   </div>

frontend/src/features/settings/ProviderPresetSelector.vue

@@ -2,6 +2,7 @@
 import { computed, ref } from 'vue'
 import type { ProviderPreset } from '@/contracts'
 import ProviderLogo from './ProviderLogo.vue'
+import { t } from '@/i18n'
 
 const props = defineProps<{ presets: ProviderPreset[]; modelValue: string }>()
 const emit = defineEmits<{ 'update:modelValue': [value: string] }>()
@@ -15,14 +16,14 @@ const filtered = computed(() => {
 
 <template>
   <div class="preset-selector">
-    <label class="field" for="provider-search"><span>提供商预设</span><input id="provider-search" v-model="search" class="input" type="search" placeholder="搜索提供商，例如 通义千问 / DeepSeek" /></label>
+    <label class="field" for="provider-search"><span>{{ t('提供商预设', 'Provider presets') }}</span><input id="provider-search" v-model="search" class="input" type="search" :placeholder="t('搜索提供商，例如 通义千问 / DeepSeek', 'Search providers, such as Qwen / DeepSeek')" /></label>
-    <div class="preset-grid" role="group" aria-label="提供商预设">
+    <div class="preset-grid" role="group" :aria-label="t('提供商预设', 'Provider presets')">
-      <button type="button" class="preset-chip" :class="{ selected: !modelValue }" :aria-pressed="!modelValue" @click="emit('update:modelValue', '')"><ProviderLogo /><span>自定义</span></button>
+      <button type="button" class="preset-chip" :class="{ selected: !modelValue }" :aria-pressed="!modelValue" @click="emit('update:modelValue', '')"><ProviderLogo /><span>{{ t('自定义', 'Custom') }}</span></button>
       <button v-for="preset in filtered" :key="preset.preset_id" type="button" class="preset-chip" :class="{ selected: modelValue === preset.preset_id }" :aria-pressed="modelValue === preset.preset_id" :title="preset.description || preset.name" :data-preset="preset.preset_id" @click="emit('update:modelValue', preset.preset_id)">
         <ProviderLogo :logo-id="preset.logo_id || preset.preset_id" /><span>{{ preset.name }}</span>
       </button>
     </div>
-    <p v-if="search && !filtered.length" class="subtle" role="status">没有匹配的预设，可以使用自定义服务。</p>
+    <p v-if="search && !filtered.length" class="subtle" role="status">{{ t('没有匹配的预设，可以使用自定义服务。', 'No matching preset. You can use a custom service.') }}</p>
   </div>
 </template>
 

frontend/src/features/settings/RequestJsonEditor.spec.ts

@@ -1,7 +1,10 @@
 // @vitest-environment happy-dom
-import { mount } from '@vue/test-utils'
+import { flushPromises, mount } from '@vue/test-utils'
-import { expect, it } from 'vitest'
+import { expect, it, vi } from 'vitest'
 import RequestJsonEditor from './RequestJsonEditor.vue'
+import { apiClient } from '@/services/apiClient'
+
+vi.mock('@/services/apiClient', () => ({apiClient:{post:vi.fn()}}))
 
 it('validates object JSON and prevents host-owned fields from being saved', async () => {
   const wrapper = mount(RequestJsonEditor, {props: {modelValue: []}})
@@ -18,3 +21,47 @@ it('validates object JSON and prevents host-owned fields from being saved', asyn
   expect(wrapper.emitted('valid')?.at(-1)).toEqual([false])
   wrapper.unmount()
 })
+
+it('ignores an imported configuration that finishes after a newer edit', async () => {
+  let finish!: (value: {request_overrides: unknown[]}) => void
+  vi.mocked(apiClient.post).mockReturnValue(new Promise(resolve => { finish = resolve }))
+  const wrapper = mount(RequestJsonEditor, {props:{modelValue:[]}})
+  const input = wrapper.get('input[type="file"]')
+  const file = new File(['{"version":1,"request_overrides":[]}'], 'rules.json', {type:'application/json'})
+  Object.defineProperty(input.element, 'files', {value:[file], configurable:true})
+  await input.trigger('change')
+  await wrapper.findAll('button').find(button => button.text() === '添加请求规则')!.trigger('click')
+  finish({request_overrides:[{capability:'embedding',body:{dimensions:384}}]})
+  await Promise.resolve(); await Promise.resolve()
+  expect(wrapper.findAll('textarea')).toHaveLength(1)
+  expect(wrapper.get('textarea').element.value).toBe('{}')
+  wrapper.unmount()
+})
+
+it('ignores an old import failure after a newer edit', async () => {
+  let fail!: (reason: Error) => void
+  vi.mocked(apiClient.post).mockReturnValue(new Promise((_resolve, reject) => { fail = reject }))
+  const wrapper = mount(RequestJsonEditor, {props:{modelValue:[]}})
+  const input = wrapper.get('input[type="file"]')
+  Object.defineProperty(input.element, 'files', {value:[new File(['{}'], 'old.json')], configurable:true})
+  await input.trigger('change')
+  await wrapper.findAll('button').find(button => button.text() === '添加请求规则')!.trigger('click')
+  fail(new Error('旧导入失败'))
+  await flushPromises()
+  expect(wrapper.text()).not.toContain('旧导入失败')
+  expect(wrapper.findAll('textarea')).toHaveLength(1)
+  wrapper.unmount()
+})
+
+it('restores defaults even from an invalid draft and reflects replacement configurations', async () => {
+  const wrapper = mount(RequestJsonEditor, {props:{modelValue:[{capability:'chat', body:{enable_thinking:false}}]}})
+  await wrapper.get('textarea').setValue('{invalid')
+  expect(wrapper.emitted('valid')?.at(-1)).toEqual([false])
+  await wrapper.findAll('button').find(button => button.text() === '恢复默认请求')!.trigger('click')
+  expect(wrapper.findAll('textarea')).toHaveLength(0)
+  expect(wrapper.emitted('update:modelValue')?.at(-1)).toEqual([[]])
+  await wrapper.setProps({modelValue:[{capability:'embedding', body:{dimensions:384}}]})
+  expect(wrapper.get('textarea').element.value).toContain('384')
+  expect(wrapper.emitted('valid')?.at(-1)).toEqual([true])
+  wrapper.unmount()
+})

frontend/src/features/settings/RequestJsonEditor.vue

@@ -1,42 +1,85 @@
 <script setup lang="ts">
 import { ref, watch } from 'vue'
+import { apiClient } from '@/services/apiClient'
 import type { RequestOverride } from '@/contracts'
+import { t } from '@/i18n'
+import FilePicker from '@/components/common/FilePicker.vue'
 const props = defineProps<{modelValue: RequestOverride[]}>()
 const emit = defineEmits<{ 'update:modelValue': [value:RequestOverride[]]; valid:[value:boolean] }>()
+const transferError = ref('')
+let published = JSON.stringify(props.modelValue)
+let generation = 0
 const rules = ref(props.modelValue.map(rule => ({...rule, draft: JSON.stringify(rule.body, null, 2), error: ''})))
 const protectedFields = new Set(['model','messages','input','system','instructions','tools','tool_choice','parallel_tool_calls','functions','function_call','file','audio','reference_file','stream','previous_response_id','conversation','background','store'])
 function publish() {
+  generation++
   let valid = true
   const result: RequestOverride[] = []
   for (const rule of rules.value) {
     try {
       const body = JSON.parse(rule.draft)
-      if (!body || typeof body !== 'object' || Array.isArray(body)) throw new Error('顶层必须为 JSON 对象')
+      if (!body || typeof body !== 'object' || Array.isArray(body)) throw new Error(t('顶层必须为 JSON 对象', 'The top level must be a JSON object'))
       const conflicts = Object.keys(body).filter(key => protectedFields.has(key))
-      if (conflicts.length) throw new Error(`运行请求管理字段不可覆盖：${conflicts.join(', ')}`)
+      if (conflicts.length) throw new Error(`${t('运行请求管理字段不可覆盖：', 'Runtime-managed fields cannot be overridden: ')}${conflicts.join(', ')}`)
       rule.error = ''
       result.push({capability:rule.capability,model:rule.model || null,stream:rule.stream ?? null,body})
     } catch(e) { rule.error = (e as Error).message; valid = false }
   }
   emit('valid', valid)
-  if(valid) emit('update:modelValue', result)
+  if(valid) { published = JSON.stringify(result); emit('update:modelValue', result) }
 }
 function add() { rules.value.push({capability:'chat',model:null,stream:null,body:{},draft:'{}',error:''}); publish() }
 function format(index:number) { try { rules.value[index].draft = JSON.stringify(JSON.parse(rules.value[index].draft), null, 2); publish() } catch { publish() } }
-watch(() => props.modelValue.length, length => { if (length === 0 && rules.value.length && rules.value.every(r => !r.error)) rules.value = [] })
+watch(() => props.modelValue, value => {
+  if (JSON.stringify(value) !== published) {
+    generation++
+    rules.value = value.map(rule => ({...rule, draft: JSON.stringify(rule.body, null, 2), error: ''}))
+    published = JSON.stringify(value)
+    emit('valid', true)
+  }
+}, {deep: true})
+function reset() { rules.value = []; transferError.value = ''; publish() }
+async function importRules(file: File | null) {
+  if (!file) return
+  const current = ++generation
+  transferError.value = ''
+  try {
+    if (file.size > 1024 * 1024) throw new Error(t('配置文件不得超过 1 MiB', 'The configuration file must not exceed 1 MiB'))
+    const parsed = JSON.parse(await file.text())
+    const validated = await apiClient.post<{request_overrides: RequestOverride[]}>('/api/providers/request-rules/validate', parsed)
+    if (current !== generation) return
+    rules.value = validated.request_overrides.map(rule => ({...rule, draft: JSON.stringify(rule.body, null, 2), error: ''}))
+    publish()
+  } catch(e) { if (current === generation) transferError.value = (e as Error).message }
+}
+async function exportRules() {
+  transferError.value = ''
+  try {
+    publish()
+    if (rules.value.some(rule => rule.error)) throw new Error(t('请先修正 JSON', 'Fix the JSON first'))
+    const validated = await apiClient.post('/api/providers/request-rules/validate', {version:1, request_overrides:JSON.parse(published)})
+    const url = URL.createObjectURL(new Blob([JSON.stringify(validated, null, 2)], {type:'application/json'}))
+    const link = document.createElement('a'); link.href = url; link.download = 'model-request-rules.json'; link.click()
+    setTimeout(() => URL.revokeObjectURL(url), 1000)
+  } catch(e) { transferError.value = (e as Error).message }
+}
+
 </script>
 <template>
-  <details class="request-json"><summary>高级：自定义请求 JSON</summary>
+  <details class="request-json ui-disclosure"><summary>{{ t('高级：自定义请求 JSON', 'Advanced: Custom request JSON') }}</summary>
-    <p class="subtle">提供商通用规则先应用，再应用模型规则。对象递归合并，数组整体替换，null 作为实际值；删除键后恢复继承。密钥继续使用独立 API Key 配置。</p>
+    <p class="subtle">{{ t('提供商通用规则先应用，再应用模型规则。对象递归合并，数组整体替换，null 作为实际值；删除键后恢复继承。密钥继续使用独立 API Key 配置。', 'Provider-wide rules are applied before model rules. Objects merge recursively, arrays replace whole values, and null is kept as a value. Delete a key to inherit it again. API keys remain in the separate credential setting.') }}</p>
     <div v-for="(rule,index) in rules" :key="index" class="rule">
-      <div class="rule-selectors"><label>能力<select v-model="rule.capability" class="select" @change="publish"><option value="chat">聊天</option><option value="embedding">Embedding</option><option value="transcription">音频转写</option><option value="speaker_matching">声纹比对</option></select></label>
+      <div class="rule-selectors"><label>{{ t('能力', 'Capability') }}<select v-model="rule.capability" class="select" @change="publish"><option value="chat">{{ t('聊天', 'Chat') }}</option><option value="embedding">Embedding</option><option value="transcription">{{ t('音频转写', 'Transcription') }}</option><option value="speaker_matching">{{ t('声纹比对', 'Speaker matching') }}</option></select></label>
-      <label>模型<input v-model="rule.model" class="input" placeholder="留空：全部模型" @input="publish" /></label>
+      <label>{{ t('模型', 'Model') }}<input v-model="rule.model" class="input" :placeholder="t('留空：全部模型', 'Blank: all models')" @input="publish" /></label>
-      <label>请求模式<select v-model="rule.stream" class="select" @change="publish"><option :value="null">全部</option><option :value="true">仅流式</option><option :value="false">仅非流式</option></select></label></div>
+      <label>{{ t('请求模式', 'Request mode') }}<select v-model="rule.stream" class="select" @change="publish"><option :value="null">{{ t('全部', 'All') }}</option><option :value="true">{{ t('仅流式', 'Streaming only') }}</option><option :value="false">{{ t('仅非流式', 'Non-streaming only') }}</option></select></label></div>
-      <textarea v-model="rule.draft" class="input json-body" rows="6" aria-label="自定义请求 JSON" spellcheck="false" placeholder='{"stream_options":{"include_usage":true}}' @input="publish" />
+      <textarea v-model="rule.draft" class="input json-body" rows="6" :aria-label="t('自定义请求 JSON', 'Custom request JSON')" spellcheck="false" placeholder='{"stream_options":{"include_usage":true}}' @input="publish" />
       <p v-if="rule.error" class="error-text" role="alert">{{ rule.error }}</p>
-      <div class="inline-actions"><button type="button" class="button-secondary" @click="format(index)">格式化</button><button type="button" class="button-danger" @click="rules.splice(index,1); publish()">删除规则</button></div>
+      <div class="inline-actions"><button type="button" class="button-secondary" @click="format(index)">{{ t('格式化', 'Format') }}</button><button type="button" class="button-danger" @click="rules.splice(index,1); publish()">{{ t('删除规则', 'Delete rule') }}</button></div>
     </div>
-    <button type="button" class="button-secondary" @click="add">添加请求规则</button>
+    <button type="button" class="button-secondary" @click="add">{{ t('添加请求规则', 'Add request rule') }}</button>
+    <div class="transfer-actions"><div class="inline-actions"><button type="button" class="button-secondary" @click="reset">{{ t('恢复默认请求', 'Restore default request') }}</button><button type="button" class="button-secondary" @click="exportRules">{{ t('导出请求配置', 'Export request settings') }}</button></div><FilePicker :file="null" :label="t('导入请求配置', 'Import request settings')" :empty-label="t('选择 JSON 文件', 'Choose a JSON file')" accept=".json,application/json" @select="importRules" /></div>
+    <p v-if="transferError" class="error-text" role="alert">{{ transferError }}</p>
+    <p class="subtle">{{ t('导入替换当前请求规则，保存提供商后生效。导出仅包含请求规则，不包含凭据引用和 API Key。', 'Importing replaces the current request rules and takes effect after saving the provider. Exports contain rules only, without credential references or API keys.') }}</p>
   </details>
 </template>
-<style scoped>.request-json{display:grid;gap:12px}.rule{padding:12px;border:1px solid var(--border-color);border-radius:8px;margin:12px 0}.rule-selectors{display:flex;gap:8px;flex-wrap:wrap;margin-bottom:10px}.rule-selectors label{display:grid;gap:5px}.json-body{font-family:monospace;width:100%}</style>
+<style scoped>.request-json{display:grid;gap:12px}.rule{padding:12px;border:1px solid var(--color-border-default);border-radius:8px;margin:12px 0}.rule-selectors{display:flex;gap:8px;flex-wrap:wrap;margin-bottom:10px}.rule-selectors label{display:grid;gap:5px}.json-body{font-family:monospace;width:100%}.transfer-actions{display:flex;flex-wrap:wrap;align-items:center;gap:var(--space-sm);justify-content:space-between}</style>

frontend/src/features/settings/SettingsView.vue

@@ -1,5 +1,5 @@
 <script setup lang="ts">
-import { onMounted, ref } from 'vue'
+import { computed, onMounted, ref } from 'vue'
 import type { ProviderConfig } from '@/contracts'
 import ProviderForm from './ProviderForm.vue'
 import ProviderLogo from './ProviderLogo.vue'
@@ -9,12 +9,13 @@ import UsageCard from './UsageCard.vue'
 import { useProviderStore } from '@/stores/provider'
 import { useSettingsStore } from '@/stores/settings'
 import { useThemeStore } from '@/stores/theme'
+import { t } from '@/i18n'
 
 type Section = 'general' | 'editor' | 'providers' | 'index' | 'permissions' | 'ai-core'
-const sections: Array<{ id: Section; label: string }> = [
+const sections = computed<Array<{ id: Section; label: string }>>(() => [
-  { id: 'general', label: '通用' }, { id: 'editor', label: '编辑器' }, { id: 'providers', label: '模型提供商' },
+  { id: 'general', label: t('通用', 'General') }, { id: 'editor', label: t('编辑器', 'Editor') }, { id: 'providers', label: t('模型提供商', 'Model Providers') },
-  { id: 'index', label: '索引与模型' }, { id: 'permissions', label: '权限' }, { id: 'ai-core', label: 'AI Core 诊断' },
+  { id: 'index', label: t('索引与模型', 'Index and Models') }, { id: 'permissions', label: t('权限', 'Permissions') }, { id: 'ai-core', label: t('AI Core 诊断', 'AI Core Diagnostics') },
-]
+])
 const activeSection = ref<Section>('general')
 const settingsStore = useSettingsStore()
 const providerStore = useProviderStore()
@@ -47,66 +48,66 @@ async function providerSaved(provider: ProviderConfig) {
   if (provider.enabled) void providerStore.loadModels(provider.provider_id).catch(() => undefined)
 }
 
