fix(usage): 列表支持显式模板查询（Sub2API/NewAPI）——对齐后端绕过 usage_kinds 的口径 — bump to v0.7.22

feat(models-dev): 高级设置新增 models.dev 在线同步 — 上下文/能力/单价即时生效 + 一键恢复内置；修复 public 快照构建漂移；确认适配 kimi-code 2.1.0 — bump to v0.7.21
chore(data): 同步 models-dev 快照至 2026-09-20（7870 模型 / 222 供应商）
2026-09-29 10:42:13 +08:00 · 2026-09-24 00:19:10 +08:00 · 2026-09-20 10:29:40 +08:00 · 2026-09-20 09:58:01 +08:00 · 2026-09-20 09:49:49 +08:00 · 2026-09-16 14:14:29 +08:00
72 changed files with 285320 additions and 259804 deletions
@@ -73,7 +73,8 @@ jobs:
        id: notes
        shell: bash
        run: |
-          FILE="release-notes-${{ github.ref_name }}.md"
+          FILE="docs/release-notes/${{ github.ref_name }}.md"
+          [ -f "$FILE" ] || FILE="release-notes-${{ github.ref_name }}.md"
          {
            echo 'body<<RELEASE_BODY_EOF'
            if [ -f "$FILE" ]; then
@@ -0,0 +1,151 @@
+# Changelog
+
+**Kimi Switch** — Windows 桌面端的 Kimi Code CLI 配置管理器，统一管理多家 LLM 供应商、模型、图标、连通性、用量统计与版本更新。
+
+本文件汇总各版本要点，按版本倒序排列；每个版本的**完整 release notes**（含安装说明、平台产物、已知问题）见 [`docs/release-notes/`](./docs/release-notes/)。
+
+---
+
+## v0.7.22 (2026-09-24)
+
+- **修复 Sub2API / NewAPI 模板查询不显示**：为无预设识别类型的自定义供应商配置 Sub2API（或 NewAPI）模板后，弹窗测试可查但供应商列表始终不显示——前端列表的查询/显示门槛只看预设 `usageKinds`，显式模板被漏掉；现在与后端口径对齐（模板查询绕过 usage_kinds），列表正常显示余额 / 配额
+
+## v0.7.21 (2026-09-24)
+
+- **models.dev 在线同步**：高级设置新增「模型参考数据」卡片——一键在线拉取 models.dev 最新快照（上下文长度 / 能力标记 / 单价），前端与仪表盘计价即时生效，无需等待新版本；支持一键恢复随版本内置数据；显示当前数据来源与版本（同步 / 内置 + 日期 + 模型/供应商数）
+- **修复构建链路数据漂移**：`fetch-models-dev.mjs` 现在同时写入 `src/lib/models-dev.json`（Rust 计价内嵌）与 `public/models-dev.json`（前端静态资源）——此前 public 副本需手工拷贝，已两次滞后于 src/lib
+- **kimi-code 2.1.0 已适配确认**：实验 flag 注册表（5 个）、config 模式、用量/quota 格式均无变化；上游 `[watch]` 默认关闭与现有开关行为一致；新增 `tui_mode`（tui.toml）不涉及配置管理面
+- models-dev 快照同步至 2026-09-23（8,126 模型 / 223 供应商）
+
+## v0.7.20 (2026-09-20)
+
+- **套餐用量阈值预警**：用量配置新增「预警阈值 %」（0-100，0/空为关闭），任一用量窗口达到阈值即红色高亮 + ⚠ 标记，低于阈值自动消除
+- **Kimi For Coding 加力钱包**：0.43.1 起解析 `boosterWallet` 并展示为独立余额行（💰 月度用量 / 总额 / 余额，含币种）
+- **供应商凭证环境变量化（`api_key_env`）**：适配 kimi-code 2.0.0 #3762，凭证来源可切换「直接填写 / 环境变量」，密钥不写入 `config.toml`，与 `api_key` 互斥；SQLite 新增列含自动迁移
+- **Kimi Code 版本跟踪器**：设置页新增版本卡片——本机已装版本、上游最新版本、适配状态徽标（绿已适配 / 黄建议升级 / 橙部分兼容 / 灰未安装），可一键打开上游 Releases
+- **供应商一键体检**：并发探活所有启用供应商的推理端点（`GET /v1/models`）并顺带重查账单，三态判定避免误报（端点不支持模型列表计中性）
+- 新增 `loop_control.compaction_max_attempts`（上游 0.43.0 #3750，默认 5）与 `[watch] enabled` 开关（上游 2.0.1 #3892）；已确认适配 kimi-code 2.0.2
+
+## v0.7.19 (2026-09-14)
+
+- **适配 kimi-code 0.43.0 实验 flag 镜像（7→5）**：移除已转正的 `auto_session_title`（上游 #3749），上游注册表现剩 `wait_for` / `tool-select` / `notify_user` / `tower` / `subagent_fork`
+- 兼容性核查：0.43.0 的 wire 持久化重建（#3737）不影响用量统计，相关记录全部保留；`loop_control.compaction_max_attempts`（#3750）保存不丢失
+- 残留的 `auto_session_title` 旧键会被上游静默忽略，无需手动清理
+
+## v0.7.18 (2026-09-14)
+
+- **会话管理「一键归档」**：支持按 1 个月前 / 半个月前 / 1 星期前 / 指定日期批量归档，替代逐个选择，并反馈已归档 / 跳过数量
+- **归档用量快照落库**：归档瞬间按「天 × 模型」粒度把 token 用量与费用写入 SQLite `archived_sessions` 表
+- 删除已归档会话释放磁盘后不再损失历史统计——趋势图、热力图、模型统计、区间合计自动合并快照
+- 快照与现存文件实时统计严格去重（只合并文件已删除的归档会话，绝不双计）
+- models-dev 快照同步至 2026-09-14（7784 模型 / 213 供应商），pricing 断言跟随 DeepSeek 官方调价（deepseek-v4-flash input 0.14→0.15 / output 0.28→0.60）
+
+## v0.7.17 (2026-09-13)
+
+- **账单查询适配 Sub2API 中转站（`balance:sub2api`）**：支持 Wei-Shaw/sub2api 面板余额查询（`GET {base}/v1/usage`），自动复用推理 `sk-` API Key，无需网页后台 Access Token
+- 解析三种模式：单 Key 总额度（quota）、5 小时 / 每日 / 7 天速率窗口（含重置倒计时）、订阅组日/周/月额度与钱包余额（USD）
+- `detect_provider` 家族规则拆分：`codingplan.site` 主域识别为 Sub2API，`ai.codingplan.site` 识别为 NewAPI，修正主域被误判导致查询 404 的问题
+- 用量配置面板新增「Sub2API 中转站」模板，可选覆盖查询地址（Base URL）
+- **模型映射「获取模型列表」新增批量选择**：全选 / 取消全选 / 反选，配合已添加标记批量接入中转站模型
+
+## v0.7.16 (2026-09-10)
+
+- **适配 kimi-code 0.42.0 实验 flag 镜像（9→6）**：移除 4 个已转正/删除的开关（子代理次主力模型、MiniDB 读模型、远程控制、搜索 Worker），新增 `notify_user`（实验性 Updates 面板，默认关闭）
+- 用量统计「模型用量统计」图表固定最大高度 360px 并支持滚动，与右侧模型用量表一致
+- models-dev 快照同步至 2026-09-10（7615 模型 / 213 供应商），`public/` 静态目录同步
+
+## v0.7.15 (2026-09-09)
+
+- **用量统计新增「昨天」选项卡**：时间范围为 今天 / 昨天 / 7 天 / 30 天 / 全部，「昨天」按本地时区严格统计昨日 0 点至今日 0 点，概览卡扩展为 5 张
+- **新增「模型用量统计」趋势选项卡**：跨供应商按模型名汇总（同一模型合并为一行），横向条形图按 Token 降序，悬停显示请求数 / Token / 缓存命中 / 费用
+- **修复 opencode Go 套餐子代理请求 400**：Go 网关强制要求 `x-opencode-session` 头，导出时凡指向 Go 端点的供应商自动补写（已有自定义值保留），按量计费的 `zen/v1` 不受影响
+- models-dev 快照同步至 2026-09-09（7583 模型 / 213 供应商）；`kimi/k2.5` 官方条目下架后暂解析到 302ai 转售价（0.66 / 3.3）
+
+## v0.7.14 (2026-09-06)
+
+- **修复关闭「启用用量查询」后仍残留查询行为**：此前关闭后前端仍发起查询并持续显示「查询失败 · 用量查询已关闭」红色错误行，现在关闭即停止查询并隐藏整行用量信息，重开后自动恢复
+- 在途查询结果在开关切换瞬间作废且不再写入共享缓存，避免过期失败状态在 5 分钟内复活
+
+## v0.7.13 (2026-09-05)
+
+- **适配 kimi-code 0.41.0**：上游移除 `file_history` flag（轮级文件历史转为恒开），镜像全链路同步删除；Auto（永不询问）模式不再拦截危险命令，设置页文案（中英）同步修正
+- **账单查询改进**：`codingplan.site` 系供应商自动识别为 NewAPI 余额查询；凭据缺失不再误报「网络异常」而给出本地化配置提示；金额显示统一为两位小数并去掉 `¤` 占位符
+- 新增 usage 错误本地化单元测试
+
+## v0.7.12 (2026-09-03)
+
+- **适配 kimi-code 0.40.1（flag 9→10）**：新增 `file_history` / `search_worker`；`secondary-model` 默认开启并显示徽章；新增危险命令守卫开关 `[permission] dangerous_command_guard`（默认开启）
+- **flag 优先级语义修正**：单 flag 环境变量 > `[experimental]` 显式配置 > 总开关环境变量 > 默认值，总开关不再锁定全部开关
+- **修复 WebUI 打开竞态**：并发调用串行化 + 后到点击自动聚焦已有窗口，失败路径不再误杀正在使用的 server，前端增加「打开中...」防重入态
+- models-dev 快照更新至 7495 模型 / 212 供应商
+
+## v0.7.11 (2026-08-29)
+
+- **适配 kimi-code 0.39.1**：`[thinking]` 保留开关改写 `keep = "off"`（修复布尔值导致整个节被 v2 校验丢弃）；`loop_control` 默认只写新键 `max_attempts_per_step` 并清理旧键；思考强度移除 `max` 档（上游迁移为 `high`）
+- **会话归档与上游 v2 对齐**：不再搬动文件，改为在 `state.json` 写入 `archived` / `archivedAt` 元数据，CLI 侧与 Kimi Switch 侧归档状态互通；旧 `.kcd-archive/` 物理归档仍可识别恢复
+- **修复子代理模型池无法添加第二个条目**：默认模型自动物化进池，不再被校验拦截
+- models.dev 快照刷新至 7482 个模型 / 207 个供应商
+
+## v0.7.10 (2026-08-29)
+
+- **修复保存配置时误删 CLI 新增的顶层配置节**：以导入时顶层键为基线，CLI 后加的 `[task]` / `[swarm]` / `[cron]` / `[tools]` / `[identity]` / `[token_counting]` 等节原样保留，界面中显式删除的节仍会移除
+- **修复未知供应商类型被改写**：`config.toml` 中新增的供应商 `type` 往返读写后原样保留（此前会被改成 `kimi`），下拉框标注「未知类型」，SQLite 快照恢复路径同步修复
+- 新增 3 个配置往返回归测试，Rust 测试 106 项全绿，前端 tsc / vitest 通过
+
+## v0.7.9 (2026-08-27)
+
+- 模型库刷新至 **7343 个模型 / 203 个供应商**，内置快照与官网静态资源同步更新
+- **新增 GLM-5.3 / GLM-5.3-Flash 全线接入**：智谱官方、OpenRouter、Cloudflare Workers AI、DeepInfra、HuggingFace、火山引擎等约 20 条渠道，GLM-5.3-Flash 定价 $0.075/$0.25（每百万 token）
+- 新增 Qwen3.8 系列、豆包 Seed 2.x 全系、DeepSeek-V4 GA、MiniMax-M2.7/M3 等共 74 个模型条目；价格更新 58 处、能力信息修正 56 处
+
+## v0.7.8 (2026-08-25)
+
+- **修复模型用量上下显示不同步**：上方用量摘要刷新后下方进度条仍停留旧数据（如 24% vs 8%），用量查询状态提升至卡片层统一持有，两处同帧更新
+- **实验功能开关同步 kimi-code 上游 7 项旗标**：新增 Tower 模式、子代理 Fork 上下文、WaitFor 工具、自动会话标题；移除已废弃的 ACP v2 开关（手写配置仍完整保留）
+- 自动刷新间隔改为每卡片恰好一份定时器，行为不变
+
+## v0.7.7 (2026-08-22)
+
+- **双区域 OAuth 支持（适配 kimi-code 0.38.0）**：适配 mainland-cn `auth.kimi.com` / global `auth.kimi.ai`（#2862），按 provider 的 oauth ref 推导凭据文件与刷新端点，global 账号（`credentials/kimi-code-env-<sha256>.json`）可正常查询用量
+- **内置登录区域选择**：应用内 Kimi 登录对话框新增「中国大陆 / 国际版 (kimi.ai)」，登录成功后按官方 CLI 行为 provision `[providers."managed:kimi-code"]`（global 写 `oauthHost`，cn 不写）
+- **修复**：手动添加模型后别名自动跟随为 `<provider>/<model-id>`；managed（OAuth）供应商自填 `api_key` 保存时不再被清空
+- models.dev 快照刷新至 **7246 模型 / 193 供应商**，新增 DeepSeek V4 Flash Vision Exp、Ox Alpha Free（opencode-go，免费）
+
+## v0.7.6 (2026-08-18)
+
+- **新增 OpenCode Go 套餐用量查询**：卡片用量页脚显示 **5 小时滚动 / 7 天 / 30 天** 三窗口的已用百分比与重置时间（数据源 `GET https://opencode.ai/zen/go/v1/usage`）
+- 已配置 OpenCode Go API Key 的老用户无需手动设置，程序按 `base_url` 自动识别并启用；已适配 opencode.ai 的 Cloudflare 拦截（携带浏览器 User-Agent）
+- 插件目录被运行中进程占用时的错误提示给出明确指引
+
+## v0.7.5 (2026-08-15)
+
+- **子代理模型池适配 kimi-code 0.36.0 新引擎**：适配 `[secondary_model]` 的 `default_model` + `[secondary_model.models]` 表 + `force`；写入时 `model` 与 `default_model` 双写同值，v1 legacy 与 v2 引擎均可识别
+- **高级设置页新增模型池管理**：升级旧配置、增删池条目、编辑路由描述（渲染进主代理 Agent/AgentSwarm 工具描述）；支持 force 强制默认模型（带二次确认）；保存前 6 类错误前置拦截
+- 新增 32 个单元测试覆盖池读写与校验逻辑（vitest）
+- **热力图详情改为按需加载**：双击方块调用后端 `get_day_detail` 聚合，不再受 7d / 30d 范围限制
+- models.dev 快照刷新至 2026-08-15（6583 模型 / 185 供应商），新增 GLM-5.3 等
+
+## v0.7.4 (2026-08-09)
+
+- **热力图方块双击弹窗**：查看当日模型用量分布，每种模型独立展示 Token 用量 / 请求次数 / 费用 / 缓存命中率（此前仅有 token 数）
+- 后端 `[dashboard] by_model` 从纯 token 数升级为结构化对象（含 requests / cost / cacheHitRate），弹窗与柱状图双击均展示全量指标
+- 范围外日期（不在当前 7d / 30d / all 内）不可双击；Escape / 遮罩 / 关闭按钮均可关闭弹窗
+
+## v0.7.3 (2026-08-07)
+
+- **Kimi Code WebUI 应用内嵌窗口**：新版 Web 界面以独立顶层窗口打开（1100×750、可缩放、居中），不再跳转外部浏览器
+- **单例 + 服务器自动管理**：窗口最多一个，重复点击自动聚焦；复用已运行的 `kimi web` 服务，自启动的服务器在窗口关闭或应用退出时自动清理
+- 「在应用内打开」与「在浏览器打开」两个入口并存（后者复用本地服务器，直接打开 `http://127.0.0.1:58627`）
+
+## v0.7.2 (2026-08-07)
+
+- 「子代理设置」升级为「**高级设置**」：聚合子代理模型指定、实验功能开关与 WebUI 快捷入口
+- **新增 Kimi Code WebUI 快捷打开**：高级设置页可将新版 Web 界面以独立窗口在应用内打开，也可在系统浏览器打开（需 kimi-code 0.33+，kimi 命令在 PATH 中）
+- **子代理模型指定**：为子代理选择次主力模型后默认绑定该模型，不再继承主模型；实验开关改为滑动开关并修复开启态不可见问题
+- 兼容 kimi-code 0.33+（v2 引擎）：`loop_control` 改用新键名 `max_attempts_per_step`（旧配置自动迁移，保留旧键兼容 legacy 引擎）；Release 附带 `install-macos.sh` 一键安装脚本
+
+## v0.7.1 (2026-08-06)
+
+- **新增「子代理模型（次主力模型）」设置**：选择后子代理默认绑定该模型，不再继承主模型，仅保存模型引用即可继承上下文长度与思考模式
+- **仪表盘用量统计**：识别子代理请求（`__secondary__`）归为「子代理模型（估算）」独立展示，并缓存次主力模型定价，删除配置后历史记录仍稳定估算
+- 自定义网关模型（未收录 models.dev）按官方同名模型跨 provider 匹配价格，不再落入兜底估算
@@ -0,0 +1,14 @@
+# KimiSwitch v0.7.18
+
+## 新增
+
+- **会话管理「一键归档」**：支持按 1 个月前 / 半个月前 / 1 星期前 / 指定日期，一键归档所有最后活跃时间早于所选时间点的会话（一次批量操作，替代逐个选择）；结果反馈已归档 / 跳过数量
+- **归档用量快照落库**：归档瞬间把每个会话的 token 用量与费用按「天 × 模型」粒度快照进 SQLite（`archived_sessions` 表）。之后删除已归档会话（释放磁盘）不再损失历史统计——仪表盘的趋势图、热力图、模型统计、区间合计均会自动合并快照数据，且与现存文件实时统计严格去重（只合并文件已删除的归档会话，绝不双计）
+
+## 修复
+
+- models-dev 快照同步至 2026-09-14（7784 模型 / 213 供应商），`public/` 前端静态目录同步；pricing 测试价格断言跟随 DeepSeek 官方调价（deepseek-v4-flash input 0.14→0.15 / output 0.28→0.60）
+
+## 备注
+
+- 归档仍是元数据语义（文件原地保留），删除文件才会触发快照合并；用量配置、账单查询等其余功能不受影响
@@ -0,0 +1,10 @@
+# KimiSwitch v0.7.19
+
+## 适配
+
+- **适配 kimi-code 0.43.0 实验 flag 镜像（7→5）**：移除已转正的 `auto_session_title` 开关（上游 #3749，AI 会话标题不再需要 flag，首轮对话后自动生成、可在重命名处重新生成）。设置页「实验性 flag」、环境变量探测（`get_experimental_env_status`）、中英文案与测试断言同步清理；上游 0.43.0 注册表现剩 5 个 flag： `wait_for` / `tool-select` / `notify_user` / `tower` / `subagent_fork`