-async function removeProvider(provider: ProviderConfig) { if (!confirm(`确定删除 Provider“${provider.name}”吗？`)) return; try { await providerStore.deleteProvider(provider.provider_id) } catch (error) { providerAction.value = error instanceof Error ? error.message : '删除失败' } }
+async function removeProvider(provider: ProviderConfig) { if (!confirm(`${t('确定删除 Provider', 'Delete Provider')} “${provider.name}”?`)) return; try { await providerStore.deleteProvider(provider.provider_id) } catch (error) { providerAction.value = error instanceof Error ? error.message : t('删除失败', 'Delete failed') } }
-async function testProvider(provider: ProviderConfig) { testResults.value[provider.provider_id] = '测试中…'; const result = await providerStore.testProvider(provider.provider_id); testResults.value[provider.provider_id] = result.success ? `连接成功${result.latency_ms ? ` · ${result.latency_ms}ms` : ''}` : `连接失败：${result.error}` }
+async function testProvider(provider: ProviderConfig) { testResults.value[provider.provider_id] = t('测试中…', 'Testing…'); const result = await providerStore.testProvider(provider.provider_id); testResults.value[provider.provider_id] = result.success ? `${t('连接成功', 'Connection succeeded')}${result.latency_ms ? ` · ${result.latency_ms}ms` : ''}` : `${t('连接失败：', 'Connection failed: ')}${result.error}` }
 async function refreshModels(provider: ProviderConfig) { await providerStore.loadModels(provider.provider_id).catch(() => undefined) }
 async function chooseDefaultModel(provider: ProviderConfig, event: Event) {
   const defaultModel = (event.target as HTMLSelectElement).value
   try { await providerStore.updateProvider(provider.provider_id, { default_model: defaultModel }) }
-  catch (error) { providerAction.value = error instanceof Error ? error.message : '默认模型更新失败' }
+  catch (error) { providerAction.value = error instanceof Error ? error.message : t('默认模型更新失败', 'Failed to update the default model') }
 }
 </script>
 
 <template>
   <section class="feature-page settings-page">
-    <header class="feature-header"><div><h1>设置</h1><p>管理应用偏好、模型、索引、权限和本地 AI Core。</p></div></header>
+    <header class="feature-header"><div><h1>{{ t('设置', 'Settings') }}</h1><p>{{ t('管理应用偏好、模型、索引、权限和本地 AI Core。', 'Manage application preferences, models, indexing, permissions, and the local AI Core.') }}</p></div></header>
     <nav class="settings-nav"><button v-for="section in sections" :key="section.id" :class="{ active: activeSection === section.id }" @click="activeSection = section.id">{{ section.label }}</button></nav>
 
-    <div v-if="activeSection === 'general'" class="panel settings-section"><h2>通用</h2><label class="setting-row"><span><strong>恢复上次 Vault</strong><small>启动后自动打开最近使用的知识库</small></span><input v-model="settingsStore.restoreLastVault" type="checkbox" /></label><div class="setting-row"><span><strong>自动保存间隔</strong><small>编辑停止后等待多久写入文件</small></span><select v-model.number="settingsStore.autoSaveInterval" class="select short"><option :value="500">0.5 秒</option><option :value="1500">1.5 秒</option><option :value="3000">3 秒</option></select></div><div class="setting-row"><span><strong>界面语言</strong><small>当前阶段支持中文和英文入口</small></span><select v-model="settingsStore.language" class="select short"><option value="zh-CN">简体中文</option><option value="en">English</option></select></div><div class="setting-row"><span><strong>版本</strong><small>Desktop / AI Core</small></span><span>{{ settingsStore.appVersion }} / {{ settingsStore.aiCoreVersion }}</span></div></div>
+    <div v-if="activeSection === 'general'" class="panel settings-section"><h2>{{ t('通用', 'General') }}</h2><label class="setting-row"><span><strong>{{ t('恢复上次 Vault', 'Restore last Vault') }}</strong><small>{{ t('启动后自动打开最近使用的知识库', 'Open the most recently used knowledge base at startup') }}</small></span><input v-model="settingsStore.restoreLastVault" type="checkbox" /></label><div class="setting-row"><span><strong>{{ t('自动保存间隔', 'Autosave interval') }}</strong><small>{{ t('编辑停止后等待多久写入文件', 'How long to wait after editing before saving') }}</small></span><select v-model.number="settingsStore.autoSaveInterval" class="select short"><option :value="500">0.5 {{ t('秒', 'sec') }}</option><option :value="1500">1.5 {{ t('秒', 'sec') }}</option><option :value="3000">3 {{ t('秒', 'sec') }}</option></select></div><div class="setting-row"><span><strong>{{ t('界面语言', 'Interface language') }}</strong><small>{{ t('切换后立即应用到界面', 'Applied to the interface immediately') }}</small></span><select v-model="settingsStore.language" class="select short"><option value="zh-CN">简体中文</option><option value="en">English</option></select></div><div class="setting-row"><span><strong>{{ t('版本', 'Version') }}</strong><small>Desktop / AI Core</small></span><span>{{ settingsStore.appVersion }} / {{ settingsStore.aiCoreVersion }}</span></div></div>
 
-    <div v-else-if="activeSection === 'editor'" class="panel settings-section"><h2>编辑器</h2><div class="setting-row"><span><strong>默认模式</strong><small>新打开文件使用的编辑器模式</small></span><select v-model="settingsStore.defaultEditorMode" class="select short"><option value="wysiwyg">写作与预览</option><option value="source">Markdown 源码</option></select></div><div class="setting-row"><span><strong>字号</strong></span><input v-model.number="themeStore.fontEditorSize" class="input short" type="number" min="12" max="32" /></div><div class="setting-row"><span><strong>行高</strong></span><input v-model.number="themeStore.lineHeight" class="input short" type="number" min="1.2" max="2.4" step="0.1" /></div><div class="setting-row"><span><strong>行宽</strong><small>Markdown 预览最大字符宽度</small></span><input v-model.number="settingsStore.editorLineWidth" class="input short" type="number" min="40" max="140" /></div><label class="setting-row"><span><strong>拼写检查</strong></span><input v-model="settingsStore.spellCheck" type="checkbox" /></label></div>
+    <div v-else-if="activeSection === 'editor'" class="panel settings-section"><h2>{{ t('编辑器', 'Editor') }}</h2><div class="setting-row"><span><strong>{{ t('默认模式', 'Default mode') }}</strong><small>{{ t('新打开文件使用的编辑器模式', 'Editor mode used for newly opened files') }}</small></span><select v-model="settingsStore.defaultEditorMode" class="select short"><option value="wysiwyg">{{ t('写作与预览', 'Writing and preview') }}</option><option value="source">{{ t('Markdown 源码', 'Markdown source') }}</option></select></div><div class="setting-row"><span><strong>{{ t('字号', 'Font size') }}</strong></span><input v-model.number="themeStore.fontEditorSize" class="input short" type="number" min="12" max="32" /></div><div class="setting-row"><span><strong>{{ t('行高', 'Line height') }}</strong></span><input v-model.number="themeStore.lineHeight" class="input short" type="number" min="1.2" max="2.4" step="0.1" /></div><div class="setting-row"><span><strong>{{ t('行宽', 'Line width') }}</strong><small>{{ t('Markdown 预览最大字符宽度', 'Maximum character width for Markdown preview') }}</small></span><input v-model.number="settingsStore.editorLineWidth" class="input short" type="number" min="40" max="140" /></div><label class="setting-row"><span><strong>{{ t('拼写检查', 'Spell check') }}</strong><small>{{ t('在写作与源码编辑器中使用系统拼写检查', 'Use system spell checking in visual and source editors') }}</small></span><input v-model="settingsStore.spellCheck" type="checkbox" /></label></div>
 
     <div v-else-if="activeSection === 'providers'" class="settings-section">
       <div class="section-head">
-        <div><h2>模型提供商</h2><p class="subtle">选择国内外提供商预设，或配置自定义 API 与独立密钥。</p></div>
+        <div><h2>{{ t('模型提供商', 'Model Providers') }}</h2><p class="subtle">{{ t('选择国内外提供商预设，或配置自定义 API 与独立密钥。', 'Choose a provider preset or configure a custom API with separate credentials.') }}</p></div>
-        <button class="button-primary" @click="openProvider()">新增 Provider</button>
+        <button class="button-primary" @click="openProvider()">{{ t('新增 Provider', 'Add Provider') }}</button>
       </div>
       <div v-if="providerStore.error || providerAction" class="error-banner">{{ providerStore.error || providerAction }}</div>
       <LocalModelSettings />
       <UsageCard />
-      <p v-if="!providerStore.providers.length" class="subtle">{{ providerStore.isLoading ? '正在加载提供商…' : '尚无可用提供商，请添加真实 API 或本地 Ollama 配置。' }}</p>
+      <p v-if="!providerStore.providers.length" class="subtle">{{ providerStore.isLoading ? t('正在加载提供商…', 'Loading providers…') : t('尚无可用提供商，请添加真实 API 或本地 Ollama 配置。', 'No providers are available. Add a real API or local Ollama configuration.') }}</p>
       <div class="provider-list">
         <article v-for="provider in providerStore.providers" :key="provider.provider_id" class="item-card provider-card">
           <div class="provider-main">
             <div class="inline-actions"><ProviderLogo :logo-id="providerStore.presets.find(preset => preset.preset_id === presetIdFor(provider))?.logo_id || presetIdFor(provider)" /><strong>{{ provider.name }}</strong><span class="badge" :class="{ success: provider.enabled }">{{ provider.provider_type }}</span></div>
-            <p class="subtle">{{ provider.base_url || '本地内置' }} · 默认模型 {{ provider.default_model || '未设置' }}</p>
+            <p class="subtle">{{ provider.base_url || t('本地内置', 'Built in locally') }} · {{ t('默认模型', 'Default model') }} {{ provider.default_model || t('未设置', 'Not set') }}</p>
             <div class="tag-list"><span v-for="(_, capability) in provider.capabilities" :key="capability" class="badge">{{ capability }}</span></div>
             <div v-if="providerStore.modelsByProvider[provider.provider_id]?.length" class="model-picker">
-              <label :for="`default-model-${provider.provider_id}`">默认模型</label>
+              <label :for="`default-model-${provider.provider_id}`">{{ t('默认模型', 'Default model') }}</label>
               <select :id="`default-model-${provider.provider_id}`" class="select" :value="provider.default_model" @change="chooseDefaultModel(provider, $event)">
-                <option value="">未设置</option>
+                <option value="">{{ t('未设置', 'Not set') }}</option>
                 <option v-for="model in providerStore.modelsByProvider[provider.provider_id]" :key="model.model_id" :value="model.model_id">{{ model.name }}</option>
               </select>
-              <span class="subtle">已获取 {{ providerStore.modelsByProvider[provider.provider_id].length }} 个模型</span>
+              <span class="subtle">{{ t('已获取', 'Loaded') }} {{ providerStore.modelsByProvider[provider.provider_id].length }} {{ t('个模型', 'models') }}</span>
             </div>
-            <p v-if="providerStore.modelErrorsByProvider[provider.provider_id]" class="error-text">模型获取失败：{{ providerStore.modelErrorsByProvider[provider.provider_id] }}</p>
+            <p v-if="providerStore.modelErrorsByProvider[provider.provider_id]" class="error-text">{{ t('模型获取失败：', 'Failed to load models: ') }}{{ providerStore.modelErrorsByProvider[provider.provider_id] }}</p>
             <p v-if="testResults[provider.provider_id]" class="test-result">{{ testResults[provider.provider_id] }}</p>
           </div>
           <div class="inline-actions provider-actions">
-            <button class="button-secondary" :disabled="providerStore.modelLoadingByProvider[provider.provider_id]" @click="refreshModels(provider)">{{ providerStore.modelLoadingByProvider[provider.provider_id] ? '获取中…' : '刷新模型' }}</button>
+            <button class="button-secondary" :disabled="providerStore.modelLoadingByProvider[provider.provider_id]" @click="refreshModels(provider)">{{ providerStore.modelLoadingByProvider[provider.provider_id] ? t('获取中…', 'Loading…') : t('刷新模型', 'Refresh models') }}</button>
-            <button class="button-secondary" @click="testProvider(provider)">测试</button>
+            <button class="button-secondary" @click="testProvider(provider)">{{ t('测试', 'Test') }}</button>
-            <button class="button-secondary" @click="openProvider(provider)">编辑</button>
+            <button class="button-secondary" @click="openProvider(provider)">{{ t('编辑', 'Edit') }}</button>
-            <button class="button-danger" @click="removeProvider(provider)">删除</button>
+            <button class="button-danger" @click="removeProvider(provider)">{{ t('删除', 'Delete') }}</button>
           </div>
         </article>
       </div>
     </div>
 
-    <div v-else-if="activeSection === 'index'" class="panel settings-section"><h2>索引与模型</h2><div class="index-summary"><div><span class="badge" :class="{ success: settingsStore.indexStatus.status === 'idle', error: settingsStore.indexStatus.status === 'error' }">{{ settingsStore.indexStatus.status }}</span><p>待处理任务 {{ settingsStore.indexStatus.pending_jobs }}</p></div><div><strong>{{ settingsStore.indexStatus.total_notes ?? '未获取' }}</strong><small>笔记</small></div><div><strong>{{ settingsStore.indexStatus.total_blocks ?? '未获取' }}</strong><small>Block</small></div></div><div v-if="settingsStore.indexStatus.error" class="error-banner">{{ settingsStore.indexStatus.error }}</div><div class="inline-actions"><button class="button-primary" @click="settingsStore.rebuildIndex('full')">重建全部</button><span class="subtle">当前后端支持全量重建。</span></div><ModelRoutingSettings /></div>
+    <div v-else-if="activeSection === 'index'" class="panel settings-section"><h2>{{ t('索引与模型', 'Index and Models') }}</h2><div class="index-summary"><div><span class="badge" :class="{ success: settingsStore.indexStatus.status === 'idle', error: settingsStore.indexStatus.status === 'error' }">{{ settingsStore.indexStatus.status }}</span><p>{{ t('待处理任务', 'Pending jobs') }} {{ settingsStore.indexStatus.pending_jobs }}</p></div><div><strong>{{ settingsStore.indexStatus.total_notes ?? t('未获取', 'Unavailable') }}</strong><small>{{ t('笔记', 'Notes') }}</small></div><div><strong>{{ settingsStore.indexStatus.total_blocks ?? t('未获取', 'Unavailable') }}</strong><small>Block</small></div></div><div v-if="settingsStore.indexStatus.error" class="error-banner">{{ settingsStore.indexStatus.error }}</div><div class="inline-actions"><button class="button-primary" @click="settingsStore.rebuildIndex('full')">{{ t('重建全部', 'Rebuild all') }}</button><span class="subtle">{{ t('当前后端支持全量重建。', 'The current backend supports a full rebuild.') }}</span></div><ModelRoutingSettings /></div>
 
-    <div v-else-if="activeSection === 'permissions'" class="panel settings-section"><h2>权限策略</h2><p class="muted section-description">以下为后端当前生效的权限策略；全局策略编辑尚未开放，运行时按实际权限请求确认。</p><p v-if="!Object.keys(settingsStore.permissionPolicy).length" class="subtle">尚未获取权限策略，请检查后端连接并重新检测。</p><div class="permission-list"><div v-for="(policy, permission) in settingsStore.permissionPolicy" :key="permission" class="setting-row"><span><strong>{{ permission }}</strong></span><span>{{ policy === 'allow' ? '允许' : policy === 'confirm' ? '每次确认' : '拒绝' }}</span></div></div></div>
+    <div v-else-if="activeSection === 'permissions'" class="panel settings-section"><h2>{{ t('权限策略', 'Permission Policy') }}</h2><p class="muted section-description">{{ t('以下为后端当前生效的权限策略；全局策略编辑尚未开放，运行时按实际权限请求确认。', 'These policies are active in the backend. Global policy editing is not yet available; runtime requests are confirmed as needed.') }}</p><p v-if="!Object.keys(settingsStore.permissionPolicy).length" class="subtle">{{ t('尚未获取权限策略，请检查后端连接并重新检测。', 'Permission policy is unavailable. Check the backend connection and try again.') }}</p><div class="permission-list"><div v-for="(policy, permission) in settingsStore.permissionPolicy" :key="permission" class="setting-row"><span><strong>{{ permission }}</strong></span><span>{{ policy === 'allow' ? t('允许', 'Allow') : policy === 'confirm' ? t('每次确认', 'Confirm each time') : t('拒绝', 'Deny') }}</span></div></div></div>
 
-    <div v-else class="panel settings-section"><h2>AI Core 诊断</h2><div v-if="settingsStore.diagnosticsError" class="error-banner">{{ settingsStore.diagnosticsError }}</div><div class="diagnostic-grid"><div class="item-card"><span class="badge" :class="{ success: settingsStore.aiCoreStatus === 'running', error: settingsStore.aiCoreStatus === 'error' }">{{ settingsStore.aiCoreStatus }}</span><h3>AI Core 连接状态</h3><p class="subtle">AI Core 不可用时，Markdown 编辑仍可继续使用。</p></div><div class="item-card"><strong>{{ settingsStore.aiCoreAddress }}</strong><h3>开发 API 地址</h3><p class="subtle">正式桌面环境由 Sidecar Manager 动态提供。</p></div></div><div class="inline-actions diagnostic-actions"><button class="button-primary" @click="settingsStore.loadDiagnostics">重新检测</button><span class="subtle">当前 Web 端不支持重启后端进程，请在运行后端的终端中操作。</span></div></div>
+    <div v-else class="panel settings-section"><h2>{{ t('AI Core 诊断', 'AI Core Diagnostics') }}</h2><div v-if="settingsStore.diagnosticsError" class="error-banner">{{ settingsStore.diagnosticsError }}</div><div class="diagnostic-grid"><div class="item-card"><span class="badge" :class="{ success: settingsStore.aiCoreStatus === 'running', error: settingsStore.aiCoreStatus === 'error' }">{{ settingsStore.aiCoreStatus }}</span><h3>{{ t('AI Core 连接状态', 'AI Core connection') }}</h3><p class="subtle">{{ t('AI Core 不可用时，Markdown 编辑仍可继续使用。', 'Markdown editing remains available when AI Core is offline.') }}</p></div><div class="item-card"><strong>{{ settingsStore.aiCoreAddress }}</strong><h3>{{ t('开发 API 地址', 'Development API address') }}</h3><p class="subtle">{{ t('正式桌面环境由 Sidecar Manager 动态提供。', 'The desktop build will provide this through Sidecar Manager.') }}</p></div></div><div class="inline-actions diagnostic-actions"><button class="button-primary" @click="settingsStore.loadDiagnostics">{{ t('重新检测', 'Check again') }}</button><span class="subtle">{{ t('当前 Web 端不支持重启后端进程，请在运行后端的终端中操作。', 'The web build cannot restart the backend. Use the terminal running it.') }}</span></div></div>
 
     <ProviderForm v-if="showProviderForm" :provider="editingProvider" :models="editingProvider ? providerStore.modelsByProvider[editingProvider.provider_id] : []" @close="showProviderForm = false" @saved="providerSaved" />
   </section>

frontend/src/features/settings/UsageCard.vue

@@ -1,7 +1,8 @@
 <script setup lang="ts">
-import { onMounted, ref } from 'vue'
+import { computed, onMounted, ref } from 'vue'
 import { apiClient } from '@/services/apiClient'
-interface Usage {totals: Record<string,number|null>;coverage:Record<string,number>;request_count:number;complete_requests:number;cache_hit_rate:number|null;cache_covered_requests:number;options:{provider_id:string;model:string;source:string}[]}
+import { t } from '@/i18n'
+interface Usage {audio_request_count:number;audio_seconds:number|null;audio_covered_requests:number;totals: Record<string,number|null>;coverage:Record<string,number>;request_count:number;complete_requests:number;cache_hit_rate:number|null;cache_covered_requests:number;options:{provider_id:string;model:string;source:string}[]}
 const data = ref<Usage | null>(null)
 const period = ref('7')
 const provider = ref('')
@@ -11,7 +12,7 @@ const start = ref('')
 const end = ref('')
 const busy = ref(false)
 const error = ref('')
-const metrics: Record<string,string> = {input_tokens:'输入 Token',output_tokens:'输出 Token',total_tokens:'总 Token',cache_hit_tokens:'缓存命中',cache_miss_tokens:'缓存未命中',cache_write_tokens:'缓存写入',reasoning_tokens:'推理 Token'}
+const metrics = computed<Record<string,string>>(() => ({input_tokens:t('输入 Token','Input tokens'),output_tokens:t('输出 Token','Output tokens'),total_tokens:t('总 Token','Total tokens'),cache_hit_tokens:t('缓存命中','Cache hits'),cache_miss_tokens:t('缓存未命中','Cache misses'),cache_write_tokens:t('缓存写入','Cache writes'),reasoning_tokens:t('推理 Token','Reasoning tokens')}))
 async function load() {
   busy.value = true; error.value = ''
   try {
@@ -19,26 +20,27 @@ async function load() {
     const from = period.value === 'custom' ? new Date(start.value) : new Date(until)
     if (period.value === 'today') from.setHours(0,0,0,0)
     else if (period.value !== 'custom') from.setDate(from.getDate() - Number(period.value))
-    if (!Number.isFinite(from.getTime()) || !Number.isFinite(until.getTime()) || until <= from) throw new Error('请选择有效的开始与结束时间。')
+    if (!Number.isFinite(from.getTime()) || !Number.isFinite(until.getTime()) || until <= from) throw new Error(t('请选择有效的开始与结束时间。', 'Choose a valid start and end time.'))
     data.value = await apiClient.get<Usage>('/api/usage', {params: {start:from.toISOString(),end:until.toISOString(),provider_id:provider.value || undefined,model:model.value || undefined,source:source.value || undefined}})
   } catch(e) { error.value = (e as Error).message } finally { busy.value = false }
 }
 onMounted(load)
 </script>
 <template>
-  <section class="panel usage-card"><header><h3>Token 消耗情况</h3><button class="button-secondary" :disabled="busy" @click="load">{{ busy ? '加载中…' : '刷新统计' }}</button></header>
+  <section class="panel usage-card"><header><h3>{{ t('Token 消耗情况', 'Token Usage') }}</h3><button class="button-secondary" :disabled="busy" @click="load">{{ busy ? t('加载中…', 'Loading…') : t('刷新统计', 'Refresh') }}</button></header>
-    <div class="filters"><label>时间<select v-model="period" class="select" @change="period !== 'custom' && load()"><option value="today">今日</option><option value="7">近 7 天</option><option value="30">近 30 天</option><option value="custom">自定义</option></select></label>
+    <div class="filters"><label>{{ t('时间', 'Period') }}<select v-model="period" class="select" @change="period !== 'custom' && load()"><option value="today">{{ t('今日', 'Today') }}</option><option value="7">{{ t('近 7 天', 'Last 7 days') }}</option><option value="30">{{ t('近 30 天', 'Last 30 days') }}</option><option value="custom">{{ t('自定义', 'Custom') }}</option></select></label>
-      <label>提供商<select v-model="provider" class="select" @change="model = ''; load()"><option value="">全部</option><option v-for="id in [...new Set(data?.options.map(o => o.provider_id) || [])]" :key="id">{{ id }}</option></select></label>
+      <label>{{ t('提供商', 'Provider') }}<select v-model="provider" class="select" @change="model = ''; load()"><option value="">{{ t('全部', 'All') }}</option><option v-for="id in [...new Set(data?.options.map(o => o.provider_id) || [])]" :key="id">{{ id }}</option></select></label>
-      <label>模型<select v-model="model" class="select" @change="load"><option value="">全部</option><option v-for="id in [...new Set(data?.options.filter(o => !provider || o.provider_id === provider).map(o => o.model) || [])]" :key="id">{{ id }}</option></select></label>
+      <label>{{ t('模型', 'Model') }}<select v-model="model" class="select" @change="load"><option value="">{{ t('全部', 'All') }}</option><option v-for="id in [...new Set(data?.options.filter(o => !provider || o.provider_id === provider).map(o => o.model) || [])]" :key="id">{{ id }}</option></select></label>
-      <label>来源<select v-model="source" class="select" @change="load"><option value="">全部</option><option value="api">远程 API</option><option value="local">本地服务</option></select></label>
+      <label>{{ t('来源', 'Source') }}<select v-model="source" class="select" @change="load"><option value="">{{ t('全部', 'All') }}</option><option value="api">{{ t('远程 API', 'Remote API') }}</option><option value="local">{{ t('本地服务', 'Local service') }}</option></select></label>
     </div>
-    <div v-if="period === 'custom'" class="filters"><label>开始<input v-model="start" class="input" type="datetime-local" /></label><label>结束<input v-model="end" class="input" type="datetime-local" /></label><button class="button-secondary" @click="load">应用时间段</button></div>
+    <div v-if="period === 'custom'" class="filters"><label>{{ t('开始', 'Start') }}<input v-model="start" class="input" type="datetime-local" /></label><label>{{ t('结束', 'End') }}<input v-model="end" class="input" type="datetime-local" /></label><button class="button-secondary" @click="load">{{ t('应用时间段', 'Apply period') }}</button></div>
     <p v-if="error" class="error-banner" role="alert">{{ error }}</p>
-    <template v-if="data"><p v-if="!data.request_count" class="subtle">该时间段没有已记录的模型请求。</p>
+    <template v-if="data"><p v-if="!data.request_count" class="subtle">{{ t('该时间段没有已记录的模型请求。', 'No model requests were recorded during this period.') }}</p>
-      <div class="usage-grid"><div v-for="(label,key) in metrics" :key="key"><small>{{ label }}</small><strong>{{ data.totals[key] === null ? '未提供' : data.totals[key]?.toLocaleString() }}</strong><small>覆盖 {{ data.coverage[key] }} / {{ data.request_count }} 次</small></div>
+      <div class="usage-grid"><div v-for="(label,key) in metrics" :key="key"><small>{{ label }}</small><strong>{{ data.totals[key] === null ? t('未提供', 'Unavailable') : data.totals[key]?.toLocaleString() }}</strong><small>{{ t('覆盖', 'Coverage') }} {{ data.coverage[key] }} / {{ data.request_count }} {{ t('次', 'requests') }}</small></div>
-      <div><small>缓存命中率</small><strong>{{ data.cache_hit_rate === null ? '未提供' : `${(data.cache_hit_rate * 100).toFixed(1)}%` }}</strong><small>覆盖 {{ data.cache_covered_requests }} 次</small></div></div>
+      <div><small>{{ t('缓存命中率', 'Cache hit rate') }}</small><strong>{{ data.cache_hit_rate === null ? t('未提供', 'Unavailable') : `${(data.cache_hit_rate * 100).toFixed(1)}%` }}</strong><small>{{ t('覆盖', 'Coverage') }} {{ data.cache_covered_requests }}</small></div></div>
-      <p class="subtle">请求 {{ data.request_count }} 次，其中完整结束 {{ data.complete_requests }} 次。输入总量包含厂商已报告的缓存，推理 Token 不重复加入输出。</p>
+      <p class="subtle">{{ t('音频调用', 'Audio calls') }} {{ data.audio_request_count ?? 0 }} · {{ t('时长', 'Duration') }} {{ data.audio_seconds == null ? t('未提供', 'Unavailable') : `${data.audio_seconds.toFixed(2)} ${t('秒', 'sec')}` }} ({{ t('覆盖', 'coverage') }} {{ data.audio_covered_requests ?? 0 }}; {{ t('重试分别计数', 'retries counted separately') }})</p>
-    </template><p class="subtle">统计为本应用观测值，不是厂商账户账单。缺失指标显示“未提供”，历史未记录的数据不补估。</p>
+      <p class="subtle">{{ t('请求', 'Requests') }} {{ data.request_count }}, {{ t('其中完整结束', 'completed') }} {{ data.complete_requests }}. {{ t('输入总量包含厂商已报告的缓存，推理 Token 不重复加入输出。', 'Input totals include provider-reported cache tokens; reasoning tokens are not added to output twice.') }}</p>
+    </template><p class="subtle">{{ t('统计为本应用观测值，不是厂商账户账单。缺失指标显示“未提供”，历史未记录的数据不补估。', 'Statistics are application observations, not provider billing. Missing metrics stay unavailable and historical gaps are not estimated.') }}</p>
   </section>
 </template>
 <style scoped>.usage-card{display:grid;gap:16px;padding:20px}.usage-card header,.filters{display:flex;gap:12px;align-items:center;flex-wrap:wrap}.usage-card header{justify-content:space-between}.filters label{display:grid;gap:5px}.usage-grid{display:grid;grid-template-columns:repeat(auto-fit,minmax(130px,1fr));gap:16px}.usage-grid>div{display:grid;gap:8px}.usage-grid strong{font-size:22px}</style>

frontend/src/features/skills/SkillsView.vue

@@ -3,36 +3,37 @@ import { Lightning } from '@element-plus/icons-vue'
 import AppIcon from '@/components/common/AppIcon.vue'
 import { onMounted, ref } from 'vue'
 import { useSkillStore } from '@/stores/skill'
+import { t } from '@/i18n'
 
 const skillStore = useSkillStore()
 const actionError = ref('')
 onMounted(() => { void skillStore.loadSkills() })
 
 async function install() {
-  const path = prompt('请输入 Skill Package 路径')?.trim()
+  const path = prompt(t('请输入 Skill Package 路径', 'Enter the Skill package path'))?.trim()
   if (!path) return
-  try { await skillStore.installSkill(path) } catch (error) { actionError.value = error instanceof Error ? error.message : '安装失败' }
+  try { await skillStore.installSkill(path) } catch (error) { actionError.value = error instanceof Error ? error.message : t('安装失败', 'Installation failed') }
 }
 async function toggle(skillId: string, enabled: boolean) {
-  try { enabled ? await skillStore.disableSkill(skillId) : await skillStore.enableSkill(skillId) } catch (error) { actionError.value = error instanceof Error ? error.message : '状态更新失败' }
+  try { enabled ? await skillStore.disableSkill(skillId) : await skillStore.enableSkill(skillId) } catch (error) { actionError.value = error instanceof Error ? error.message : t('状态更新失败', 'Status update failed') }
 }
 async function uninstall(skillId: string, name: string) {
-  if (!confirm(`确定卸载 Skill“${name}”吗？`)) return
+  if (!confirm(`${t('确定卸载 Skill', 'Uninstall Skill')} “${name}”?`)) return
-  try { await skillStore.uninstallSkill(skillId) } catch (error) { actionError.value = error instanceof Error ? error.message : '卸载失败' }
+  try { await skillStore.uninstallSkill(skillId) } catch (error) { actionError.value = error instanceof Error ? error.message : t('卸载失败', 'Uninstall failed') }
 }
 </script>
 
 <template>
   <section class="feature-page">
-    <header class="feature-header"><div><h1>Skill 管理</h1><p>查看工作流使用的 Tool、权限、检索配置和模型要求。</p></div><button class="button-primary" @click="install">安装 Skill</button></header>
+    <header class="feature-header"><div><h1>{{ t('Skill 管理', 'Skill Management') }}</h1><p>{{ t('查看工作流使用的 Tool、权限、检索配置和模型要求。', 'Review the tools, permissions, retrieval settings, and model requirements used by workflows.') }}</p></div><button class="button-primary" @click="install">{{ t('安装 Skill', 'Install Skill') }}</button></header>
     <div v-if="skillStore.error || actionError" class="error-banner">{{ skillStore.error || actionError }}</div>
     <div v-if="skillStore.selectedSkill" class="panel detail-panel">
-      <div class="detail-head"><div><span class="badge" :class="{ success: skillStore.selectedSkill.status === 'ready', error: skillStore.selectedSkill.status === 'error', warning: skillStore.selectedSkill.status.includes('missing') }">{{ skillStore.selectedSkill.status }}</span><h2>{{ skillStore.selectedSkill.icon }} {{ skillStore.selectedSkill.name }}</h2><p class="muted">v{{ skillStore.selectedSkill.version }} · {{ skillStore.selectedSkill.author || '未知作者' }}</p></div><div class="inline-actions"><button class="button-secondary" @click="toggle(skillStore.selectedSkill.skill_id, skillStore.selectedSkill.enabled)">{{ skillStore.selectedSkill.enabled ? '停用' : '启用' }}</button><button class="button-danger" @click="uninstall(skillStore.selectedSkill.skill_id, skillStore.selectedSkill.name)">卸载</button></div></div>
+      <div class="detail-head"><div><span class="badge" :class="{ success: skillStore.selectedSkill.status === 'ready', error: skillStore.selectedSkill.status === 'error', warning: skillStore.selectedSkill.status.includes('missing') }">{{ skillStore.selectedSkill.status }}</span><h2>{{ skillStore.selectedSkill.icon }} {{ skillStore.selectedSkill.name }}</h2><p class="muted">v{{ skillStore.selectedSkill.version }} · {{ skillStore.selectedSkill.author || t('未知作者', 'Unknown author') }}</p></div><div class="inline-actions"><button class="button-secondary" @click="toggle(skillStore.selectedSkill.skill_id, skillStore.selectedSkill.enabled)">{{ skillStore.selectedSkill.enabled ? t('停用', 'Disable') : t('启用', 'Enable') }}</button><button class="button-danger" @click="uninstall(skillStore.selectedSkill.skill_id, skillStore.selectedSkill.name)">{{ t('卸载', 'Uninstall') }}</button></div></div>
       <p class="description">{{ skillStore.selectedSkill.description }}</p>
-      <div class="detail-grid"><div><h3>工具</h3><div class="tag-list"><span v-for="tool in skillStore.selectedSkill.tools" :key="tool" class="badge info">{{ tool }}</span></div></div><div><h3>权限</h3><div class="tag-list"><span v-for="permission in skillStore.selectedSkill.permissions" :key="permission" class="badge warning">{{ permission }}</span></div></div><div><h3>检索配置</h3><pre>{{ JSON.stringify(skillStore.selectedSkill.retrieval_config, null, 2) }}</pre></div><div><h3>模型能力</h3><div class="tag-list"><span v-for="cap in skillStore.selectedSkill.model_requirements?.capabilities" :key="cap" class="badge">{{ cap }}</span></div></div></div>
+      <div class="detail-grid"><div><h3>{{ t('工具', 'Tools') }}</h3><div class="tag-list"><span v-for="tool in skillStore.selectedSkill.tools" :key="tool" class="badge info">{{ tool }}</span></div></div><div><h3>{{ t('权限', 'Permissions') }}</h3><div class="tag-list"><span v-for="permission in skillStore.selectedSkill.permissions" :key="permission" class="badge warning">{{ permission }}</span></div></div><div><h3>{{ t('检索配置', 'Retrieval Settings') }}</h3><pre>{{ JSON.stringify(skillStore.selectedSkill.retrieval_config, null, 2) }}</pre></div><div><h3>{{ t('模型能力', 'Model Capabilities') }}</h3><div class="tag-list"><span v-for="cap in skillStore.selectedSkill.model_requirements?.capabilities" :key="cap" class="badge">{{ cap }}</span></div></div></div>
-      <div v-if="skillStore.selectedSkill.missing_dependencies?.length" class="error-banner dependencies">缺失依赖：{{ skillStore.selectedSkill.missing_dependencies.join('、') }}</div>
+      <div v-if="skillStore.selectedSkill.missing_dependencies?.length" class="error-banner dependencies">{{ t('缺失依赖：', 'Missing dependencies: ') }}{{ skillStore.selectedSkill.missing_dependencies.join(', ') }}</div>
     </div>
-    <div v-else-if="!skillStore.skills.length" class="empty-state"><div><strong>{{ skillStore.isLoading ? '正在加载…' : skillStore.error ? '加载失败' : '尚未安装' }}</strong><button class="button-secondary" @click="skillStore.loadSkills">重新加载</button></div></div>
+    <div v-else-if="!skillStore.skills.length" class="empty-state"><div><strong>{{ skillStore.isLoading ? t('正在加载…', 'Loading…') : skillStore.error ? t('加载失败', 'Load failed') : t('尚未安装', 'Not installed') }}</strong><button class="button-secondary" @click="skillStore.loadSkills">{{ t('重新加载', 'Reload') }}</button></div></div>
     <div v-else class="feature-grid"><article v-for="skill in skillStore.skills" :key="skill.skill_id" class="item-card extension-card" @click="skillStore.selectSkill(skill.skill_id)"><div class="extension-title"><AppIcon :icon="Lightning" :size="22" /><div><strong>{{ skill.name }}</strong><p>v{{ skill.version }}</p></div><span class="badge" :class="{ success: skill.status === 'ready', warning: skill.status === 'dependency_missing' }">{{ skill.status }}</span></div><p class="muted">{{ skill.description }}</p><div class="tag-list"><span v-for="permission in skill.permissions.slice(0, 3)" :key="permission" class="badge">{{ permission }}</span></div></article></div>
   </section>
 </template>