+- **兼容性核查结论（无需改动）**：0.43.0 的 wire 持久化重建（#3737）不影响用量统计——`usage.record` / `token_counting.measured` / `context.apply_compaction` / `turn.prompt` 记录全部保留（`turn.prompt` 仅新增可选 `turnId` 字段）；新增配置项 `loop_control.compaction_max_attempts`（#3750）经现有合并式写入验证保存不丢失；MCP `deferred` 字段（#3667）不涉及本应用管理的配置节
+
+## 备注
+
+- 若此前在 `[experimental]` 中显式开启过 `auto_session_title`，残留键会被上游静默忽略，无需手动清理
@@ -0,0 +1,23 @@
+# KimiSwitch v0.7.20
+
+## 新增
+
+- **套餐用量阈值预警**：用量配置面板新增「预警阈值 %」（0-100，0 或空 = 关闭）。每次账单查询后，任一百分比用量窗口达到阈值即红色高亮 + ⚠ 标记，低于阈值自动消除；自动查询间隔会周期性重复评估
+- **Kimi For Coding 加力钱包**：0.43.1 起服务端 quota 响应中的 `boosterWallet`（加力钱包）现在解析展示为一行余额（💰 月度用量 / 总额 / 余额，含币种），与四个用量窗口并列
+- **供应商凭证环境变量化（`api_key_env`）**：适配 kimi-code 2.0.0 #3762——供应商编辑页「凭证来源」可切换「直接填写 / 环境变量」，密钥不写进 config.toml；导出时与 `api_key` 互斥；账单查询 / 获取模型列表会实时解析环境变量（变量未设置给出明确错误）；SQLite 新增列（含自动迁移）
+- **Kimi Code 版本跟踪器**：设置页新增「Kimi Code 版本」卡片——本机已装版本（`kimi --version` 检测）、上游最新版本、适配状态徽标（绿已适配 / 黄建议升级 KimiSwitch / 橙部分兼容 / 灰未安装），可一键打开上游 Releases
+- **供应商一键体检**：供应商列表新增「一键体检」按钮——并发探活所有启用供应商的推理端点（`GET /v1/models`）并顺带重新查询账单，每行状态点 + 延迟 tooltip + 顶部汇总。状态判定为三态：账单或探活任一真实通过 = 绿；鉴权失败必红；端点不支持模型列表（404）计中性不误报
+
+## 设置
+
+- **压缩重试次数**：Agent 设置新增 `loop_control.compaction_max_attempts`（上游 0.43.0 #3750，默认 5，0/空不写键）
+- **文件监听开关**：新增 `[watch] enabled` 开关（上游 2.0.1 #3892，2.0.2 起上游默认关闭），并探测 `KIMI_CODE_WATCH` 环境变量
+
+## 适配
+
+- **kimi-code 2.0.2 已适配确认**：wire 记录格式、实验 flag（5 个）、config 模式均无变化；`api_key_env` / `[watch]` / `[database]` 新增配置节经现有合并式写入验证无损往返
+
+## 备注
+
+- 版本跟踪器的「最新版本」查询直连 GitHub，受限网络环境下可能显示 `—`（已装版本与状态判定不受影响）
+- 体检跳过大模型平台类供应商（anthropic / google-genai / vertexai 无 `/v1/models` 路由），显示灰点「未检测」
@@ -0,0 +1,23 @@
+# KimiSwitch v0.7.21
+
+## 新增
+
+- **models.dev 在线同步**：高级设置新增「模型参考数据（models.dev）」卡片——一键在线拉取 models.dev 最新快照，新模型发布后无需等待 KimiSwitch 发版：
+  - 覆盖三类参考数据：模型上下文长度、能力标记（思考 / 工具 / 图像 / 视频）、单价（$/M tokens）
+  - 同步后立即生效：供应商编辑页的参数自动填充、仪表盘用量计价（Rust 侧价格索引按快照文件 mtime 自动重建，无需重启应用）
+  - 数据落在 `~/.kimi-switch/models-dev.json`，打包内置快照始终作为兜底；无效副本自动回退
+  - 支持一键「恢复内置数据」，随时可重新同步
+  - 状态行显示当前数据来源（在线同步 / 随版本内置）、快照日期与模型 / 供应商数量
+  - 代理环境自动适配：依次读取 `HTTPS_PROXY` / `HTTP_PROXY` 环境变量与 git `http.proxy` 配置
+
+## 修复
+
+- **构建链路数据漂移**：`fetch-models-dev.mjs` 此前只写 `src/lib/models-dev.json`（Rust 计价内嵌），前端实际加载的 `public/models-dev.json` 需手工拷贝，历史上已两次滞后。现在两个文件由脚本同步写入，彻底消除漂移
+
+## 适配
+
+- **kimi-code 2.1.0 已适配确认**：实验 flag 注册表（5 个）、config 模式、用量 / quota 格式均无变化；上游 `[watch]` 默认关闭与现有开关行为一致；新增 `tui_mode`（`~/.kimi-code/tui.toml`）不涉及配置管理面，暂无适配需求
+
+## 数据
+
+- models-dev 快照同步至 2026-09-23（8,126 模型 / 223 供应商）
@@ -0,0 +1,9 @@
+# KimiSwitch v0.7.22
+
+## 修复
+
+- **Sub2API / NewAPI 模板查询在供应商列表不显示**：为没有预设识别类型的自定义供应商（如自家中转站）配置「Sub2API 中转站」或「NewAPI 中转站」模板后，配置弹窗里「测试查询」正常返回余额，但回到供应商列表却始终空白。根因：列表的查询与显示门槛只认预设注入的 `usageKinds`，而后端模板查询本就绕过 `usageKinds` 直接执行——前端口径与后端不一致。现在二者对齐，显式模板的供应商在列表正常显示余额 / 配额窗口、支持自动查询间隔与预警阈值。
+
+## 备注
+
+- 仅前端修复（`useUsageQuery` 支持判定 + 卡片渲染门槛），后端查询逻辑无变化；已补 3 个单测覆盖支持判定。
@@ -1,7 +1,7 @@
 {
  "name": "kimiswitch",
  "private": true,
-  "version": "0.7.17",
+  "version": "0.7.22",
  "type": "module",
  "scripts": {
    "dev": "vite",
@@ -1,9 +1,17 @@
 #!/usr/bin/env node
 /**
 * Fetch the latest model reference data from models.dev and write:
- *   1. src/lib/models-dev.json      — compact per-model snapshot (frontend)
- *   2. src/lib/models-dev-full.json — provider-grouped full list incl. pricing
- *   3. src/lib/models-dev.last-good.json — local backup of the last success
+ *   1. src/lib/models-dev.json      — compact per-model snapshot; embedded
+ *                                     into the Rust binary by dashboard.rs
+ *                                     (include_str!) and loaded by the
+ *                                     frontend as fallback
+ *   2. public/models-dev.json       — the same snapshot as a static asset;
+ *                                     this is what the frontend actually
+ *                                     serves at runtime (must stay in sync
+ *                                     with #1 — they serve different halves
+ *                                     of the app)
+ *   3. src/lib/models-dev-full.json — provider-grouped full list incl. pricing
+ *   4. src/lib/models-dev.last-good.json — local backup of the last success
 *
 * The data source is https://models.dev/api.json (176 providers; each model
 * carries `cost` = { input, output, cache_read?, cache_write? } in $/M tokens).
@@ -63,6 +71,8 @@ if (proxy && process.env.NODE_USE_ENV_PROXY !== "1") {
 const SOURCE_URL = "https://models.dev/api.json";
 const ROOT = join(dirname(fileURLToPath(import.meta.url)), "..");
 const SNAPSHOT = join(ROOT, "src", "lib", "models-dev.json");
+// Static-asset copy served to the frontend (see header — must equal SNAPSHOT).
+const SNAPSHOT_PUBLIC = join(ROOT, "public", "models-dev.json");
 const FULL = join(ROOT, "src", "lib", "models-dev-full.json");
 // Local backup of the last successful snapshot. Not committed to git (see
 // .gitignore); the committed models-dev.json itself is the versioned fallback.
@@ -168,12 +178,15 @@ async function main() {
  }