frontend/src/features/tasks/TaskFiltersPanel.vue

@@ -1,12 +1,13 @@
 <script setup lang="ts">
 import { useTaskStore } from '@/stores/task'
+import { t } from '@/i18n'
 const taskStore = useTaskStore()
 </script>
 
 <template>
   <div class="sidebar-panel filters">
-    <div class="field"><label>状态</label><select v-model="taskStore.filterStatus" class="select"><option value="all">全部</option><option value="todo">待办</option><option value="in_progress">进行中</option><option value="done">已完成</option><option value="cancelled">已取消</option></select></div>
+    <div class="field"><label>{{ t('状态', 'Status') }}</label><select v-model="taskStore.filterStatus" class="select"><option value="all">{{ t('全部', 'All') }}</option><option value="todo">{{ t('待办', 'To do') }}</option><option value="in_progress">{{ t('进行中', 'In progress') }}</option><option value="done">{{ t('已完成', 'Completed') }}</option><option value="cancelled">{{ t('已取消', 'Cancelled') }}</option></select></div>
-    <div class="task-counts"><p><span>待办</span><strong>{{ taskStore.todoTasks.length }}</strong></p><p><span>进行中</span><strong>{{ taskStore.inProgressTasks.length }}</strong></p><p><span>已完成</span><strong>{{ taskStore.doneTasks.length }}</strong></p></div>
+    <div class="task-counts"><p><span>{{ t('待办', 'To do') }}</span><strong>{{ taskStore.todoTasks.length }}</strong></p><p><span>{{ t('进行中', 'In progress') }}</span><strong>{{ taskStore.inProgressTasks.length }}</strong></p><p><span>{{ t('已完成', 'Completed') }}</span><strong>{{ taskStore.doneTasks.length }}</strong></p></div>
   </div>
 </template>
 

frontend/src/features/tasks/TasksView.vue

@@ -2,6 +2,7 @@
 import { onMounted, reactive, ref } from 'vue'
 import type { TaskItem, TaskStatus } from '@/contracts'
 import { useTaskStore } from '@/stores/task'
+import { localeTag, t } from '@/i18n'
 
 const taskStore = useTaskStore()
 const showForm = ref(false)
@@ -20,32 +21,32 @@ async function saveTask() {
     if (editingId.value) await taskStore.updateTask(editingId.value, { ...form, due_date: form.due_date || undefined, note_id: form.note_id || null })
     else await taskStore.createTask({ ...form, due_date: form.due_date || undefined, note_id: form.note_id || undefined })
     showForm.value = false; resetForm()
-  } catch (error) { actionError.value = error instanceof Error ? error.message : '任务保存失败' }
+  } catch (error) { actionError.value = error instanceof Error ? error.message : t('任务保存失败', 'Failed to save task') }
 }
 
 async function setStatus(task: TaskItem, status: TaskStatus) {
-  try { await taskStore.updateTask(task.task_id, { status }) } catch (error) { actionError.value = error instanceof Error ? error.message : '状态更新失败' }
+  try { await taskStore.updateTask(task.task_id, { status }) } catch (error) { actionError.value = error instanceof Error ? error.message : t('状态更新失败', 'Failed to update status') }
 }
 
 async function remove(task: TaskItem) {
-  if (!confirm(`确定删除任务“${task.title}”吗？`)) return
+  if (!confirm(`${t('确定删除任务', 'Delete task')} “${task.title}”?`)) return
-  try { await taskStore.deleteTask(task.task_id) } catch (error) { actionError.value = error instanceof Error ? error.message : '任务删除失败' }
+  try { await taskStore.deleteTask(task.task_id) } catch (error) { actionError.value = error instanceof Error ? error.message : t('任务删除失败', 'Failed to delete task') }
 }
 </script>
 
 <template>
   <section class="feature-page">
-    <header class="feature-header"><div><h1>任务</h1><p>管理用户、笔记和 Agent 产生的行动项。</p></div><button class="button-primary" @click="resetForm(); showForm = true">＋ 新建任务</button></header>
+    <header class="feature-header"><div><h1>{{ t('任务', 'Tasks') }}</h1><p>{{ t('管理用户、笔记和 Agent 产生的行动项。', 'Manage action items created by users, notes, and agents.') }}</p></div><button class="button-primary" @click="resetForm(); showForm = true">＋ {{ t('新建任务', 'New task') }}</button></header>
     <div v-if="taskStore.error || actionError" class="error-banner">{{ taskStore.error || actionError }}</div>
     <div v-if="taskStore.filteredTasks.length" class="task-list">
       <article v-for="task in taskStore.filteredTasks" :key="task.task_id" class="item-card task-card">
-        <button class="status-check" :class="{ done: task.status === 'done' }" title="切换完成状态" @click="setStatus(task, task.status === 'done' ? 'todo' : 'done')">{{ task.status === 'done' ? '✓' : '' }}</button>
+        <button class="status-check" :class="{ done: task.status === 'done' }" :title="t('切换完成状态', 'Toggle completion')" @click="setStatus(task, task.status === 'done' ? 'todo' : 'done')">{{ task.status === 'done' ? '✓' : '' }}</button>
-        <div class="task-content"><div class="task-title"><strong :class="{ completed: task.status === 'done' }">{{ task.title }}</strong></div><p v-if="task.description" class="muted">{{ task.description }}</p><div class="subtle"><span>{{ task.status }}</span><span v-if="task.due_date">截止 {{ new Date(task.due_date).toLocaleString() }}</span><span v-if="task.note_id">关联 Note：{{ task.note_id }}</span></div></div>
+        <div class="task-content"><div class="task-title"><strong :class="{ completed: task.status === 'done' }">{{ task.title }}</strong></div><p v-if="task.description" class="muted">{{ task.description }}</p><div class="subtle"><span>{{ task.status }}</span><span v-if="task.due_date">{{ t('截止', 'Due') }} {{ new Date(task.due_date).toLocaleString(localeTag()) }}</span><span v-if="task.note_id">{{ t('关联 Note', 'Linked Note') }}: {{ task.note_id }}</span></div></div>
-        <div class="inline-actions"><button class="icon-button" @click="editTask(task)">编辑</button><button class="button-danger" @click="remove(task)">删除</button></div>
+        <div class="inline-actions"><button class="icon-button" @click="editTask(task)">{{ t('编辑', 'Edit') }}</button><button class="button-danger" @click="remove(task)">{{ t('删除', 'Delete') }}</button></div>
       </article>
     </div>
-    <div v-else class="empty-state"><div><strong>{{ taskStore.isLoading ? '正在加载任务…' : '没有符合条件的任务' }}</strong><p>创建一项任务，或调整左侧筛选条件。</p></div></div>
+    <div v-else class="empty-state"><div><strong>{{ taskStore.isLoading ? t('正在加载任务…', 'Loading tasks…') : t('没有符合条件的任务', 'No matching tasks') }}</strong><p>{{ t('创建一项任务，或调整左侧筛选条件。', 'Create a task or adjust the filters.') }}</p></div></div>
-    <div v-if="showForm" class="modal-backdrop" @click.self="showForm = false"><div class="modal"><h2>{{ editingId ? '编辑任务' : '新建任务' }}</h2><form @submit.prevent="saveTask"><div class="field"><label>标题</label><input v-model="form.title" class="input" required /></div><div class="field"><label>描述</label><textarea v-model="form.description" class="textarea" /></div><div class="field"><label>截止时间</label><input v-model="form.due_date" class="input" type="datetime-local" /></div><div class="field"><label>关联 Note ID</label><input v-model="form.note_id" class="input" /></div><div class="inline-actions"><button class="button-primary">保存</button><button type="button" class="button-secondary" @click="showForm = false">取消</button></div></form></div></div>
+    <div v-if="showForm" class="modal-backdrop" @click.self="showForm = false"><div class="modal"><h2>{{ editingId ? t('编辑任务', 'Edit task') : t('新建任务', 'New task') }}</h2><form @submit.prevent="saveTask"><div class="field"><label>{{ t('标题', 'Title') }}</label><input v-model="form.title" class="input" required /></div><div class="field"><label>{{ t('描述', 'Description') }}</label><textarea v-model="form.description" class="textarea" /></div><div class="field"><label>{{ t('截止时间', 'Due date') }}</label><input v-model="form.due_date" class="input" type="datetime-local" /></div><div class="field"><label>{{ t('关联 Note ID', 'Linked Note ID') }}</label><input v-model="form.note_id" class="input" /></div><div class="inline-actions"><button class="button-primary">{{ t('保存', 'Save') }}</button><button type="button" class="button-secondary" @click="showForm = false">{{ t('取消', 'Cancel') }}</button></div></form></div></div>
   </section>
 </template>
 

frontend/src/features/themes/ThemesView.vue

@@ -2,6 +2,7 @@
 import { computed } from 'vue'
 import MarkdownContent from '@/components/common/MarkdownContent.vue'
 import { useThemeStore } from '@/stores/theme'
+import { t } from '@/i18n'
 
 const themeStore = useThemeStore()
 const shikiPreview = `\`\`\`typescript
@@ -14,25 +15,25 @@ const codeThemeLabel = computed(() => themeStore.resolvedCodeBlockTheme === 'git
 
 <template>
   <section class="feature-page">
-    <header class="feature-header"><div><h1>主题</h1><p>预览并切换 Design Token，编辑器偏好会即时生效。</p></div><button class="button-secondary" @click="themeStore.resetToDefault">恢复默认</button></header>
+    <header class="feature-header"><div><h1>{{ t('主题', 'Themes') }}</h1><p>{{ t('预览并切换 Design Token，编辑器偏好会即时生效。', 'Preview and switch design tokens. Editor preferences apply immediately.') }}</p></div><button class="button-secondary" @click="themeStore.resetToDefault">{{ t('恢复默认', 'Reset defaults') }}</button></header>
     <div class="feature-grid themes">
       <button v-for="theme in themeStore.themes" :key="theme.theme_id" class="item-card theme-card" :class="{ selected: themeStore.currentThemeId === theme.theme_id }" @click="themeStore.applyTheme(theme.theme_id)">
         <div class="theme-preview" :class="`preview-${theme.theme_id}`"><span></span><span></span><span></span><div></div></div>
-        <div class="theme-info"><div><strong>{{ theme.name }}</strong><p class="subtle">{{ theme.description }}</p></div><span v-if="themeStore.currentThemeId === theme.theme_id" class="badge success">使用中</span></div>
+        <div class="theme-info"><div><strong>{{ theme.name }}</strong><p class="subtle">{{ theme.description }}</p></div><span v-if="themeStore.currentThemeId === theme.theme_id" class="badge success">{{ t('使用中', 'Active') }}</span></div>
-        <p class="subtle">v{{ theme.version }} · {{ theme.builtin ? '内置主题' : theme.author }}</p>
+        <p class="subtle">v{{ theme.version }} · {{ theme.builtin ? t('内置主题', 'Built-in theme') : theme.author }}</p>
       </button>
     </div>
     <div class="panel preference-panel">
-      <h2 class="panel-title">编辑器外观</h2>
+      <h2 class="panel-title">{{ t('编辑器外观', 'Editor Appearance') }}</h2>
       <div class="form-grid">
-        <div class="field"><label>字号：{{ themeStore.fontEditorSize }}px</label><input v-model.number="themeStore.fontEditorSize" type="range" min="12" max="24" /></div>
+        <div class="field"><label>{{ t('字号', 'Font size') }}: {{ themeStore.fontEditorSize }}px</label><input v-model.number="themeStore.fontEditorSize" type="range" min="12" max="24" /></div>
-        <div class="field"><label>行高：{{ themeStore.lineHeight }}</label><input v-model.number="themeStore.lineHeight" type="range" min="1.2" max="2.2" step="0.1" /></div>
+        <div class="field"><label>{{ t('行高', 'Line height') }}: {{ themeStore.lineHeight }}</label><input v-model.number="themeStore.lineHeight" type="range" min="1.2" max="2.2" step="0.1" /></div>
-        <div class="field"><label>字体</label><select v-model="themeStore.fontEditorFamily" class="select"><option value="system-ui">系统字体</option><option value="serif">衬线字体</option><option value="var(--font-ui-mono)">等宽字体</option></select></div>
+        <div class="field"><label>{{ t('字体', 'Font') }}</label><select v-model="themeStore.fontEditorFamily" class="select"><option value="system-ui">{{ t('系统字体', 'System font') }}</option><option value="serif">{{ t('衬线字体', 'Serif') }}</option><option value="var(--font-ui-mono)">{{ t('等宽字体', 'Monospace') }}</option></select></div>
-        <div class="field"><label>代码块样式</label><select v-model="themeStore.codeBlockTheme" class="select"><option value="auto">跟随主题</option><option value="github-light">GitHub Light</option><option value="github-dark">GitHub Dark</option></select><small>Markdown 渲染使用对应的 Shiki GitHub 主题</small></div>
+        <div class="field"><label>{{ t('代码块样式', 'Code block style') }}</label><select v-model="themeStore.codeBlockTheme" class="select"><option value="auto">{{ t('跟随主题', 'Follow theme') }}</option><option value="github-light">GitHub Light</option><option value="github-dark">GitHub Dark</option></select><small>{{ t('Markdown 渲染使用对应的 Shiki GitHub 主题', 'Markdown rendering uses the matching Shiki GitHub theme') }}</small></div>
       </div>
       <div class="editor-preview" :style="{ fontSize: `${themeStore.fontEditorSize}px`, lineHeight: themeStore.lineHeight, fontFamily: themeStore.fontEditorFamily }">
-        <div class="preview-heading"><h3>主题预览</h3><span class="badge info">{{ codeThemeLabel }}</span></div>
+        <div class="preview-heading"><h3>{{ t('主题预览', 'Theme Preview') }}</h3><span class="badge info">{{ codeThemeLabel }}</span></div>
-        <p>知识的价值不只在于保存，更在于被重新发现和使用。</p>
+        <p>{{ t('知识的价值不只在于保存，更在于被重新发现和使用。', 'Knowledge gains value when it can be rediscovered and used.') }}</p>
         <MarkdownContent class="code-theme-preview" :source="shikiPreview" />
       </div>
     </div>

frontend/src/features/vault/VaultEntry.vue

@@ -6,6 +6,7 @@ import { useThemeStore } from '@/stores/theme'
 import { useSettingsStore } from '@/stores/settings'
 import { ArrowRight, Document, Folder, FolderOpened, Moon, Sunny } from '@element-plus/icons-vue'
 import AppIcon from '@/components/common/AppIcon.vue'
+import { t } from '@/i18n'
 
 const router = useRouter()
 const workspaceStore = useWorkspaceStore()
@@ -55,15 +56,15 @@ async function openFolderPicker() {
       <div class="brand-section">
         <div class="logo"><AppIcon :icon="Document" :size="56" /></div>
         <h1 class="app-title">NotesAgent</h1>
-        <p class="app-subtitle">本地优先的 AI 笔记软件</p>
+        <p class="app-subtitle">{{ t('本地优先的 AI 笔记软件', 'A local-first AI note-taking app') }}</p>
       </div>
 