  await mkdir(dirname(SNAPSHOT), { recursive: true });
+  await mkdir(dirname(SNAPSHOT_PUBLIC), { recursive: true });
  const json = JSON.stringify(snapshot, null, 2) + "\n";
  await writeFile(SNAPSHOT, json, "utf8");
+  await writeFile(SNAPSHOT_PUBLIC, json, "utf8");
  await writeFile(FULL, JSON.stringify(full, null, 2) + "\n", "utf8");
  await writeFile(BACKUP, json, "utf8");

  console.log(`models.dev snapshot: ${modelCount} models -> ${SNAPSHOT}`);
+  console.log(`models.dev static asset: -> ${SNAPSHOT_PUBLIC}`);
  console.log(
    `models.dev full list: ${Object.keys(full.providers).length} providers -> ${FULL}`,
  );
@@ -3,7 +3,8 @@
 # (git.codingplan.site/admin/KimiCodeSwitch).
 #
 # Usage: publish-gitea.sh <tag> <asset> [asset...]
-#   tag    : e.g. v0.7.15; notes read from release-notes-<tag>.md
+#   tag    : e.g. v0.7.15; notes read from docs/release-notes/<tag>.md
+#            (legacy root release-notes-<tag>.md as fallback)
 #   assets : files to attach (MSI, install-macos.sh, ...)
 #
 # Auth: reuses the stored git credential for git.codingplan.site
@@ -15,7 +16,8 @@ REPO="admin/KimiCodeSwitch"
 TAG="${1:?usage: publish-gitea.sh <tag> <asset> [asset...]}"
 shift

-NOTES_FILE="release-notes-${TAG}.md"
+NOTES_FILE="docs/release-notes/${TAG}.md"
+[ -f "$NOTES_FILE" ] || NOTES_FILE="release-notes-${TAG}.md"
 json_escape() {
  python - "$1" <<'PYEOF'
 import json, sys
@@ -1978,7 +1978,7 @@ dependencies = [

 [[package]]
 name = "kimiswitch"
-version = "0.7.17"
+version = "0.7.22"
 dependencies = [
 "anyhow",
 "chrono",
@@ -1,6 +1,6 @@
 [package]
 name = "kimiswitch"
-version = "0.7.17"
+version = "0.7.22"
 description = "Kimi Switch - model config manager"
 authors = ["codingplan.site"]
 edition = "2021"
@@ -1,11 +1,13 @@
 use chrono::{DateTime, Datelike, Local, TimeZone};
+use crate::db::{ArchivedSessionSnapshot, DayModelStat, DayStat};
 use regex::Regex;
 use serde::{Deserialize, Serialize};
 use std::collections::HashMap;
 use std::fs;
 use std::io::{BufRead, BufReader};
 use std::path::{Path, PathBuf};
-use std::sync::{Mutex, OnceLock};
+use std::sync::{Arc, Mutex, OnceLock, RwLock};
+use std::time::SystemTime;
 use walkdir::WalkDir;

 // ---------------------------------------------------------------------------
@@ -277,6 +279,15 @@ pub struct ActionResponse {
    pub deleted: Option<bool>,
 }

+/// Outcome of the bulk "archive everything older than X" command.
+#[derive(Debug, Serialize, Deserialize, Clone)]
+#[serde(rename_all = "camelCase")]
+pub struct BulkArchiveResult {
+    pub archived: usize,
+    pub skipped: usize,
+    pub errors: Vec<String>,
+}
+
 #[derive(Debug, Serialize, Deserialize, Clone)]
 #[serde(rename_all = "camelCase")]
 pub struct PreviewResult {
@@ -524,50 +535,62 @@ struct ModelsDevCost {
    cache_write: Option<f64>,
 }

-/// Compiled-in models.dev snapshot (`src/lib/models-dev.json`, generated by
-/// scripts/fetch-models-dev.mjs before every build via the `pretauri` hook).
-/// Keyed by "<provider>/<model>", lowercased.
-const MODELS_DEV_SNAPSHOT: &str = include_str!("../../src/lib/models-dev.json");
-
-fn models_dev_cost_index() -> &'static HashMap<String, ModelsDevCost> {
-    static INDEX: OnceLock<HashMap<String, ModelsDevCost>> = OnceLock::new();
-    INDEX.get_or_init(|| {
-        let mut map = HashMap::new();
-        let Ok(v) = serde_json::from_str::<serde_json::Value>(MODELS_DEV_SNAPSHOT) else {
-            return map;
-        };
-        let Some(obj) = v.as_object() else {
-            return map;
-        };
-        for (key, entry) in obj {
-            // Skip the "last_updated" metadata key and entries without cost.
-            if !entry.is_object() || entry.get("cost").is_none() {
-                continue;
+/// models.dev price index over the effective snapshot (runtime-synced copy
+/// when present, else the compiled-in `src/lib/models-dev.json` generated by
+/// scripts/fetch-models-dev.mjs). Keyed by "<provider>/<model>", lowercased.
+/// Cached and rebuilt only when the synced file's mtime changes, so an online
+/// sync (models_dev::sync_models_dev) refreshes prices without a restart.
+fn models_dev_cost_index() -> Arc<HashMap<String, ModelsDevCost>> {
+    static CACHE: OnceLock<RwLock<Option<(Option<SystemTime>, Arc<HashMap<String, ModelsDevCost>>)>>>=
+        OnceLock::new();
+    let cache = CACHE.get_or_init(|| RwLock::new(None));
+    let stamp = fs::metadata(crate::models_dev::synced_path())
+        .and_then(|m| m.modified())
+        .ok();
+    {
+        let cached = cache.read().expect("price index lock");
+        if let Some((cached_stamp, idx)) = cached.as_ref() {
+            if *cached_stamp == stamp {
+                return idx.clone();
            }
-            let Some(cost) = entry.get("cost").and_then(|c| c.as_object()) else {
-                continue;
-            };
-            let num = |k: &str| cost.get(k).and_then(|x| x.as_f64());
-            let (Some(input), Some(output)) = (num("input"), num("output")) else {
-                continue;
-            };
-            map.insert(
-                key.to_ascii_lowercase(),
-                ModelsDevCost {
-                    input,
-                    output,
-                    cache_read: num("cache_read").unwrap_or(0.0),
-                    cache_write: num("cache_write"),
-                },
-            );
        }
-        map
-    })
+    }
+    let mut map = HashMap::new();
+    let v = serde_json::from_str::<serde_json::Value>(&crate::models_dev::effective_snapshot());
+    if let Ok(v) = v {
+        if let Some(obj) = v.as_object() {
+            for (key, entry) in obj {
+                // Skip the "last_updated" metadata key and entries without cost.
+                if !entry.is_object() || entry.get("cost").is_none() {
+                    continue;
+                }
+                let Some(cost) = entry.get("cost").and_then(|c| c.as_object()) else {
+                    continue;
+                };
+                let num = |k: &str| cost.get(k).and_then(|x| x.as_f64());
+                let (Some(input), Some(output)) = (num("input"), num("output")) else {
+                    continue;
+                };
+                map.insert(
+                    key.to_ascii_lowercase(),
+                    ModelsDevCost {
+                        input,
+                        output,
+                        cache_read: num("cache_read").unwrap_or(0.0),
+                        cache_write: num("cache_write"),
+                    },
+                );
+            }
+        }
+    }
+    let idx = Arc::new(map);
+    *cache.write().expect("price index lock") = Some((stamp, idx.clone()));
+    idx
 }

-/// Warm the compiled-in models.dev price index off the first dashboard open:
-/// parsing the ~1.3 MB snapshot costs tens of milliseconds, so build it on a
-/// background thread at startup instead of lazily on the first get_summary.
+/// Warm the models.dev price index off the first dashboard open: parsing the
+/// ~1.3 MB snapshot costs tens of milliseconds, so build it on a background
+/// thread at startup instead of lazily on the first get_summary.
 pub fn warm_price_index() {
    std::thread::spawn(|| {
        let _ = models_dev_cost_index();
@@ -598,14 +621,15 @@ fn provider_rank(key: &str) -> usize {
 ///    (deterministic tie-break). Returns None when nothing matches.
 fn models_dev_lookup(model_name: &str) -> Option<(String, ModelsDevCost)> {
    let lower = model_name.to_ascii_lowercase();
-    if let Some(cost) = models_dev_cost_index().get(&lower) {
+    let index = models_dev_cost_index();
+    if let Some(cost) = index.get(&lower) {
        return Some((lower, *cost));
    }
    let bare = model_name.rsplit_once('/').map(|(_, b)| b).unwrap_or(model_name);
    let bare_l = bare.to_ascii_lowercase();
    // Match on the model part after the last '/', not the full key: aliases
    // strip the family prefix (record "k2.5" vs models.dev "kimi-k2.5").
-    let mut matches: Vec<(&String, &ModelsDevCost)> = models_dev_cost_index()
+    let mut matches: Vec<(&String, &ModelsDevCost)> = index
        .iter()
        .filter(|(key, _)| {
            key.rsplit_once('/')
@@ -685,7 +709,8 @@ fn match_price(model_name: &str) -> (String, f64, f64, f64, bool) {
 fn priced_suffix_lookup(model_name: &str) -> Option<(String, ModelsDevCost)> {
    let bare = model_name.rsplit_once('/').map(|(_, b)| b).unwrap_or(model_name);
    let bare_l = bare.to_ascii_lowercase();
-    let mut matches: Vec<(&String, &ModelsDevCost)> = models_dev_cost_index()
+    let index = models_dev_cost_index();
+    let mut matches: Vec<(&String, &ModelsDevCost)> = index
        .iter()
        .filter(|(key, cost)| {
            (cost.input > 0.0 || cost.output > 0.0)
@@ -853,67 +878,8 @@ fn scan_usage(home: &Path) -> (Vec<UsageRecord>, ScanMeta) {
            Ok(c) => c,
            Err(_) => continue,
        };
-        for line in content.lines() {
-            lines_seen += 1;
-            if !line.contains("\"usage.record\"") { continue; }
-            let obj: serde_json::Value = match serde_json::from_str(line) {
-                Ok(v) => v,
-                Err(_) => continue,
-            };
-            if obj.get("type").and_then(|v| v.as_str()) != Some("usage.record") { continue; }
-            let scope = obj.get("usageScope").and_then(|v| v.as_str()).unwrap_or("turn");
-            if scope != "turn" { continue; }
-
-            let model_raw = obj.get("model").and_then(|v| v.as_str()).unwrap_or("unknown").to_string();
-            // Kimi Code writes the internal marker `__secondary__` as the model
-            // key for usage records emitted by subagents bound to the
-            // configured secondary model. The record stays keyed on that
-            // stable marker so historical records keep their semantics if the
-            // secondary config changes later; cost billing follows the
-            // currently configured secondary model, and the record is flagged
-            // is_secondary so the UI shows it as its own "subagent model"
-            // entry.
-            let is_secondary = model_raw == "__secondary__";
-            let from_env = model_raw == "__kimi_env_model__";
-            let price_model = if is_secondary {
-                secondary_alias.clone().unwrap_or_else(|| model_raw.clone())
-            } else {
-                model_raw.clone()
-            };
-            let time = obj.get("time").and_then(|v| v.as_u64()).unwrap_or(0);
-            let usage = &obj["usage"];
-            let input_other = usage.get("inputOther").and_then(|v| v.as_u64()).unwrap_or(0);
-            let output = usage.get("output").and_then(|v| v.as_u64()).unwrap_or(0);
-            let cache_read = usage.get("inputCacheRead").and_then(|v| v.as_u64()).unwrap_or(0);
-            let cache_create = usage.get("inputCacheCreation").and_then(|v| v.as_u64()).unwrap_or(0);
-
-            // Resolve model display
-            let bare = model_raw.rsplit_once('/').map(|x| x.1).unwrap_or(&model_raw);
-            let (cost, est) = cost_for_usage(input_other, output, cache_read, cache_create, &price_model);
-            // Subagent records are inherently estimates: the `__secondary__`
-            // marker carries no model identity, so the cost is flagged as
-            // estimated even when the current secondary config resolves to a
-            // priced listing.
-            let est = est || is_secondary;
-            let (pid, _ch, _ip, _op, _) = match_price(&price_model);
-
-            records.push(UsageRecord {
-                time,
-                model: model_raw.clone(),
-                model_resolved: bare.to_string(),
-                model_display: model_raw.clone(),
-                provider: resolve_provider(&price_model, &alias2prov),
-                from_env,
-                is_secondary,
-                input_other,
-                output,
-                input_cache_read: cache_read,
-                input_cache_creation: cache_create,
-                cost_usd: cost,
-                cost_estimated: est,
-                price_id: pid,
-            });
-        }
+        lines_seen += content.lines().count();
+        records.extend(parse_wire_usage_records(&content, &secondary_alias, &alias2prov));
    }

    records.sort_by(|a, b| b.time.cmp(&a.time));
@@ -926,6 +892,180 @@ fn scan_usage(home: &Path) -> (Vec<UsageRecord>, ScanMeta) {
    })
 }

+/// Parse one `wire.jsonl` body into usage records. Extracted from
+/// `scan_usage` so a single session can be snapshotted without walking the
+/// whole sessions tree; the record filter chain is unchanged.
+fn parse_wire_usage_records(
+    content: &str,
+    secondary_alias: &Option<String>,
+    alias2prov: &HashMap<String, String>,
+) -> Vec<UsageRecord> {
+    let mut records = Vec::new();
+    for line in content.lines() {
+        if !line.contains("\"usage.record\"") { continue; }
+        let obj: serde_json::Value = match serde_json::from_str(line) {
+            Ok(v) => v,
+            Err(_) => continue,
+        };
+        if obj.get("type").and_then(|v| v.as_str()) != Some("usage.record") { continue; }
+        let scope = obj.get("usageScope").and_then(|v| v.as_str()).unwrap_or("turn");
+        if scope != "turn" { continue; }
+
+        let model_raw = obj.get("model").and_then(|v| v.as_str()).unwrap_or("unknown").to_string();
+        // Kimi Code writes the internal marker `__secondary__` as the model
+        // key for usage records emitted by subagents bound to the
+        // configured secondary model. The record stays keyed on that
+        // stable marker so historical records keep their semantics if the
+        // secondary config changes later; cost billing follows the
+        // currently configured secondary model, and the record is flagged
+        // is_secondary so the UI shows it as its own "subagent model"
+        // entry.
+        let is_secondary = model_raw == "__secondary__";
+        let from_env = model_raw == "__kimi_env_model__";
+        let price_model = if is_secondary {
+            secondary_alias.clone().unwrap_or_else(|| model_raw.clone())
+        } else {
+            model_raw.clone()
+        };
+        let time = obj.get("time").and_then(|v| v.as_u64()).unwrap_or(0);
+        let usage = &obj["usage"];
+        let input_other = usage.get("inputOther").and_then(|v| v.as_u64()).unwrap_or(0);
+        let output = usage.get("output").and_then(|v| v.as_u64()).unwrap_or(0);
+        let cache_read = usage.get("inputCacheRead").and_then(|v| v.as_u64()).unwrap_or(0);
+        let cache_create = usage.get("inputCacheCreation").and_then(|v| v.as_u64()).unwrap_or(0);
+
+        // Resolve model display
+        let bare = model_raw.rsplit_once('/').map(|x| x.1).unwrap_or(&model_raw);
+        let (cost, est) = cost_for_usage(input_other, output, cache_read, cache_create, &price_model);
+        // Subagent records are inherently estimates: the `__secondary__`
+        // marker carries no model identity, so the cost is flagged as
+        // estimated even when the current secondary config resolves to a
+        // priced listing.
+        let est = est || is_secondary;
+        let (pid, _ch, _ip, _op, _) = match_price(&price_model);
+
+        records.push(UsageRecord {
+            time,
+            model: model_raw.clone(),
+            model_resolved: bare.to_string(),
+            model_display: model_raw.clone(),
+            provider: resolve_provider(&price_model, &alias2prov),
+            from_env,
+            is_secondary,
+            input_other,
+            output,
+            input_cache_read: cache_read,
+            input_cache_creation: cache_create,
+            cost_usd: cost,
+            cost_estimated: est,
+            price_id: pid,
+        });
+    }
+    records
+}
+
+/// Aggregate raw usage records into the per-day, per-model snapshot stored
+/// for an archived session.
+fn snapshot_from_records(records: &[UsageRecord]) -> Vec<DayStat> {
+    let mut by_day: HashMap<String, (u64, HashMap<String, DayModelStat>)> = HashMap::new();
+    for r in records {
+        let entry = by_day
+            .entry(day_key(r.time))
+            .or_insert_with(|| (local_midnight_ms(r.time, 0), HashMap::new()));
+        let m = entry.1.entry(r.model.clone()).or_insert_with(|| DayModelStat {
+            model: r.model.clone(),
+            requests: 0,
+            input_other: 0,
+            output: 0,
+            input_cache_read: 0,
+            input_cache_creation: 0,
+            cost_usd: 0.0,
+        });
+        m.requests += 1;
+        m.input_other += r.input_other;
+        m.output += r.output;
+        m.input_cache_read += r.input_cache_read;
+        m.input_cache_creation += r.input_cache_creation;
+        m.cost_usd += r.cost_usd;
+    }
+    let mut days: Vec<DayStat> = by_day
+        .into_iter()
+        .map(|(day, (day_start_ms, models))| {
+            let mut models: Vec<DayModelStat> = models.into_values().collect();
+            models.sort_by(|a, b| a.model.cmp(&b.model));
+            DayStat { day, day_start_ms, models }
+        })
+        .collect();
+    days.sort_by(|a, b| a.day.cmp(&b.day));
+    days
+}
+
+/// Synthesize usage records from stored archive snapshots. Only snapshots
+/// whose session directory is gone are synthesized: sessions still on disk
+/// are covered by the live wire.jsonl scan, and emitting both would double
+/// count their usage.
+fn synthesize_snapshot_records(rows: &[ArchivedSessionSnapshot], home: &Path) -> Vec<UsageRecord> {
+    let root = sessions_root(home);
+    let legacy_root = root.join(".kcd-archive");
+    let alias2prov = build_alias_provider_map(home);
+    let secondary_alias = read_secondary_model_alias(home);
+    let mut out = Vec::new();
+    for row in rows {
+        if root.join(&row.workspace_id).join(&row.session_id).exists() { continue; }
+        if legacy_root.join(&row.workspace_id).join(&row.session_id).exists() { continue; }
+        for day in &row.day_stats {
+            for m in &day.models {
+                let is_secondary = m.model == "__secondary__";
+                let price_model = if is_secondary {
+                    secondary_alias.clone().unwrap_or_else(|| m.model.clone())
+                } else {
+                    m.model.clone()
+                };
+                let bare = m.model.rsplit_once('/').map(|x| x.1).unwrap_or(&m.model).to_string();
+                out.push(UsageRecord {
+                    time: day.day_start_ms,
+                    model: m.model.clone(),
+                    model_resolved: bare,
+                    model_display: m.model.clone(),
+                    provider: resolve_provider(&price_model, &alias2prov),
+                    from_env: m.model == "__kimi_env_model__",
+                    is_secondary,
+                    input_other: m.input_other,
+                    output: m.output,
+                    input_cache_read: m.input_cache_read,
+                    input_cache_creation: m.input_cache_creation,
+                    cost_usd: m.cost_usd,
+                    cost_estimated: true,
+                    price_id: String::new(),
+                });
+            }
+        }
+    }
+    out
+}
+
+/// Append the synthesized records for `rows` to a live scan result and
+/// restore the newest-first ordering `scan_usage` produces.
+fn merge_snapshot_records(
+    mut records: Vec<UsageRecord>,
+    rows: &[ArchivedSessionSnapshot],
+    home: &Path,
+) -> Vec<UsageRecord> {
+    // Nothing archived yet: keep the scan result untouched (and skip the
+    // config reads the synthesis would do).
+    if rows.is_empty() { return records; }
+    records.extend(synthesize_snapshot_records(rows, home));
+    records.sort_by(|a, b| b.time.cmp(&a.time));
+    records
+}
+
+/// Live scan records plus the stored usage of archived sessions whose files
+/// have been deleted, so dashboard stats survive session deletion.
+fn records_with_archive_merge(home: &Path, records: Vec<UsageRecord>) -> Vec<UsageRecord> {
+    let rows = crate::db::list_archived_sessions().unwrap_or_default();
+    merge_snapshot_records(records, &rows, home)
+}
+
 // ---------------------------------------------------------------------------
 // Scan cache
 // ---------------------------------------------------------------------------
@@ -1332,6 +1472,147 @@ fn set_session_archived_meta(session_dir: &Path, archived: bool) -> Result<(), S
    Ok(())
 }