       <div class="vault-card">
-        <h2 class="card-title">选择知识库</h2>
+        <h2 class="card-title">{{ t('选择知识库', 'Select Knowledge Base') }}</h2>
-        <p class="card-desc">Web 联调模式连接 AI Core 当前配置的 Vault</p>
+        <p class="card-desc">{{ t('Web 联调模式连接 AI Core 当前配置的 Vault', 'Web development mode connects to the Vault configured in AI Core') }}</p>
 
         <div v-if="workspaceStore.recentVaults.length" class="recent-vaults">
-          <div class="section-label">最近打开</div>
+          <div class="section-label">{{ t('最近打开', 'Recently opened') }}</div>
           <div class="vault-list">
             <button
               v-for="vault in workspaceStore.recentVaults"
@@ -84,15 +85,15 @@ async function openFolderPicker() {
 
         <div class="actions">
           <button class="btn btn-primary" @click="openFolderPicker" :disabled="isLoading || !workspaceStore.recentVaults.length">
-            <AppIcon :icon="FolderOpened" /> 打开后端 Vault
+            <AppIcon :icon="FolderOpened" /> {{ t('打开后端 Vault', 'Open backend Vault') }}
           </button>
         </div>
 
         <div class="ai-core-status">
           <span class="status-dot" :class="aiCoreStatus" />
-          <span v-if="aiCoreStatus === 'checking'">正在检查 AI Core 状态...</span>
+          <span v-if="aiCoreStatus === 'checking'">{{ t('正在检查 AI Core 状态...', 'Checking AI Core status...') }}</span>
-          <span v-else-if="aiCoreStatus === 'running'" class="status-running">AI Core 运行正常</span>
+          <span v-else-if="aiCoreStatus === 'running'" class="status-running">{{ t('AI Core 运行正常', 'AI Core is running') }}</span>
-          <span v-else class="status-stopped">AI Core 未启动（编辑功能仍可用）</span>
+          <span v-else class="status-stopped">{{ t('AI Core 未启动（编辑功能仍可用）', 'AI Core is offline (editing remains available)') }}</span>
         </div>
       </div>
 
@@ -100,7 +101,7 @@ async function openFolderPicker() {
         <span>v0.1.0</span>
         <button class="theme-toggle" @click="themeStore.toggleTheme()">
           <AppIcon :icon="themeStore.isDark ? Sunny : Moon" :size="15" />
-          {{ themeStore.isDark ? '浅色' : '深色' }}
+          {{ themeStore.isDark ? t('浅色', 'Light') : t('深色', 'Dark') }}
         </button>
       </div>
     </div>

frontend/src/features/workspace/FileTreePanel.vue

@@ -8,6 +8,7 @@ import { useWorkspaceStore } from '@/stores/workspace'
 import FileTreeNode from './FileTreeNode.vue'
 import { DocumentAdd, FolderAdd } from '@element-plus/icons-vue'
 import AppIcon from '@/components/common/AppIcon.vue'
+import { t } from '@/i18n'
 
 const workspaceStore = useWorkspaceStore()
 const editorStore = useEditorStore()
@@ -91,7 +92,7 @@ function closeContextMenu() { contextTarget.value = null }
 async function renameTarget() {
   const node = contextTarget.value
   if (!node) return
-  const newName = window.prompt('新名称', node.name)?.trim()
+  const newName = window.prompt(t('新名称', 'New name'), node.name)?.trim()
   if (newName && newName !== node.name) {
     const normalizedName = node.type === 'file' && !newName.toLowerCase().endsWith('.md') ? `${newName}.md` : newName
     const oldPath = node.path
@@ -113,7 +114,7 @@ async function renameTarget() {
 async function deleteTarget() {
   const node = contextTarget.value
   if (!node) return
-  if (!window.confirm(`确定要删除“${node.name}”吗？`)) return closeContextMenu()
+  if (!window.confirm(`${t('确定要删除', 'Delete')} “${node.name}”?`)) return closeContextMenu()
   await workspaceService.deleteFile(node.path)
   const activeWasRemoved = workspaceStore.closePath(node.path)
   workspaceStore.removeFromTree(node.path)
@@ -137,13 +138,13 @@ function containingFolder(path: string): string {
 <template>
   <section class="file-tree-panel" @click="closeContextMenu">
     <div class="toolbar">
-      <button type="button" title="新建笔记" aria-label="新建笔记" @click.stop="beginCreate('file', selectedFolderPath)"><AppIcon :icon="DocumentAdd" /></button>
+      <button type="button" :title="t('新建笔记', 'New note')" :aria-label="t('新建笔记', 'New note')" @click.stop="beginCreate('file', selectedFolderPath)"><AppIcon :icon="DocumentAdd" /></button>
-      <button type="button" title="新建文件夹" aria-label="新建文件夹" @click.stop="beginCreate('folder', selectedFolderPath)"><AppIcon :icon="FolderAdd" /></button>
+      <button type="button" :title="t('新建文件夹', 'New folder')" :aria-label="t('新建文件夹', 'New folder')" @click.stop="beginCreate('folder', selectedFolderPath)"><AppIcon :icon="FolderAdd" /></button>
     </div>
     <form v-if="newItemType" class="new-item" @submit.prevent="createItem">
-      <input v-model="newItemName" :placeholder="newItemType === 'file' ? '笔记名称' : '文件夹名称'" autofocus />
+      <input v-model="newItemName" :placeholder="newItemType === 'file' ? t('笔记名称', 'Note name') : t('文件夹名称', 'Folder name')" autofocus />
-      <button type="submit">创建</button>
+      <button type="submit">{{ t('创建', 'Create') }}</button>
-      <button type="button" @click="newItemType = null">取消</button>
+      <button type="button" @click="newItemType = null">{{ t('取消', 'Cancel') }}</button>
     </form>
     <div class="tree">
       <FileTreeNode v-for="node in workspaceStore.fileTree" :key="node.id" :node="node"
@@ -152,8 +153,8 @@ function containingFolder(path: string): string {
     <Teleport to="body">
       <div v-if="contextTarget" class="context-menu"
         :style="{ left: `${contextMenuPosition.x}px`, top: `${contextMenuPosition.y}px` }" @click.stop>
-        <button @click="renameTarget">重命名</button>
+        <button @click="renameTarget">{{ t('重命名', 'Rename') }}</button>
-        <button class="danger" @click="deleteTarget">删除</button>
+        <button class="danger" @click="deleteTarget">{{ t('删除', 'Delete') }}</button>
       </div>
     </Teleport>
   </section>

frontend/src/features/workspace/WorkspaceView.vue

@@ -4,6 +4,7 @@ import EditorHeader from '@/features/editor/EditorHeader.vue'
 import EditorPane from '@/features/editor/EditorPane.vue'
 import { EditPen } from '@element-plus/icons-vue'
 import AppIcon from '@/components/common/AppIcon.vue'
+import { t } from '@/i18n'
 
 const workspaceStore = useWorkspaceStore()
 </script>
@@ -17,8 +18,8 @@ const workspaceStore = useWorkspaceStore()
     <div v-else class="empty-workspace">
       <div class="empty-content">
         <AppIcon class="empty-icon" :icon="EditPen" :size="48" />
-        <h2>开始写作</h2>
+        <h2>{{ t('开始写作', 'Start writing') }}</h2>
-        <p>从左侧文件树选择笔记，或创建新的笔记</p>
+        <p>{{ t('从左侧文件树选择笔记，或创建新的笔记', 'Select a note from the file tree or create a new one') }}</p>
       </div>
     </div>
   </div>

frontend/src/i18n.spec.ts

@@ -0,0 +1,41 @@
+// @vitest-environment happy-dom
+import { afterEach, beforeEach, describe, expect, it } from 'vitest'
+import { mount } from '@vue/test-utils'
+import { createPinia, setActivePinia } from 'pinia'
+import { createMemoryHistory, createRouter } from 'vue-router'
+import { nextTick } from 'vue'
+import PrimarySidebar from '@/components/common/PrimarySidebar.vue'
+import { appLocale, t } from '@/i18n'
+import { useSettingsStore } from '@/stores/settings'
+
+beforeEach(() => {
+  localStorage.clear()
+  appLocale.value = 'zh-CN'
+  setActivePinia(createPinia())
+})
+
+afterEach(() => {
+  appLocale.value = 'zh-CN'
+  localStorage.clear()
+})
+
+describe('interface locale', () => {
+  it('changes shared labels and the document language immediately', async () => {
+    const router = createRouter({
+      history: createMemoryHistory(),
+      routes: [{ path: '/', name: 'workspace', component: { template: '<div />' } }],
+    })
+    await router.push('/')
+    await router.isReady()
+    const wrapper = mount(PrimarySidebar, { global: { plugins: [router] } })
+    const settings = useSettingsStore()
+
+    expect(wrapper.text()).toContain('工作区')
+    settings.language = 'en'
+    await nextTick()
+
+    expect(t('工作区', 'Workspace')).toBe('Workspace')
+    expect(wrapper.text()).toContain('Workspace')
+    expect(document.documentElement.lang).toBe('en')
+  })
+})

frontend/src/i18n.ts

@@ -0,0 +1,28 @@
+import { ref, watch } from 'vue'
+
+export type AppLocale = 'zh-CN' | 'en'
+
+function storedLocale(): AppLocale {
+  if (typeof localStorage === 'undefined') return 'zh-CN'
+  try {
+    const saved = JSON.parse(localStorage.getItem('app-settings') ?? '{}') as { language?: unknown }
+    return saved.language === 'en' ? 'en' : 'zh-CN'
+  } catch {
+    return 'zh-CN'
+  }
+}
+
+export const appLocale = ref<AppLocale>(storedLocale())
+
+watch(appLocale, (value) => {
+  if (typeof document !== 'undefined') document.documentElement.lang = value
+}, { immediate: true })
+
+/** Keep the Chinese source beside its English translation while the UI is migrated. */
+export function t(zh: string, en: string): string {
+  return appLocale.value === 'en' ? en : zh
+}
+
+export function localeTag(): string {
+  return appLocale.value === 'en' ? 'en' : 'zh-CN'
+}

frontend/src/main.ts

@@ -5,6 +5,10 @@ import router from './router'
 import './styles/tokens.css'
 import './styles/features.css'
 import { useThemeStore } from './stores/theme'
+import { useSettingsStore } from './stores/settings'
+import { watch } from 'vue'
+import { appLocale } from './i18n'
+import { updateDocumentTitle } from './router'
 
 const app = createApp(App)
 const pinia = createPinia()
@@ -13,6 +17,12 @@ app.use(pinia)
 app.use(router)
 
 const themeStore = useThemeStore()
+const settingsStore = useSettingsStore()
 themeStore.initTheme()
+watch(appLocale, () => updateDocumentTitle())
+watch(() => settingsStore.spellCheck, (enabled) => {
+  document.body.spellcheck = enabled
+  document.body.setAttribute('spellcheck', String(enabled))
+}, { immediate: true })
 
 app.mount('#app')

frontend/src/router/index.ts

@@ -1,5 +1,6 @@
 import { createRouter, createWebHashHistory } from 'vue-router'
 import { useWorkspaceStore } from '@/stores/workspace'
+import { t } from '@/i18n'
 
 const routes = [
   { path: '/media', name: 'media', component: () => import('@/features/media/MediaView.vue'), meta: { title: '音视频转写', requiresVault: true } },
@@ -87,10 +88,26 @@ router.beforeEach((to) => {
   return true
 })
 
-router.afterEach((to) => {
+export function updateDocumentTitle(to = router.currentRoute.value) {
   const baseTitle = 'NotesAgent'
-  const title = to.meta.title as string | undefined
+  const titles: Record<string, string> = {
+    media: t('音视频转写', 'Media Transcription'),
+    'vault-entry': t('选择知识库', 'Select Knowledge Base'),
+    workspace: t('工作区', 'Workspace'),
+    search: t('搜索', 'Search'),
+    chat: t('AI 对话', 'AI Chat'),
+    agent: 'Agent Trace',
+    tasks: t('任务', 'Tasks'),
+    skills: t('Skill 管理', 'Skill Management'),
+    'mcp-servers': t('MCP 服务器', 'MCP Servers'),
+    plugins: t('Plugin 与 MCP', 'Plugins and MCP'),
+    themes: t('主题管理', 'Theme Management'),
+    settings: t('设置', 'Settings'),
+  }
+  const title = titles[String(to.name ?? '')] ?? (to.meta.title as string | undefined)
   document.title = title ? `${title} · ${baseTitle}` : baseTitle
-})
+}
+
+router.afterEach(updateDocumentTitle)
 
 export default router

frontend/src/services/chatService.ts

@@ -1,10 +1,14 @@
 import { SseClient } from './sseClient'
-import type { ModelEvent } from '@/contracts'
+import { apiClient } from './apiClient'
+import type { ChatMessage, Conversation, ModelEvent, PageMeta } from '@/contracts'
 
 export interface ChatRequest {
   provider_id: string
   model: string
   conversation_id?: string
+  user_message_id?: string
+  assistant_message_id?: string
+  conversation_title?: string
   system?: string
   messages: Array<{
     role: 'system' | 'user' | 'assistant' | 'tool'
@@ -18,6 +22,25 @@ export interface ChatRequest {
   max_tokens?: number
 }
 
+export function listConversations(offset = 0, limit = 100) {
+  return apiClient.get<{ items: Conversation[]; page: PageMeta }>('/api/chat/conversations', { params: { limit, offset } })
+}
+
+export function createConversation(conversation: Pick<Conversation, 'conversation_id' | 'title'>) {
+  return apiClient.post<Conversation>('/api/chat/conversations', {
+    conversation_id: conversation.conversation_id,
+    title: conversation.title,
+  })
+}
+
+export function listConversationMessages(conversationId: string, offset = 0, limit = 500) {
+  return apiClient.get<{ items: ChatMessage[]; page: PageMeta }>(`/api/chat/conversations/${encodeURIComponent(conversationId)}/messages`, { params: { limit, offset } })
+}
+
+export function removeConversation(conversationId: string) {
+  return apiClient.delete(`/api/chat/conversations/${encodeURIComponent(conversationId)}`)
+}
+
 export function streamChat(
   request: ChatRequest,
   handlers: {

frontend/src/services/mediaService.spec.ts

@@ -0,0 +1,42 @@
+// @vitest-environment happy-dom
+import { afterEach, expect, it, vi } from 'vitest'
+import { createMediaSubmission, mediaService, type MediaJob } from './mediaService'
+
+afterEach(() => vi.restoreAllMocks())
+
+it('reuses upload and job identities after lost responses, until explicitly reset', async () => {
+  const upload = vi.spyOn(mediaService, 'upload').mockRejectedValueOnce(new Error('response lost'))
+    .mockResolvedValue({attachment_id:'uploaded'})
+  const create = vi.spyOn(mediaService, 'create').mockRejectedValueOnce(new Error('response lost'))
+    .mockResolvedValue({job_id:'same-job'} as MediaJob)
+  const submission = createMediaSubmission()
+  const file = new File(['audio'], 'lecture.wav')
+  const options = {local_only:true}
+  await expect(submission.submit(file, options)).rejects.toThrow('response lost')
+  await expect(submission.submit(file, options)).rejects.toThrow('response lost')
+  expect(await submission.submit(file, options)).toEqual({job_id:'same-job'})
+  expect(upload).toHaveBeenCalledTimes(2)
+  expect(upload.mock.calls[0][1]).toBe(upload.mock.calls[1][1])
+  expect(create.mock.calls[0][0]).toEqual(create.mock.calls[1][0])
+  submission.reset()
+  await submission.submit(file, options)
+  expect(upload.mock.calls[2][1]).not.toBe(upload.mock.calls[1][1])
+  expect(create.mock.calls[2][0]).not.toEqual(create.mock.calls[1][0])
+})
+
+it('freezes options across upload and treats changed options as a new request', async () => {
+  let release!: (value:{attachment_id:string}) => void
+  vi.spyOn(mediaService, 'upload').mockImplementationOnce(() => new Promise(resolve => { release = resolve }))
+    .mockResolvedValue({attachment_id:'next'})
+  const create = vi.spyOn(mediaService, 'create').mockResolvedValue({job_id:'job'} as MediaJob)
+  const submission = createMediaSubmission()
+  const file = new File(['audio'], 'lecture.wav')
+  const options = {local_only:true}
+  const pending = submission.submit(file, options)
+  options.local_only = false
+  release({attachment_id:'first'})
+  await pending
+  expect(create.mock.calls[0][0]).toMatchObject({local_only:true})
+  await submission.submit(file, options)
+  expect(create.mock.calls[1][0]).toMatchObject({local_only:false})
+})

frontend/src/services/mediaService.ts

@@ -1,4 +1,5 @@
 import { apiClient, resolveApiUrl } from './apiClient'
+import { t } from '@/i18n'
 
 export interface Segment { segment_id: string; start_time: number; end_time: number; text: string; speaker: string | null; language?: string }
 export interface MediaJob {
@@ -18,15 +19,33 @@ export const mediaService = {
     revision: job.revision, text: job.text, segments: job.segments, speaker_names: job.speaker_names,
   }),
   revisions: (id: string) => apiClient.get<{items: MediaJob[]}>(`/api/media/transcriptions/${encodeURIComponent(id)}/revisions`),
-  note: (id: string, title: string) => apiClient.post<{note_id: string; title: string}>(`/api/media/transcriptions/${encodeURIComponent(id)}/notes`, { title }),
+  note: (id: string, title: string, update_existing = false) => apiClient.post<{note_id: string; title: string}>(`/api/media/transcriptions/${encodeURIComponent(id)}/notes`, { title, update_existing }),
   audio: (id: string) => resolveApiUrl(`/api/media/attachments/${encodeURIComponent(id)}`),
   impact: (id: string) => apiClient.get<{message:string;retained_note_ids:string[]}>(`/api/media/attachments/${encodeURIComponent(id)}/cleanup-impact`),
   purge: (id: string) => apiClient.delete(`/api/media/attachments/${encodeURIComponent(id)}`),
-  async upload(file: File) {
+  async upload(file: File, idempotencyKey?: string) {
     const response = await fetch(resolveApiUrl(`/api/media/attachments?filename=${encodeURIComponent(file.name)}`), {
-      method: 'POST', headers: {'Content-Type': 'application/octet-stream'}, body: file,
+      method: 'POST', headers: {'Content-Type': 'application/octet-stream', ...(idempotencyKey ? {'Idempotency-Key': idempotencyKey} : {})}, body: file,
     })
-    if (!response.ok) throw new Error((await response.json())?.error?.message || '附件上传失败')
+  if (!response.ok) throw new Error((await response.json())?.error?.message || t('附件上传失败', 'Attachment upload failed'))
     return await response.json() as {attachment_id: string}
   },
 }
+
+// Keep one identity until the input/options change, including a lost HTTP response.
+// Payloads remain in memory; durable uploads/jobs are owned by the backend.
+export function createMediaSubmission() {
+  let pending: {file: File; options: string; uploadKey: string; jobKey: string; attachmentId?: string} | null = null
+  return {
+    reset() { pending = null },
+    async submit(file: File, options: Record<string, unknown>) {
+      const serialized = JSON.stringify(options)
+      if (!pending || pending.file !== file || pending.options !== serialized) {
+        pending = {file, options: serialized, uploadKey: crypto.randomUUID(), jobKey: crypto.randomUUID()}
+      }
+      const current = pending
+      if (!current.attachmentId) current.attachmentId = (await mediaService.upload(file, current.uploadKey)).attachment_id
+      return mediaService.create({...JSON.parse(current.options), attachment_id: current.attachmentId, idempotency_key: current.jobKey})
+    },
+  }
+}

frontend/src/services/workspaceService.ts

@@ -7,6 +7,7 @@ import type {
   OperationResponse,
 } from '@/contracts'
 import apiClient from './apiClient'
+import { t } from '@/i18n'
 import * as noteService from './noteService'
 