+/// Parse a `state.json` timestamp into epoch ms. kimi-code writes epoch-ms
+/// numbers, other builds write ISO-8601 strings — accept both, plus numeric
+/// strings.
+fn parse_state_time_ms(value: &serde_json::Value) -> Option<u64> {
+    match value {
+        serde_json::Value::Number(n) => n.as_u64().or_else(|| n.as_f64().map(|f| f.max(0.0) as u64)),
+        serde_json::Value::String(s) => parse_time_string_ms(s),
+        _ => None,
+    }
+}
+
+fn parse_time_string_ms(s: &str) -> Option<u64> {
+    let t = s.trim();
+    if t.is_empty() { return None; }
+    if let Ok(n) = t.parse::<u64>() {
+        // Numeric string: epoch ms when it is already millisecond-sized.
+        return Some(if n >= 1e12 as u64 { n } else { n * 1000 });
+    }
+    if let Ok(dt) = DateTime::parse_from_rfc3339(t) {
+        return Some(dt.timestamp_millis().max(0) as u64);
+    }
+    for fmt in [
+        "%Y-%m-%dT%H:%M:%S%.f",
+        "%Y-%m-%dT%H:%M:%S",
+        "%Y-%m-%d %H:%M:%S%.f",
+        "%Y-%m-%d %H:%M:%S",
+    ] {
+        if let Ok(naive) = chrono::NaiveDateTime::parse_from_str(t, fmt) {
+            return Some(naive.and_utc().timestamp_millis().max(0) as u64);
+        }
+    }
+    None
+}
+
+/// Read a session's `state.json` as a JSON object (None when missing or
+/// unparseable).
+fn read_state_obj(session_dir: &Path) -> Option<serde_json::Value> {
+    let content = fs::read_to_string(session_dir.join("state.json")).ok()?;
+    let obj: serde_json::Value = serde_json::from_str(&content).ok()?;
+    if obj.is_object() { Some(obj) } else { None }
+}
+
+/// Session directory mtime in epoch ms — the fallback when `state.json` has
+/// no parseable `updatedAt`.
+fn dir_mtime_ms(dir: &Path) -> Option<u64> {
+    dir.metadata().ok().and_then(|m| m.modified().ok())
+        .and_then(|t| t.duration_since(std::time::UNIX_EPOCH).ok())
+        .map(|d| d.as_millis() as u64)
+}
+
+/// Usage records of one session: every `wire.jsonl` below its directory
+/// (agent subdirectories included), with the same blob/task filtering the
+/// full-tree scan applies.
+fn parse_session_wire_usage(
+    dir: &Path,
+    secondary_alias: &Option<String>,
+    alias2prov: &HashMap<String, String>,
+) -> Vec<UsageRecord> {
+    let mut records = Vec::new();
+    for entry in WalkDir::new(dir).into_iter().filter_map(|e| e.ok()) {
+        if entry.file_name() != "wire.jsonl" || !entry.file_type().is_file() { continue; }
+        let p = entry.path();
+        if p.to_string_lossy().contains("blobs") || p.to_string_lossy().contains("tasks") { continue; }
+        let Ok(content) = fs::read_to_string(p) else { continue };
+        records.extend(parse_wire_usage_records(&content, secondary_alias, alias2prov));
+    }
+    records
+}
+
+/// Bulk-archive every active session whose last activity predates `cutoff_ms`
+/// (epoch ms) and store a usage snapshot per session in SQLite, so dashboard
+/// stats keep counting their history after the directories are deleted.
+/// Sessions already archived, in `.kcd-archive`, or without a readable
+/// `state.json` are left alone.
+fn archive_sessions_before_cmd(home: &Path, cutoff_ms: u64) -> BulkArchiveResult {
+    let root = sessions_root(home);
+    let alias2prov = build_alias_provider_map(home);
+    let secondary_alias = read_secondary_model_alias(home);
+    let now_ms = std::time::SystemTime::now()
+        .duration_since(std::time::UNIX_EPOCH).unwrap_or_default().as_millis() as u64;
+    let mut result = BulkArchiveResult { archived: 0, skipped: 0, errors: Vec::new() };
+
+    for wid in list_workspace_dirs(home) {
+        for entry in safe_read_dir(&root.join(&wid)) {
+            if !entry.file_type().map(|t| t.is_dir()).unwrap_or(false) { continue; }
+            let sid = entry.file_name().to_string_lossy().to_string();
+            if !session_re().is_match(&sid) { continue; }
+            let dir = entry.path();
+            let Some(state) = read_state_obj(&dir) else {
+                result.skipped += 1;
+                continue;
+            };
+            if state.get("archived").and_then(|v| v.as_bool()).unwrap_or(false) { continue; }
+            let Some(updated_ms) = state
+                .get("updatedAt")
+                .and_then(parse_state_time_ms)
+                .or_else(|| dir_mtime_ms(&dir))
+            else {
+                result.skipped += 1;
+                continue;
+            };
+            if updated_ms >= cutoff_ms {
+                result.skipped += 1;
+                continue;
+            }
+
+            if let Err(e) = set_session_archived_meta(&dir, true) {
+                result.errors.push(format!("{wid}/{sid}: {e}"));
+                continue;
+            }
+            let day_stats = snapshot_from_records(&parse_session_wire_usage(
+                &dir, &secondary_alias, &alias2prov,
+            ));
+            let mut total_tokens = 0u64;
+            let mut total_cost_usd = 0.0f64;
+            for day in &day_stats {
+                for m in &day.models {
+                    total_tokens += m.input_other + m.output + m.input_cache_read + m.input_cache_creation;
+                    total_cost_usd += m.cost_usd;
+                }
+            }
+            let snapshot = ArchivedSessionSnapshot {
+                session_id: sid.clone(),
+                workspace_id: wid.clone(),
+                title: state.get("title").and_then(|v| v.as_str()).map(|s| s.to_string()),
+                archived_at_ms: now_ms,
+                updated_at_ms: Some(updated_ms),
+                created_at_ms: state.get("createdAt").and_then(parse_state_time_ms),
+                total_tokens,
+                total_cost_usd,
+                day_stats,
+            };
+            result.archived += 1;
+            if let Err(e) = crate::db::upsert_archived_session(&snapshot) {
+                result.errors.push(format!("{wid}/{sid}: snapshot: {e}"));
+            }
+        }
+    }
+    result
+}
+
 fn humanize_workspace(id: &str) -> String {
    let re = Regex::new(r"^wd_(.+)_[0-9a-fA-F]{8,}$").unwrap();
    if let Some(caps) = re.captures(id) {
@@ -1876,6 +2157,7 @@ pub fn get_summary(home_override: Option<String>, range: Option<String>, refresh
    let r = range.unwrap_or_else(|| "30d".into());