 /** Web 联调只连接 AI Core 配置的单一 Vault；多 Vault 选择由 Tauri Host 接管。 */
@@ -72,7 +73,7 @@ async function requireNoteId(filePath: string): Promise<string> {
     await refreshTree()
     noteId = noteIdByPath.get(path)
   }
-  if (!noteId) throw new Error(`笔记尚未建立后端索引：${path}`)
+  if (!noteId) throw new Error(`${t('笔记尚未建立后端索引：', 'The note has not been indexed by the backend: ')}${path}`)
   return noteId
 }
 
@@ -174,7 +175,7 @@ export async function deleteFile(pathValue: string): Promise<void> {
 export async function moveFile(sourcePath: string, targetPath: string): Promise<void> {
   const source = normalizePublicPath(sourcePath)
   if (typeByPath.get(source) !== 'file') {
-    throw new Error('当前阶段只支持移动笔记文件。')
+    throw new Error(t('当前阶段只支持移动笔记文件。', 'Only note files can be moved at this stage.'))
   }
   await noteService.moveNote(await requireNoteId(source), relativePath(targetPath))
   await refreshTree()

frontend/src/stores/agent.ts

@@ -3,6 +3,7 @@ import { ref, computed } from 'vue'
 import type { AgentRun, AgentEvent, ToolDefinition, PermissionRequest, ToolCall } from '@/contracts'
 import * as agentService from '@/services/agentService'
 import type { SseClient } from '@/services/sseClient'
+import { t } from '@/i18n'
 
 export const useAgentStore = defineStore('agent', () => {
   const runs = ref<AgentRun[]>([])
@@ -93,7 +94,7 @@ export const useAgentStore = defineStore('agent', () => {
         tool_name: String(call.name ?? 'unknown'),
         permission: String(data.permission ?? ''),
         parameters: (call.arguments ?? {}) as Record<string, unknown>,
-        impact: '该工具需要获得权限后才能继续执行。',
+        impact: t('该工具需要获得权限后才能继续执行。', 'This tool requires permission before it can continue.'),
       }
       if (run) run.status = 'waiting_permission'
     } else if (['RunCompleted', 'RunFailed', 'RunCancelled'].includes(event.event)) {

frontend/src/stores/chat.spec.ts

@@ -1,36 +1,85 @@
 import { beforeEach, expect, it, vi } from 'vitest'
 import { createPinia, setActivePinia } from 'pinia'
 import { useChatStore } from './chat'
-import { streamChat } from '@/services/chatService'
+import {
+  createConversation,
+  listConversationMessages,
+  listConversations,
+  removeConversation,
+  streamChat,
+} from '@/services/chatService'
+import type { ChatMessage, Conversation } from '@/contracts'
 import type { SseClient } from '@/services/sseClient'
 
-vi.mock('@/services/chatService', () => ({ streamChat: vi.fn() }))
+vi.mock('@/services/chatService', () => ({
+  createConversation: vi.fn(),
+  listConversationMessages: vi.fn(),
+  listConversations: vi.fn(),
+  removeConversation: vi.fn(),
+  streamChat: vi.fn(),
+}))
+
+const page = { total: 0, limit: 100, offset: 0 }
+
+function deferred<T>() {
+  let resolve!: (value: T) => void
+  let reject!: (reason: Error) => void
+  const promise = new Promise<T>((done, fail) => { resolve = done; reject = fail })
+  return { promise, resolve, reject }
+}
+
 beforeEach(() => {
   setActivePinia(createPinia())
   vi.mocked(streamChat).mockReset().mockReturnValue({ cancel: vi.fn() } as unknown as SseClient)
+  vi.mocked(listConversations).mockReset().mockResolvedValue({ items: [], page })
+  vi.mocked(listConversationMessages).mockReset().mockResolvedValue({ items: [], page: { ...page, limit: 1000 } })
+  vi.mocked(createConversation).mockReset().mockImplementation(async value => ({
+    ...value, created_at: new Date().toISOString(), updated_at: new Date().toISOString(), message_count: 0,
+  }))
+  vi.mocked(removeConversation).mockReset().mockResolvedValue(undefined)
 })
 
-it('sends real user history, applies streaming changes, and restores it when switching conversations', async () => {
+it('sends persistent message ids and restores messages from the backend', async () => {
   const store = useChatStore()
   store.selectedProviderId = 'real'
   store.selectedModel = 'configured-model'
   await store.sendMessage('user input')
   const [request, handlers] = vi.mocked(streamChat).mock.calls[0]!
   expect(request.use_rag).toBe(true)
-  handlers.onEvent?.({ event: 'Citation', sequence: 0, timestamp: '', data: { note_id: 'note', block_id: 'block', file_path: 'note.md', content: 'real evidence' } })
+  expect(request.user_message_id).toBe(store.messages[0]?.message_id)
-  expect(store.messages[1]?.citations?.[0]?.content).toBe('real evidence')
+  expect(request.assistant_message_id).toBe(store.messages[1]?.message_id)
   expect(request.messages).toEqual([{ role: 'user', content: 'user input' }])
-  handlers.onEvent?.({ event: 'TextDelta', sequence: 0, timestamp: '', data: { text: 'real response' } })
+  handlers.onEvent?.({ event: 'Citation', sequence: 0, timestamp: '', data: { note_id: 'note', block_id: 'block', file_path: 'note.md', heading_path: ['Heading'], content: 'real evidence' } })
-  expect(store.messages[1]?.content).toBe('real response')
+  handlers.onEvent?.({ event: 'TextDelta', sequence: 1, timestamp: '', data: { text: 'real response' } })
   handlers.onDone?.()
+
+  const persisted = store.messages.map(message => ({ ...message })) as ChatMessage[]
+  vi.mocked(listConversationMessages).mockResolvedValueOnce({ items: persisted, page: { total: 2, limit: 1000, offset: 0 } })
   const id = store.activeConversationId!
-  store.createNewConversation()
+  await store.createNewConversation()
   expect(store.messages).toEqual([])
   await store.setActiveConversation(id)
-  expect(store.messages.map(m => m.content)).toEqual(['user input', 'real response'])
+  expect(store.messages.map(message => message.content)).toEqual(['user input', 'real response'])
+  expect(store.messages[1]?.citations?.[0]?.heading_path).toBe('Heading')
 })
 
-it('does not send without a provider and ignores late callbacks from a cancelled conversation', async () => {
+it('loads the newest persisted conversation on initialization', async () => {
+  const conversation: Conversation = {
+    conversation_id: 'persisted', title: 'Saved', created_at: '2026-01-01T00:00:00Z',
+    updated_at: '2026-01-02T00:00:00Z', message_count: 1,
+  }
+  vi.mocked(listConversations).mockResolvedValue({ items: [conversation], page: { ...page, total: 1 } })
+  vi.mocked(listConversationMessages).mockResolvedValue({
+    items: [{ message_id: 'm1', conversation_id: 'persisted', role: 'user', content: 'saved text', created_at: '2026-01-01T00:00:00Z' }],
+    page: { total: 1, limit: 1000, offset: 0 },
+  })
+  const store = useChatStore()
+  await store.loadConversations()
+  expect(store.activeConversationId).toBe('persisted')
+  expect(store.messages[0]?.content).toBe('saved text')
+})
+
+it('does not send without a provider and ignores callbacks from a cancelled conversation', async () => {
   const store = useChatStore()
   await store.sendMessage('no provider')
   expect(streamChat).not.toHaveBeenCalled()
@@ -38,9 +87,197 @@ it('does not send without a provider and ignores late callbacks from a cancelled
   store.selectedModel = 'configured-model'
   await store.sendMessage('first')
   const old = vi.mocked(streamChat).mock.calls[0]![1]
-  store.createNewConversation()
+  await store.createNewConversation()
   await store.sendMessage('second')
   old.onDone?.()
   expect(store.isStreaming).toBe(true)
   expect(store.messages[0]?.content).toBe('second')
 })
+
+it('keeps a conversation visible when backend deletion fails', async () => {
+  const store = useChatStore()
+  await store.createNewConversation()
+  const id = store.activeConversationId!
+  vi.mocked(removeConversation).mockRejectedValueOnce(new Error('offline'))
+  await store.deleteConversation(id)
+  expect(store.conversations.some(item => item.conversation_id === id)).toBe(true)
+  expect(store.historyError).toBe('offline')
+})
+
+it('blocks sends until history is loaded, then includes that history', async () => {
+  const store = useChatStore()
+  store.selectedProviderId = 'real'
+  store.selectedModel = 'model'
+  await store.createNewConversation()
+  const id = store.activeConversationId!
+  const history = deferred<Awaited<ReturnType<typeof listConversationMessages>>>()
+  vi.mocked(listConversationMessages).mockReturnValueOnce(history.promise)
+  const loading = store.setActiveConversation(id)
+  store.inputText = 'followup'
+  expect(store.canSend).toBe(false)
+  await store.sendMessage(store.inputText)
+  expect(streamChat).not.toHaveBeenCalled()
+  expect(store.inputText).toBe('followup')
+  history.resolve({ items: [{ message_id: 'old', conversation_id: id, role: 'user', content: 'previous context', created_at: '' }], page: { ...page, total: 1 } })
+  await loading
+  expect(store.canSend).toBe(true)
+  await store.sendMessage(store.inputText)
+  expect(vi.mocked(streamChat).mock.calls[0]![0].messages).toEqual([
+    { role: 'user', content: 'previous context' }, { role: 'user', content: 'followup' },
+  ])
+  expect(store.messages.map(m => m.content)).toEqual(['previous context', 'followup', ''])
+})
+
+it('keeps sending blocked after history failure until a successful retry', async () => {
+  const store = useChatStore()
+  store.selectedProviderId = 'real'
+  store.selectedModel = 'model'
+  await store.createNewConversation()
+  const id = store.activeConversationId!
+  vi.mocked(listConversationMessages).mockRejectedValueOnce(new Error('offline'))
+  await store.setActiveConversation(id)
+  await store.sendMessage('followup')
+  expect(streamChat).not.toHaveBeenCalled()
+  expect(store.historyError).toBe('offline')
+  expect(store.canSend).toBe(false)
+  await store.setActiveConversation(id)
+  expect(store.canSend).toBe(true)
+})
+
+it.each(['switch', 'stop', 'delete'] as const)('cancels a pending send on %s without touching another send', async action => {
+  const store = useChatStore()
+  store.selectedProviderId = 'real'
+  store.selectedModel = 'model'
+  await store.createNewConversation()
+  const b = store.activeConversationId!
+  const creation = deferred<Conversation>()
+  vi.mocked(createConversation).mockReturnValueOnce(creation.promise)
+  const creating = store.createNewConversation()
+  const a = store.activeConversationId!
+  const saved = { ...store.activeConversation! }
+  const sending = store.sendMessage('belongs to a')
+  expect(store.isPreparing).toBe(true)
+  const deleting = action === 'delete' ? store.deleteConversation(a) : undefined
+  if (action === 'stop') store.stopGeneration()
+  await store.setActiveConversation(b)
+  await store.sendMessage('belongs to b')
+  creation.resolve(saved)
+  await Promise.all([creating, sending, deleting])
+  expect(store.activeConversationId).toBe(b)
+  expect(store.messages.every(m => m.conversation_id === b)).toBe(true)
+  expect(store.messages[0]?.content).toBe('belongs to b')
+  expect(streamChat).toHaveBeenCalledTimes(1)
+  expect(store.isStreaming).toBe(true)
+})
+
+it('locks the initial send while creating its conversation and allows retry after failure', async () => {
+  const store = useChatStore()
+  store.selectedProviderId = 'real'
+  store.selectedModel = 'model'
+  const creation = deferred<Conversation>()
+  vi.mocked(createConversation).mockReturnValueOnce(creation.promise)
+  const sending = store.sendMessage('first')
+  await store.sendMessage('duplicate')
+  expect(createConversation).toHaveBeenCalledTimes(1)
+  expect(streamChat).not.toHaveBeenCalled()
+  creation.resolve({ ...store.activeConversation! })
+  await sending
+  expect(streamChat).toHaveBeenCalledTimes(1)
+  store.stopGeneration()
+  store.activeConversationId = null
+  vi.mocked(createConversation).mockRejectedValueOnce(new Error('offline'))
+  await store.sendMessage('retry')
+  expect(store.isPreparing).toBe(false)
+  expect(store.canSend).toBe(true)
+  await store.sendMessage('retry')
+  expect(streamChat).toHaveBeenCalledTimes(2)
+})
+
+it('stops the first send before creation completes without switching conversations', async () => {
+  const store = useChatStore()
+  store.selectedProviderId = 'real'
+  store.selectedModel = 'model'
+  const creation = deferred<Conversation>()
+  vi.mocked(createConversation).mockReturnValueOnce(creation.promise)
+  const sending = store.sendMessage('cancelled')
+  const saved = { ...store.activeConversation! }
+  store.stopGeneration()
+  creation.resolve(saved)
+  await sending
+  expect(streamChat).not.toHaveBeenCalled()
+  expect(store.messages).toEqual([])
+  expect(store.canSend).toBe(true)
+})
+
+it('ignores old history after switching to a new conversation and sending', async () => {
+  const store = useChatStore()
+  store.selectedProviderId = 'real'
+  store.selectedModel = 'model'
+  await store.createNewConversation()
+  const id = store.activeConversationId!
+  const history = deferred<Awaited<ReturnType<typeof listConversationMessages>>>()
+  vi.mocked(listConversationMessages).mockReturnValueOnce(history.promise)
+  const loading = store.setActiveConversation(id)
+  await store.createNewConversation()
+  await store.sendMessage('new question')
+  history.resolve({ items: [], page })
+  await loading
+  expect(store.messages.map(m => m.content)).toEqual(['new question', ''])
+  expect(store.isStreaming).toBe(true)
+})
+
+it.each(['success', 'failure'])('blocks sends and duplicate deletes until deletion ends with %s', async outcome => {
+  const store = useChatStore()
+  store.selectedProviderId = 'real'
+  store.selectedModel = 'model'
+  await store.createNewConversation()
+  const id = store.activeConversationId!
+  const removal = deferred<Awaited<ReturnType<typeof removeConversation>>>()
+  vi.mocked(removeConversation).mockReturnValueOnce(removal.promise)
+  const deleting = store.deleteConversation(id)
+  store.inputText = 'keep this draft'
+  expect(store.canSend).toBe(false)
+  await store.sendMessage(store.inputText)
+  await store.deleteConversation(id)
+  expect(streamChat).not.toHaveBeenCalled()
+  expect(removeConversation).toHaveBeenCalledTimes(1)
+  expect(store.inputText).toBe('keep this draft')
+  if (outcome === 'success') removal.resolve(undefined)
+  else removal.reject(new Error('offline'))
+  await deleting
+  expect(store.isStreaming).toBe(false)
+  expect(store.canSend).toBe(true)
+  expect(store.activeConversationId).toBe(outcome === 'success' ? null : id)
+  await store.sendMessage(store.inputText)
+  expect(streamChat).toHaveBeenCalledTimes(1)
+  const request = vi.mocked(streamChat).mock.calls[0]![0]
+  if (outcome === 'success') expect(request.conversation_id).not.toBe(id)
+  else expect(request.conversation_id).toBe(id)
+})
+
+it('keeps a deleting conversation blocked after reselecting it without blocking other conversations', async () => {
+  const store = useChatStore()
+  store.selectedProviderId = 'real'
+  store.selectedModel = 'model'
+  await store.createNewConversation()
+  const a = store.activeConversationId!
+  await store.createNewConversation()
+  const b = store.activeConversationId!
+  const removal = deferred<Awaited<ReturnType<typeof removeConversation>>>()
+  vi.mocked(removeConversation).mockReturnValueOnce(removal.promise)
+  const deleting = store.deleteConversation(a)
+  await store.setActiveConversation(a)
+  expect(store.canSend).toBe(false)
+  await store.sendMessage('blocked')
+  expect(streamChat).not.toHaveBeenCalled()
+  await store.setActiveConversation(b)
+  expect(store.canSend).toBe(true)
+  await store.sendMessage('belongs to b')
+  const client = vi.mocked(streamChat).mock.results[0]!.value as SseClient
+  removal.resolve(undefined)
+  await deleting
+  expect(store.activeConversationId).toBe(b)
+  expect(store.messages[0]?.content).toBe('belongs to b')
+  expect(store.isStreaming).toBe(true)
+  expect(client.cancel).not.toHaveBeenCalled()
+})