    let (records, meta) = scan_usage_cached(&home, refresh);
+    let records = records_with_archive_merge(&home, records);
    let t1 = std::time::Instant::now();
    let stats = aggregate(&records, &r, now_ms);
    let all_stats = aggregate(&records, "all", now_ms);
@@ -1983,6 +2265,7 @@ fn build_day_detail(date: &str, records: &[UsageRecord]) -> Option<DailyRow> {
 pub fn get_day_detail(home_override: Option<String>, date: String) -> Option<DailyRow> {
    let home = resolve_kimi_home(home_override);
    let (records, _) = scan_usage_cached(&home, false);
+    let records = records_with_archive_merge(&home, records);
    build_day_detail(&date, &records)
 }

@@ -1998,6 +2281,14 @@ pub fn archive_session(home_override: Option<String>, workspace_id: String, sess
    archive_session_cmd(&home, &workspace_id, &session_id)
 }

+/// Bulk-archive every active session whose last activity predates `cutoff_ms`
+/// (epoch ms) across all workspaces, storing a usage snapshot per session.
+#[tauri::command]
+pub fn archive_sessions_before(home_override: Option<String>, cutoff_ms: u64) -> BulkArchiveResult {
+    let home = resolve_kimi_home(home_override);
+    archive_sessions_before_cmd(&home, cutoff_ms)
+}
+
 #[tauri::command]
 pub fn unarchive_session(home_override: Option<String>, workspace_id: String, session_id: String) -> Result<ActionResponse, String> {
    let home = resolve_kimi_home(home_override);
@@ -2073,8 +2364,8 @@ mod pricing_tests {
        let (id, _, input, output, est) = match_price("CodingPlan.site/deepseek-v4-flash");
        assert!(!est, "resolved via models.dev, not estimated");
        assert_eq!(id, "deepseek/deepseek-v4-flash");
-        assert_eq!(input, 0.14);
-        assert_eq!(output, 0.28);
+        assert_eq!(input, 0.15);
+        assert_eq!(output, 0.6);
    }