frontend/src/stores/chat.ts

@@ -1,92 +1,206 @@
+import { computed, reactive, ref } from 'vue'
 import { defineStore } from 'pinia'
-import { ref, computed, reactive } from 'vue'
+import type { ChatMessage, Citation, Conversation } from '@/contracts'
-import type { ChatMessage, Conversation } from '@/contracts'
+import {
-import { streamChat } from '@/services/chatService'
+  createConversation as createConversationApi,
+  listConversationMessages,
+  listConversations as listConversationsApi,
+  removeConversation,
+  streamChat,
+} from '@/services/chatService'
 import type { SseClient } from '@/services/sseClient'
+import { t } from '@/i18n'
 
 export const useChatStore = defineStore('chat', () => {
   const conversations = ref<Conversation[]>([])
   const activeConversationId = ref<string | null>(null)
   const messages = ref<ChatMessage[]>([])
   const isStreaming = ref(false)
+  const isPreparing = ref(false)
+  const messagesReady = ref(true)
+  const deletingConversations = reactive(new Set<string>())
+  const canSend = computed(() => messagesReady.value && !isPreparing.value && !isStreaming.value
+    && (!activeConversationId.value || !deletingConversations.has(activeConversationId.value)))
   const inputText = ref('')
   const useRag = ref(true)
   const selectedSkillId = ref<string | null>(null)
   const selectedProviderId = ref('')
   const selectedModel = ref('')
+  const historyError = ref('')
+  let initialized = false
+  let loading: Promise<void> | null = null
+  let loadVersion = 0
   let sseClient: SseClient | null = null
   let streamVersion = 0
-
+  const pendingCreates = new Map<string, Promise<void>>()
-  // User-created conversations live in this browser session; no fabricated history.
-  const history = reactive<Record<string, ChatMessage[]>>({})
 
   const activeConversation = computed(() =>
-    conversations.value.find((c) => c.conversation_id === activeConversationId.value) || null
+    conversations.value.find(item => item.conversation_id === activeConversationId.value) || null
   )
-
   const sortedConversations = computed(() =>
     [...conversations.value].sort((a, b) => b.updated_at.localeCompare(a.updated_at))
   )
 
+  function normalizeMessage(message: ChatMessage): ChatMessage {
+    return {
+      ...message,
+      citations: message.citations?.map(citation => ({
+        ...citation,
+        heading_path: Array.isArray(citation.heading_path)
+          ? citation.heading_path.join(' / ')
+          : citation.heading_path,
+      } as Citation)),
+    }
+  }
+
+  async function fetchAllConversations() {
+    const items: Conversation[] = []
+    while (true) {
+      const result = await listConversationsApi(items.length, 100)
+      items.push(...result.items)
+      if (!result.items.length || items.length >= result.page.total) return items
+    }
+  }
+
+  async function fetchAllMessages(conversationId: string) {
+    const items: ChatMessage[] = []
+    while (true) {
+      const result = await listConversationMessages(conversationId, items.length, 500)
+      items.push(...result.items)
+      if (!result.items.length || items.length >= result.page.total) return items
+    }
+  }
+
+  async function loadConversations(force = false) {
+    if (loading) return loading
+    if (initialized && !force) return
+    const version = ++loadVersion
+    messagesReady.value = false
+    loading = (async () => {
+      historyError.value = ''
+      try {
+        const items = await fetchAllConversations()
+        if (version !== loadVersion) return
+        conversations.value = items
+        initialized = true
+        const selected = activeConversationId.value && items.some(item => item.conversation_id === activeConversationId.value)
+          ? activeConversationId.value
+          : items[0]?.conversation_id || null
+        if (selected) await setActiveConversation(selected)
+        else { activeConversationId.value = null; messages.value = []; messagesReady.value = true }
+      } catch (error) {
+        if (version === loadVersion) historyError.value = error instanceof Error ? error.message : t('聊天记录加载失败', 'Failed to load chat history')
+      } finally {
+        loading = null
+      }
+    })()
+    return loading
+  }
+
   async function setActiveConversation(id: string) {
     stopGeneration()
+    const version = ++loadVersion
     activeConversationId.value = id
-    messages.value = history[id] ?? []
+    messagesReady.value = false
+    messages.value = []
+    historyError.value = ''
+    try {
+      const loadedMessages = await fetchAllMessages(id)
+      if (version === loadVersion && activeConversationId.value === id) {
+        messages.value = loadedMessages.map(normalizeMessage)
+        messagesReady.value = true
+      }
+    } catch (error) {
+      if (version === loadVersion) historyError.value = error instanceof Error ? error.message : t('消息加载失败', 'Failed to load messages')
+    }
+  }
+
+  function addLocalConversation(title: string) {
+    loadVersion++
+    const now = new Date().toISOString()
+    const conversation: Conversation = {
+      conversation_id: crypto.randomUUID(), title, created_at: now, updated_at: now, message_count: 0,
+    }
+    conversations.value.unshift(conversation)
+    activeConversationId.value = conversation.conversation_id
+    messages.value = []
+    messagesReady.value = true
+    return conversation
+  }
+
+  async function persistConversation(conversation: Conversation) {
+    const promise = createConversationApi(conversation).then(saved => {
+      const index = conversations.value.findIndex(item => item.conversation_id === saved.conversation_id)
+      if (index >= 0) Object.assign(conversations.value[index]!, saved)
+    }).catch(error => {
+      conversations.value = conversations.value.filter(item => item.conversation_id !== conversation.conversation_id)
+      if (activeConversationId.value === conversation.conversation_id) {
+        activeConversationId.value = null
+        messages.value = []
+      }
+      historyError.value = error instanceof Error ? error.message : t('会话创建失败', 'Failed to create conversation')
+      throw error
+    }).finally(() => pendingCreates.delete(conversation.conversation_id))
+    pendingCreates.set(conversation.conversation_id, promise)
+    return promise
+  }
+
+  async function createNewConversation() {
+    stopGeneration()
+    historyError.value = ''
+    const conversation = addLocalConversation(t('新对话', 'New conversation'))
+    try { await persistConversation(conversation) } catch { /* exposed through historyError */ }
   }
 
   async function sendMessage(text: string) {
-    if (!text.trim() || isStreaming.value || !selectedProviderId.value || !selectedModel.value) return
+    const content = text.trim()
-    const conversationId = activeConversationId.value || crypto.randomUUID()
+    if (!content || !canSend.value || !selectedProviderId.value || !selectedModel.value) return
-
+    const version = ++streamVersion
-    if (!activeConversationId.value) {
+    isPreparing.value = true
-      const newConv: Conversation = {
+    historyError.value = ''
-        conversation_id: conversationId,
+    let conversation = activeConversation.value
-        title: text.slice(0, 30),
+    try {
-        created_at: new Date().toISOString(),
+      if (!conversation) {
-        updated_at: new Date().toISOString(),
+        conversation = addLocalConversation(content.slice(0, 30))
-        message_count: 0,
+        await persistConversation(conversation)
+      } else if (pendingCreates.has(conversation.conversation_id)) {
+        await pendingCreates.get(conversation.conversation_id)
       }
-      conversations.value.unshift(newConv)
+    } catch { return }
-      activeConversationId.value = conversationId
+    finally {
+      if (version === streamVersion) isPreparing.value = false
     }
+    // Switching, stopping or deleting cancels sends still waiting for creation.
+    if (version !== streamVersion || activeConversationId.value !== conversation.conversation_id) return
 
-    history[conversationId] = messages.value
+    const conversationId = conversation.conversation_id
-    const conversationMessages = messages.value
+    if (conversation.message_count === 0) conversation.title = content.slice(0, 30)
     const userMsg: ChatMessage = {
-      message_id: crypto.randomUUID(),
+      message_id: crypto.randomUUID(), conversation_id: conversationId, role: 'user', content,
-      conversation_id: conversationId,
-      role: 'user',
-      content: text,
       created_at: new Date().toISOString(),
     }
-    messages.value.push(userMsg)
+    const aiMsg = reactive<ChatMessage>({
+      message_id: crypto.randomUUID(), conversation_id: conversationId, role: 'assistant', content: '',
+      created_at: new Date().toISOString(), citations: [], tool_calls: [],
+    })
+    messages.value.push(userMsg, aiMsg)
     inputText.value = ''
     isStreaming.value = true
-    const conversation = conversations.value.find(c => c.conversation_id === conversationId)
+    conversation.updated_at = new Date().toISOString()
-    if (conversation) { conversation.updated_at = new Date().toISOString(); conversation.message_count = messages.value.length }
+    conversation.message_count = messages.value.length
 
-    // 先插入占位消息，随后将 SSE 增量原位合并，避免每个 token 重建消息列表。
-    const aiMsg = reactive<ChatMessage>({
-      message_id: crypto.randomUUID(),
-      conversation_id: conversationId,
-      role: 'assistant',
-      content: '',
-      created_at: new Date().toISOString(),
-      citations: [],
-      tool_calls: [],
-    })
-    messages.value.push(aiMsg)
-
-    const version = ++streamVersion
     const argumentBuffers = new Map<string, string>()
     sseClient = streamChat({
       provider_id: selectedProviderId.value,
       model: selectedModel.value,
       conversation_id: conversationId,
+      user_message_id: userMsg.message_id,
+      assistant_message_id: aiMsg.message_id,
+      conversation_title: conversation.title,
       use_rag: useRag.value,
       messages: messages.value
-        .filter((message) => message.message_id !== aiMsg.message_id)
+        .filter(message => message.message_id !== aiMsg.message_id)
-        .map((message) => ({ role: message.role, content: message.content })),
+        .map(message => ({ role: message.role, content: message.content })),
     }, {
       onEvent(event) {
         if (version !== streamVersion) return
@@ -94,25 +208,21 @@ export const useChatStore = defineStore('chat', () => {
         if (event.event === 'ThinkingDelta') aiMsg.thinking = `${aiMsg.thinking ?? ''}${String(event.data.text ?? '')}`
         if (event.event === 'ToolCallStart') {
           aiMsg.tool_calls?.push({
-            tool_call_id: String(event.data.tool_call_id ?? ''),
+            tool_call_id: String(event.data.tool_call_id ?? ''), name: String(event.data.name ?? 'unknown'),
-            name: String(event.data.name ?? 'unknown'),
+            parameters: (event.data.arguments ?? {}) as Record<string, unknown>, status: 'running',
-            parameters: (event.data.arguments ?? {}) as Record<string, unknown>,
-            status: 'running',
           })
         }
         if (event.event === 'ToolCallDelta') {
-          const call = aiMsg.tool_calls?.find((item) => item.tool_call_id === event.data.tool_call_id)
+          const call = aiMsg.tool_calls?.find(item => item.tool_call_id === event.data.tool_call_id)
           if (call && typeof event.data.arguments_delta === 'string') {
             const buffer = (argumentBuffers.get(call.tool_call_id) ?? '') + event.data.arguments_delta
             argumentBuffers.set(call.tool_call_id, buffer)
             try { call.parameters = JSON.parse(buffer) } catch { /* incomplete JSON fragment */ }
           }
-          if (call && event.data.arguments && typeof event.data.arguments === 'object') {
+          if (call && event.data.arguments && typeof event.data.arguments === 'object') Object.assign(call.parameters, event.data.arguments)
-            Object.assign(call.parameters, event.data.arguments)
-          }
         }
         if (event.event === 'ToolCallEnd') {
-          const call = aiMsg.tool_calls?.find((item) => item.tool_call_id === event.data.tool_call_id)
+          const call = aiMsg.tool_calls?.find(item => item.tool_call_id === event.data.tool_call_id)
           if (call) call.status = 'completed'
         }
         if (event.event === 'Usage') {
@@ -128,21 +238,18 @@ export const useChatStore = defineStore('chat', () => {
             content: String(event.data.content ?? event.data.snippet ?? ''),
           })
         }
-        if (event.event === 'Error') aiMsg.content += `\n\n生成失败：${String(event.data.message ?? '未知错误')}`
+        if (event.event === 'Error') aiMsg.content += `\n\n${t('生成失败：', 'Generation failed: ')}${String(event.data.message ?? t('未知错误', 'Unknown error'))}`
       },
       onError(error) {
         if (version !== streamVersion) return
-        aiMsg.content += `\n\n连接失败：${error.message}`
+        aiMsg.content += `\n\n${t('连接失败：', 'Connection failed: ')}${error.message}`
         isStreaming.value = false
         sseClient = null
       },
       onDone() {
         if (version !== streamVersion) return
-        const conversation = conversations.value.find((item) => item.conversation_id === conversationId)
+        conversation!.message_count = messages.value.length
-        if (conversation) {
+        conversation!.updated_at = new Date().toISOString()
-          conversation.message_count = conversationMessages.length
-          conversation.updated_at = new Date().toISOString()
-        }
         isStreaming.value = false
         sseClient = null
       },
@@ -151,57 +258,35 @@ export const useChatStore = defineStore('chat', () => {
 
   function stopGeneration() {
     streamVersion++
-    if (sseClient) {
+    isPreparing.value = false
-      sseClient.cancel()
+    if (sseClient) { sseClient.cancel(); sseClient = null }
-      sseClient = null
-    }
     isStreaming.value = false
   }
 
-  function createNewConversation() {
+  async function deleteConversation(id: string) {
-    stopGeneration()
+    if (deletingConversations.has(id)) return
-    const newConv: Conversation = {
+    deletingConversations.add(id)
-      conversation_id: crypto.randomUUID(),
-      title: '新对话',
-      created_at: new Date().toISOString(),
-      updated_at: new Date().toISOString(),
-      message_count: 0,
-    }
-    conversations.value.unshift(newConv)
-    activeConversationId.value = newConv.conversation_id
-    history[newConv.conversation_id] = []
-    messages.value = history[newConv.conversation_id]
-  }
-
-  function deleteConversation(id: string) {
     if (activeConversationId.value === id) stopGeneration()
-    delete history[id]
+    historyError.value = ''
-    const idx = conversations.value.findIndex((c) => c.conversation_id === id)
+    try {
-    if (idx > -1) {
+      if (pendingCreates.has(id)) await pendingCreates.get(id)
-      conversations.value.splice(idx, 1)
+      await removeConversation(id)
+      conversations.value = conversations.value.filter(item => item.conversation_id !== id)
       if (activeConversationId.value === id) {
-        activeConversationId.value = conversations.value[0]?.conversation_id || null
+        const next = sortedConversations.value[0]
-        messages.value = conversations.value[0] ? history[conversations.value[0].conversation_id] || [] : []
+        if (next) await setActiveConversation(next.conversation_id)
+        else { loadVersion++; activeConversationId.value = null; messages.value = []; messagesReady.value = true }
       }
+    } catch (error) {
+      historyError.value = error instanceof Error ? error.message : t('会话删除失败', 'Failed to delete conversation')
+    } finally {
+      deletingConversations.delete(id)
     }
   }
 
   return {
-    conversations,
+    conversations, activeConversationId, activeConversation, sortedConversations, messages,
-    activeConversationId,
+    isStreaming, isPreparing, canSend, inputText, useRag, selectedSkillId, selectedProviderId, selectedModel, historyError,
-    activeConversation,
+    loadConversations, setActiveConversation, sendMessage, stopGeneration, createNewConversation, deleteConversation,
-    sortedConversations,
-    messages,
-    isStreaming,
-    inputText,
-    useRag,
-    selectedSkillId,
-    selectedProviderId,
-    selectedModel,
-    setActiveConversation,
-    sendMessage,
-    stopGeneration,
-    createNewConversation,
-    deleteConversation,
   }
 })

frontend/src/stores/editor.ts

@@ -2,6 +2,7 @@ import { defineStore } from 'pinia'
 import { ref, computed } from 'vue'
 import type { SaveStatus } from '@/contracts'
 import * as workspaceService from '@/services/workspaceService'
+import { t } from '@/i18n'
 
 export const useEditorStore = defineStore('editor', () => {
   const mode = ref<'wysiwyg' | 'source'>('wysiwyg')
@@ -76,12 +77,12 @@ export const useEditorStore = defineStore('editor', () => {
       saveTimer = null
     }
     if (saveStatus.value === 'conflict') {
-      throw new Error('当前文件存在编辑冲突，请处理后再切换文件。')
+      throw new Error(t('当前文件存在编辑冲突，请处理后再切换文件。', 'The current file has an editing conflict. Resolve it before switching files.'))
     }
     if (pendingSave) await pendingSave
     if (saveStatus.value === 'dirty' || saveStatus.value === 'save_failed') await save()
     if (saveStatus.value === 'dirty' || saveStatus.value === 'save_failed') {
-      throw new Error('当前文件保存失败，已阻止切换以避免内容丢失。')
+      throw new Error(t('当前文件保存失败，已阻止切换以避免内容丢失。', 'The current file could not be saved. Switching was blocked to prevent data loss.'))
     }
     // 版本号使较慢的旧读取不能覆盖用户后选择的新文件。
     const version = ++loadVersion

frontend/src/stores/plugin.ts

@@ -2,6 +2,7 @@ import { defineStore } from 'pinia'
 import { ref, computed } from 'vue'
 import type { Plugin } from '@/contracts'
 import * as pluginService from '@/services/pluginService'
+import { t } from '@/i18n'
 
 export const usePluginStore = defineStore('plugin', () => {
   const plugins = ref<Plugin[]>([])
@@ -23,7 +24,7 @@ export const usePluginStore = defineStore('plugin', () => {
       plugins.value = await pluginService.listPlugins()
       error.value = null
     } catch (reason) {
-      error.value = reason instanceof Error ? reason.message : 'Plugin 加载失败'
+      error.value = reason instanceof Error ? reason.message : t('Plugin 加载失败', 'Failed to load Plugins')
     } finally {
       isLoading.value = false
     }

frontend/src/stores/provider.ts

@@ -3,6 +3,7 @@ import { ref, computed } from 'vue'
 import type { ProviderConfig, ModelInfo, ProviderPreset } from '@/contracts'
 import { createProvider, deleteProvider as deleteProviderRequest, getCredentialStatus, listModels, listProviderPresets, listProviders, putCredential, testProvider as testProviderRequest, updateProvider as updateProviderRequest } from '@/services/providerService'
 import { ApiErrorClass } from '@/services/apiClient'
+import { t } from '@/i18n'
 
 export const useProviderStore = defineStore('provider', () => {
   const providers = ref<ProviderConfig[]>([])
@@ -29,7 +30,7 @@ export const useProviderStore = defineStore('provider', () => {
       }
       error.value = null
     } catch (reason) {
-      error.value = reason instanceof Error ? reason.message : 'Provider 加载失败'
+      error.value = reason instanceof Error ? reason.message : t('Provider 加载失败', 'Failed to load Providers')
     } finally {
       isLoading.value = false
     }
@@ -39,7 +40,7 @@ export const useProviderStore = defineStore('provider', () => {
     try {
       presets.value = await listProviderPresets()
     } catch (reason) {
-      error.value = reason instanceof Error ? reason.message : 'Provider 预设加载失败'
+      error.value = reason instanceof Error ? reason.message : t('Provider 预设加载失败', 'Failed to load Provider presets')
     }
   }
 
@@ -55,11 +56,11 @@ export const useProviderStore = defineStore('provider', () => {
     } catch (reason) {
       const provider = providers.value.find((item) => item.provider_id === providerId)
       const credentialId = provider?.credential_id
-      let message = reason instanceof Error ? reason.message : '模型列表获取失败'
+      let message = reason instanceof Error ? reason.message : t('模型列表获取失败', 'Failed to load the model list')
       if (reason instanceof ApiErrorClass && reason.code === 'PROVIDER_CREDENTIAL_MISSING') {
-        message = '尚未配置 API Key，请编辑该 Provider 后填写并保存。'
+        message = t('尚未配置 API Key，请编辑该 Provider 后填写并保存。', 'No API key is configured. Edit this Provider, enter a key, and save it.')
       } else if (reason instanceof ApiErrorClass && reason.code === 'PROVIDER_AUTH_FAILED') {
-        message = `鉴权失败，请检查凭据“${credentialId || '未设置'}”对应的 API Key 是否有效。`
+        message = t(`鉴权失败，请检查凭据“${credentialId || '未设置'}”对应的 API Key 是否有效。`, `Authentication failed. Check the API key for credential “${credentialId || 'not set'}”.`)
       }
       modelErrorsByProvider.value[providerId] = message
       throw reason

frontend/src/stores/search.ts

@@ -3,6 +3,7 @@ import { ref } from 'vue'
 import type { SearchResult, SearchRequest } from '@/contracts'
 import * as searchService from '@/services/searchService'
 import { ApiErrorClass } from '@/services/apiClient'
+import { t } from '@/i18n'
 
 const VECTOR_ERROR_CODES = new Set([
   'SEMANTIC_INDEX_UNAVAILABLE',
@@ -28,7 +29,7 @@ export const useSearchStore = defineStore('search', () => {
       if (version !== historyVersion) return
       recentQueries.value = response.queries
       historyError.value = ''
-    } catch { if (version === historyVersion) historyError.value = '无法读取应用搜索记录，请检查后端连接。' }
+    } catch { if (version === historyVersion) historyError.value = t('无法读取应用搜索记录，请检查后端连接。', 'Could not load search history. Check the backend connection.') }
   }
   async function clearHistory() {
     const version = ++historyVersion
@@ -36,7 +37,7 @@ export const useSearchStore = defineStore('search', () => {
       await searchService.clearHistory()
       if (version !== historyVersion) return
       recentQueries.value = []; historyError.value = ''
-    } catch { if (version === historyVersion) historyError.value = '清空搜索记录失败，请重试。' }
+    } catch { if (version === historyVersion) historyError.value = t('清空搜索记录失败，请重试。', 'Failed to clear search history. Please retry.') }
   }
   const error = ref<string | null>(null)
   const vectorUnavailable = ref(false)
@@ -71,12 +72,12 @@ export const useSearchStore = defineStore('search', () => {
           selectedIndex.value = 0
         } catch (fallbackError) {
           if (version !== searchVersion) return
-          error.value = fallbackError instanceof Error ? fallbackError.message : '全文检索降级失败'
+          error.value = fallbackError instanceof Error ? fallbackError.message : t('全文检索降级失败', 'Full-text search fallback failed')
           results.value = []
           total.value = 0
         }
       } else {
-        error.value = reason instanceof Error ? reason.message : '搜索失败'
+        error.value = reason instanceof Error ? reason.message : t('搜索失败', 'Search failed')
         results.value = []
         total.value = 0
       }

frontend/src/stores/settings.ts

@@ -5,6 +5,7 @@ import { resolveApiUrl } from '@/services/apiClient'
 import packageInfo from '../../package.json'
 import * as indexService from '@/services/indexService'
 import * as systemService from '@/services/systemService'
+import { appLocale, t } from '@/i18n'
 
 export const useSettingsStore = defineStore('settings', () => {
   const saved = (() => {
@@ -14,9 +15,9 @@ export const useSettingsStore = defineStore('settings', () => {
   // General
   const restoreLastVault = ref(saved.restoreLastVault !== false)
   const autoSaveInterval = ref(typeof saved.autoSaveInterval === 'number' ? saved.autoSaveInterval : 1500)
-  const language = ref<'zh-CN' | 'en'>(saved.language === 'en' ? 'en' : 'zh-CN')
+  const language = appLocale
   const appVersion = ref(packageInfo.version)
-  const aiCoreVersion = ref('未获取')
+  const aiCoreVersion = ref('—')
 
   // Editor
   const defaultEditorMode = ref<'wysiwyg' | 'source'>(saved.defaultEditorMode === 'source' ? 'source' : 'wysiwyg')
@@ -47,10 +48,10 @@ export const useSettingsStore = defineStore('settings', () => {
     ])
     const [health, status, index, policy] = results
     aiCoreStatus.value = health.status === 'fulfilled' && health.value.status === 'ok' ? 'running' : 'error'
-    aiCoreVersion.value = status.status === 'fulfilled' ? status.value.version : '未获取'
+    aiCoreVersion.value = status.status === 'fulfilled' ? status.value.version : '—'
     indexStatus.value = index.status === 'fulfilled' ? index.value : emptyIndex()
     permissionPolicy.value = policy.status === 'fulfilled' ? policy.value : {}
-    diagnosticsError.value = results.filter(item => item.status === 'rejected').map(item => item.reason instanceof Error ? item.reason.message : '后端请求失败').join('；') || null
+    diagnosticsError.value = results.filter(item => item.status === 'rejected').map(item => item.reason instanceof Error ? item.reason.message : t('后端请求失败', 'Backend request failed')).join(t('；', '; ')) || null
   }
 
   function setAutoSaveInterval(ms: number) {
@@ -68,7 +69,7 @@ export const useSettingsStore = defineStore('settings', () => {
       indexStatus.value = await indexService.getIndexStatus()
     } catch (reason) {
       indexStatus.value.status = 'error'
-      indexStatus.value.error = reason instanceof Error ? reason.message : '索引重建失败'
+      indexStatus.value.error = reason instanceof Error ? reason.message : t('索引重建失败', 'Index rebuild failed')
     }
   }
 

frontend/src/stores/skill.ts

@@ -2,6 +2,7 @@ import { defineStore } from 'pinia'
 import { ref, computed } from 'vue'
 import type { Skill } from '@/contracts'
 import * as skillService from '@/services/skillService'
+import { t } from '@/i18n'
 
 export const useSkillStore = defineStore('skill', () => {
   const skills = ref<Skill[]>([])
@@ -23,7 +24,7 @@ export const useSkillStore = defineStore('skill', () => {
       skills.value = await skillService.listSkills()
       error.value = null
     } catch (reason) {
-      error.value = reason instanceof Error ? reason.message : 'Skill 加载失败'
+      error.value = reason instanceof Error ? reason.message : t('Skill 加载失败', 'Failed to load Skills')
     } finally {
       isLoading.value = false
     }

frontend/src/stores/task.ts

@@ -2,6 +2,7 @@ import { defineStore } from 'pinia'
 import { ref, computed } from 'vue'
 import type { TaskItem, TaskStatus, TaskPriority, TaskSource } from '@/contracts'
 import { createTask as createTaskRequest, deleteTask as deleteTaskRequest, listTasks, updateTask as updateTaskRequest } from '@/services/taskService'
+import { t } from '@/i18n'
 
 export const useTaskStore = defineStore('task', () => {
   const tasks = ref<TaskItem[]>([])
@@ -31,7 +32,7 @@ export const useTaskStore = defineStore('task', () => {
       tasks.value = resp.items
       error.value = null
     } catch (reason) {
-      error.value = reason instanceof Error ? reason.message : '任务加载失败'
+      error.value = reason instanceof Error ? reason.message : t('任务加载失败', 'Failed to load tasks')
     } finally {
       isLoading.value = false
     }

frontend/src/stores/theme.ts

@@ -1,11 +1,12 @@
 import { defineStore } from 'pinia'
 import { ref, computed, watch } from 'vue'
 import type { ThemeConfig } from '@/contracts'
+import { t } from '@/i18n'
 
-const builtinThemes: ThemeConfig[] = [
+const builtinThemes = (): ThemeConfig[] => [
-  { theme_id: 'light', name: '浅色', version: '1.0.0', description: '默认浅色主题', is_dark: false, builtin: true, code_theme: 'github-light' },
+  { theme_id: 'light', name: t('浅色', 'Light'), version: '1.0.0', description: t('默认浅色主题', 'Default light theme'), is_dark: false, builtin: true, code_theme: 'github-light' },
-  { theme_id: 'dark', name: '深色', version: '1.0.0', description: '默认深色主题', is_dark: true, builtin: true, code_theme: 'github-dark' },
+  { theme_id: 'dark', name: t('深色', 'Dark'), version: '1.0.0', description: t('默认深色主题', 'Default dark theme'), is_dark: true, builtin: true, code_theme: 'github-dark' },
-  { theme_id: 'sepia', name: '护眼', version: '1.0.0', description: '护眼暖色调', is_dark: false, builtin: true, code_theme: 'github-light' },
+  { theme_id: 'sepia', name: t('护眼', 'Sepia'), version: '1.0.0', description: t('护眼暖色调', 'Warm, low-glare theme'), is_dark: false, builtin: true, code_theme: 'github-light' },
 ]
 
 export type CodeBlockThemePreference = 'auto' | 'github-light' | 'github-dark'
@@ -15,7 +16,7 @@ function isCodeBlockThemePreference(value: unknown): value is CodeBlockThemePref
 }
 
 export const useThemeStore = defineStore('theme', () => {
-  const themes = ref<ThemeConfig[]>(builtinThemes)
+  const themes = computed<ThemeConfig[]>(builtinThemes)
   const currentThemeId = ref<string>('light')
   const fontEditorSize = ref(15)
   const fontEditorFamily = ref('system-ui')

frontend/src/styles/features.css

@@ -85,6 +85,44 @@ button:disabled { cursor: not-allowed; opacity: .55; box-shadow: none; transform
 .input:focus, .select:focus, .textarea:focus { border-color: var(--color-border-focus); box-shadow: 0 0 0 3px var(--color-accent-soft); background: var(--color-surface-primary); }
 .form-grid { display: grid; grid-template-columns: repeat(auto-fit, minmax(180px, 1fr)); gap: var(--space-md); }
 
+input[type='checkbox'], input[type='radio'] {
+  appearance: none;
+  display: inline-grid;
+  place-content: center;
+  width: 18px;
+  height: 18px;
+  flex: 0 0 18px;
+  border: 1px solid var(--color-border-default);
+  background: var(--color-background-primary);
+  transition: border-color var(--motion-fast), background-color var(--motion-fast), box-shadow var(--motion-fast);
+}
+input[type='checkbox'] { border-radius: 5px; }
+input[type='radio'] { border-radius: 50%; }
+input[type='checkbox']::before, input[type='radio']::before { content: ''; width: 10px; height: 10px; transform: scale(0); transition: transform var(--motion-fast); }
+input[type='checkbox']::before { clip-path: polygon(14% 44%, 0 59%, 39% 96%, 100% 20%, 84% 7%, 37% 68%); background: var(--color-text-inverse); }
+input[type='radio']::before { border-radius: 50%; background: var(--color-text-inverse); }
+input[type='checkbox']:hover, input[type='radio']:hover { border-color: var(--color-accent-primary); }
+input[type='checkbox']:checked, input[type='radio']:checked { border-color: var(--color-accent-primary); background: var(--color-accent-primary); }
+input[type='checkbox']:checked::before, input[type='radio']:checked::before { transform: scale(1); }
+input[type='checkbox']:disabled, input[type='radio']:disabled { cursor: not-allowed; opacity: .55; }
+
+input[type='range'] { appearance: none; height: 20px; background: transparent; cursor: pointer; }
+input[type='range']::-webkit-slider-runnable-track { height: 5px; border-radius: var(--radius-full); background: var(--color-background-tertiary); }
+input[type='range']::-webkit-slider-thumb { appearance: none; width: 16px; height: 16px; margin-top: -5.5px; border: 2px solid var(--color-surface-primary); border-radius: 50%; background: var(--color-accent-primary); box-shadow: 0 1px 4px color-mix(in srgb, var(--color-text-primary) 24%, transparent); }
+
+progress { appearance: none; width: 100%; height: 8px; overflow: hidden; border: 0; border-radius: var(--radius-full); background: var(--color-background-tertiary); }
+progress::-webkit-progress-bar { border-radius: var(--radius-full); background: var(--color-background-tertiary); }
+progress::-webkit-progress-value { border-radius: var(--radius-full); background: linear-gradient(90deg, var(--color-accent-primary), var(--color-accent-secondary)); }
+progress:not([value]) { background: linear-gradient(90deg, var(--color-background-tertiary) 25%, var(--color-accent-secondary) 50%, var(--color-background-tertiary) 75%); background-size: 200% 100%; animation: progress-pulse 1.2s linear infinite; }
+
+.ui-disclosure { padding: var(--space-md); border: 1px solid var(--color-border-default); border-radius: var(--radius-md); background: var(--color-background-secondary); }
+.ui-disclosure > summary { display: flex; align-items: center; justify-content: space-between; gap: var(--space-sm); list-style: none; color: var(--color-text-secondary); font-weight: 600; cursor: pointer; }
+.ui-disclosure > summary::-webkit-details-marker { display: none; }
+.ui-disclosure > summary::after { content: ''; width: 8px; height: 8px; flex: 0 0 auto; border-right: 2px solid currentColor; border-bottom: 2px solid currentColor; transform: rotate(45deg); transition: transform var(--motion-fast); }
+.ui-disclosure[open] > summary::after { transform: rotate(225deg); }
+.ui-disclosure[open] > summary { margin-bottom: var(--space-md); color: var(--color-text-primary); }
+.ui-disclosure > :not(summary) + :not(summary) { margin-top: var(--space-sm); }
+
 .badge {
   display: inline-flex;
   align-items: center;
@@ -143,6 +181,8 @@ button:disabled { cursor: not-allowed; opacity: .55; box-shadow: none; transform
   to { opacity: 1; transform: translateY(0) scale(1); }
 }
 
+@keyframes progress-pulse { from { background-position: 100% 0; } to { background-position: -100% 0; } }
+
 @media (max-width: 900px) {
   .feature-page { padding: var(--space-lg); }
   .split-view { grid-template-columns: 1fr; }