    #[test]
@@ -2432,3 +2723,271 @@ mod range_and_model_totals_tests {
        assert_eq!(total, 3 * 200);
    }
 }
+
+#[cfg(test)]
+mod archive_snapshot_tests {
+    use super::*;
+
+    const WID: &str = "wd_proj_a1b2c3d4e5f6";
+    const LIVE_SID: &str = "session_11111111-2222-3333-4444-555555555555";
+    const GONE_SID: &str = "session_99999999-8888-7777-6666-555555555555";
+
+    fn now_ms() -> u64 {
+        std::time::SystemTime::now()
+            .duration_since(std::time::UNIX_EPOCH).unwrap().as_millis() as u64
+    }
+
+    fn rec(time: u64, model: &str, tokens: (u64, u64, u64, u64), cost: f64) -> UsageRecord {
+        UsageRecord {
+            time,
+            model: model.to_string(),
+            input_other: tokens.0,
+            output: tokens.1,
+            input_cache_read: tokens.2,
+            input_cache_creation: tokens.3,
+            cost_usd: cost,
+            cost_estimated: false,
+            price_id: String::new(),
+            model_resolved: model.rsplit_once('/').map(|x| x.1).unwrap_or(model).to_string(),
+            model_display: model.to_string(),
+            provider: None,
+            from_env: false,
+            is_secondary: false,
+        }
+    }
+
+    fn snapshot_row(
+        sid: &str,
+        day_start_ms: u64,
+        model: &str,
+        tokens: u64,
+    ) -> ArchivedSessionSnapshot {
+        ArchivedSessionSnapshot {
+            session_id: sid.to_string(),
+            workspace_id: WID.to_string(),
+            title: None,
+            archived_at_ms: 0,
+            updated_at_ms: None,
+            created_at_ms: None,
+            total_tokens: tokens,
+            total_cost_usd: 0.0,
+            day_stats: vec![DayStat {
+                day: day_key(day_start_ms),
+                day_start_ms,
+                models: vec![DayModelStat {
+                    model: model.to_string(),
+                    requests: 2,
+                    input_other: tokens,
+                    output: 0,
+                    input_cache_read: 0,
+                    input_cache_creation: 0,
+                    cost_usd: 1.5,
+                }],
+            }],
+        }
+    }
+
+    fn write_state(dir: &Path, updated_ms: u64, archived: bool) {
+        let state = serde_json::json!({
+            "id": dir.file_name().unwrap().to_string_lossy(),
+            "createdAt": updated_ms,
+            "updatedAt": updated_ms,
+            "archived": archived,
+            "title": "archived candidate",
+        });
+        fs::write(dir.join("state.json"), state.to_string()).unwrap();
+    }
+
+    fn read_state(dir: &Path) -> serde_json::Value {
+        serde_json::from_str(&fs::read_to_string(dir.join("state.json")).unwrap()).unwrap()
+    }
+
+    #[test]
+    fn snapshot_groups_records_by_local_day_and_model() {
+        let now = now_ms();
+        let d1 = local_midnight_ms(now, 2);
+        let d2 = local_midnight_ms(now, 1);
+        let records = vec![
+            rec(d1 + 3_600_000, "openai/gpt-x", (100, 10, 5, 1), 0.5),
+            rec(d1 + 7_200_000, "openai/gpt-x", (200, 20, 0, 0), 0.25),
+            rec(d1 + 60_000, "moonshotai/kimi-k3", (1, 2, 3, 4), 0.01),
+            rec(d2 + 1_000, "openai/gpt-x", (7, 7, 7, 7), 0.07),
+        ];
+        let days = snapshot_from_records(&records);
+        assert_eq!(days.len(), 2, "one bucket per local day");
+
+        let first = &days[0];
+        assert_eq!(first.day, day_key(d1));
+        assert_eq!(first.day_start_ms, d1, "day bucket starts at local midnight");
+        assert_eq!(first.models.len(), 2);
+        let gpt = first.models.iter().find(|m| m.model == "openai/gpt-x").unwrap();
+        assert_eq!(gpt.requests, 2);
+        assert_eq!(gpt.input_other, 300);
+        assert_eq!(gpt.output, 30);
+        assert_eq!(gpt.input_cache_read, 5);
+        assert_eq!(gpt.input_cache_creation, 1);
+        assert_eq!(gpt.cost_usd, 0.75);
+        // Deterministic ordering: models sorted by raw model key.
+        assert_eq!(first.models[0].model, "moonshotai/kimi-k3");
+
+        assert_eq!(days[1].day, day_key(d2));
+        assert_eq!(days[1].day_start_ms, d2);
+        assert_eq!(days[1].models.len(), 1);
+        assert_eq!(days[1].models[0].input_other, 7);
+    }
+
+    #[test]
+    fn synthesize_only_covers_sessions_whose_files_are_gone() {
+        let td = tempfile::tempdir().unwrap();
+        let home = td.path().join("home");
+        fs::create_dir_all(home.join("sessions").join(WID).join(LIVE_SID)).unwrap();
+
+        let day_start = 1_700_000_000_000u64;
+        let rows = vec![
+            snapshot_row(LIVE_SID, day_start, "openai/gpt-x", 100),
+            snapshot_row(GONE_SID, day_start, "openai/gpt-x", 200),
+        ];
+        let out = synthesize_snapshot_records(&rows, &home);
+        assert_eq!(out.len(), 1, "live session dirs stay with the live scan");
+
+        let r = &out[0];
+        assert_eq!(r.time, day_start);
+        assert_eq!(r.model, "openai/gpt-x");
+        assert_eq!(r.model_resolved, "gpt-x");
+        assert_eq!(r.model_display, "openai/gpt-x");
+        assert_eq!(r.provider.as_deref(), Some("openai"));
+        assert_eq!(r.input_other, 200);
+        assert_eq!(r.cost_usd, 1.5);
+        assert!(r.cost_estimated, "snapshot costs are estimates");
+        assert_eq!(r.price_id, "");
+        assert!(!r.is_secondary);
+        assert!(!r.from_env);
+    }
+
+    #[test]
+    fn merge_appends_missing_sessions_newest_first_without_double_counting() {
+        let td = tempfile::tempdir().unwrap();
+        let home = td.path().join("home");
+        fs::create_dir_all(home.join("sessions").join(WID).join(LIVE_SID)).unwrap();
+
+        let now = now_ms();
+        let old_day = local_midnight_ms(now, 5);
+        let new_day = local_midnight_ms(now, 1);
+        let gone_old = "session_22222222-2222-3333-4444-555555555555";
+        let gone_new = "session_33333333-2222-3333-4444-555555555555";
+        let rows = vec![
+            snapshot_row(LIVE_SID, new_day, "openai/gpt-x", 999),
+            snapshot_row(gone_old, old_day, "openai/gpt-x", 10),
+            snapshot_row(gone_new, new_day, "openai/gpt-x", 20),
+        ];
+        let live = vec![rec(new_day + 5_000, "openai/gpt-x", (1, 1, 1, 1), 0.01)];
+        let merged = merge_snapshot_records(live, &rows, &home);
+
+        assert_eq!(merged.len(), 3, "live record plus the two missing sessions");
+        assert_eq!(merged[0].time, new_day + 5_000, "newest record first");
+        assert_eq!(merged[1].time, new_day);
+        assert_eq!(merged[2].time, old_day);
+        assert_eq!(merged[1].input_other, 20);
+        assert_eq!(merged[2].input_other, 10);
+        assert!(
+            merged.iter().all(|r| r.input_other != 999),
+            "a live session's stored snapshot must never be synthesized"
+        );
+    }
+
+    #[test]
+    fn state_times_parse_epoch_numbers_and_iso_strings() {
+        assert_eq!(
+            parse_state_time_ms(&serde_json::json!(1_700_000_000_000u64)),
+            Some(1_700_000_000_000)
+        );
+        assert_eq!(
+            parse_state_time_ms(&serde_json::json!("1700000000000")),
+            Some(1_700_000_000_000)
+        );
+        assert_eq!(
+            parse_state_time_ms(&serde_json::json!("2025-01-02T03:04:05.000Z")),
+            Some(1_735_787_045_000)
+        );
+        assert_eq!(
+            parse_state_time_ms(&serde_json::json!("2025-01-02T03:04:05")),
+            Some(1_735_787_045_000)
+        );
+        assert_eq!(
+            parse_state_time_ms(&serde_json::json!("2025-01-02 03:04:05")),
+            Some(1_735_787_045_000)
+        );
+        assert_eq!(parse_state_time_ms(&serde_json::json!(null)), None);
+        assert_eq!(parse_state_time_ms(&serde_json::json!("not-a-date")), None);
+    }
+
+    #[test]
+    fn bulk_archive_flags_old_sessions_and_snapshots_their_usage() {
+        let td = tempfile::tempdir().unwrap();
+        let home = td.path().join("home");
+        // Redirect the SQLite store so the test never touches the user's DB.
+        std::env::set_var("KIMI_SWITCH_DB_PATH", td.path().join("test.db"));
+
+        let ws = home.join("sessions").join(WID);
+        let now = now_ms();
+        let old_ms = now - 60 * 24 * 3600 * 1000;
+
+        let old_dir = ws.join(LIVE_SID);
+        fs::create_dir_all(&old_dir).unwrap();
+        write_state(&old_dir, old_ms, false);
+        let wire = serde_json::json!({
+            "type": "usage.record",
+            "usageScope": "turn",
+            "time": old_ms,
+            "model": "openai/gpt-x",
+            "usage": {
+                "inputOther": 100,
+                "output": 50,
+                "inputCacheRead": 10,
+                "inputCacheCreation": 5,
+            },
+        });
+        fs::write(old_dir.join("wire.jsonl"), format!("{wire}\n")).unwrap();
+
+        let new_sid = "session_44444444-2222-3333-4444-555555555555";
+        let new_dir = ws.join(new_sid);
+        fs::create_dir_all(&new_dir).unwrap();
+        write_state(&new_dir, now - 3_600_000, false);
+
+        let cutoff = now - 30 * 24 * 3600 * 1000;
+        let res = archive_sessions_before_cmd(&home, cutoff);
+        assert_eq!(res.archived, 1, "only the stale session is archived");
+        assert_eq!(res.skipped, 1, "the recent session is reported as skipped");
+        assert!(res.errors.is_empty(), "{:?}", res.errors);
+
+        // Metadata archive happened in place, the recent session is untouched.
+        assert_eq!(read_state(&old_dir)["archived"], true);
+        assert!(read_state(&old_dir)["archivedAt"].as_u64().unwrap() > 0);
+        assert_eq!(read_state(&new_dir)["archived"], false);
+
+        // Usage snapshot landed in SQLite with the session's totals.
+        let rows = crate::db::list_archived_sessions().unwrap();
+        let row = rows.iter().find(|r| r.session_id == LIVE_SID).expect("snapshot row");
+        assert_eq!(row.workspace_id, WID);
+        assert_eq!(row.title.as_deref(), Some("archived candidate"));
+        assert_eq!(row.updated_at_ms, Some(old_ms));
+        assert_eq!(row.total_tokens, 165);
+        assert!(row.total_cost_usd > 0.0);
+        assert_eq!(row.day_stats.len(), 1);
+        assert_eq!(row.day_stats[0].day, day_key(old_ms));
+        assert_eq!(row.day_stats[0].models[0].requests, 1);
+        assert_eq!(row.day_stats[0].models[0].input_other, 100);
+
+        // While the files are still there, the live scan owns the usage.
+        let merged = records_with_archive_merge(&home, vec![]);
+        assert!(merged.is_empty(), "no synthesized records while the dir exists");
+
+        // Once the archived directory is deleted, stats come from the snapshot.
+        fs::remove_dir_all(&old_dir).unwrap();
+        let merged = records_with_archive_merge(&home, vec![]);
+        assert_eq!(merged.len(), 1);
+        assert_eq!(merged[0].input_other, 100);
+        assert_eq!(merged[0].output, 50);
+        assert_eq!(day_key(merged[0].time), day_key(old_ms));
+    }
+}
@@ -10,6 +10,7 @@ use std::path::PathBuf;
 use anyhow::Context;
 use indexmap::IndexMap;
 use rusqlite::{params, Connection};
+use serde::{Deserialize, Serialize};
 use serde_json::Value;

 use crate::models::{Agent, Config, Model, Provider, ProviderType};
@@ -23,6 +24,13 @@ pub fn kimi_switch_data_dir() -> PathBuf {
 }

 pub fn db_path() -> PathBuf {
+    // Tests (and power users) can point the store elsewhere; production always
+    // resolves to the default location under the home directory.
+    if let Some(p) = std::env::var_os("KIMI_SWITCH_DB_PATH") {
+        if !p.is_empty() {
+            return PathBuf::from(p);
+        }
+    }
    kimi_switch_data_dir().join("kimi-switch.db")
 }

@@ -31,6 +39,10 @@ pub fn db_path() -> PathBuf {
 /// configuration. Safe to call on every startup — it is a no-op once the new
 /// directory exists.
 fn migrate_legacy_data_dir() {
+    // A redirected store (tests) must never touch the real home directory.
+    if std::env::var_os("KIMI_SWITCH_DB_PATH").is_some() {
+        return;
+    }
    let new_dir = kimi_switch_data_dir();
    if new_dir.exists() {
        return;
@@ -69,6 +81,7 @@ pub fn init_db() -> DbResult<Connection> {
            provider_type TEXT NOT NULL,
            base_url TEXT,
            api_key TEXT,
+            api_key_env TEXT,
            env TEXT,
            note TEXT,
            official_url TEXT,
@@ -106,11 +119,32 @@ pub fn init_db() -> DbResult<Connection> {
        [],
    )?;

+    // Usage snapshots taken when sessions are bulk-archived: keep the session's
+    // token/cost history queryable after its directory (and wire.jsonl) is
+    // deleted from disk.
+    conn.execute(
+        "CREATE TABLE IF NOT EXISTS archived_sessions (
+            session_id     TEXT PRIMARY KEY,
+            workspace_id   TEXT NOT NULL,
+            title          TEXT,
+            archived_at_ms INTEGER NOT NULL,
+            updated_at_ms  INTEGER,
+            created_at_ms  INTEGER,
+            total_tokens   INTEGER NOT NULL DEFAULT 0,
+            total_cost_usd REAL NOT NULL DEFAULT 0,
+            day_stats      TEXT NOT NULL DEFAULT '[]'
+        )",
+        [],
+    )?;
+
    // Add icon columns for provider icon picker (introduced in v0.3.1).
    // SQLite has no ADD COLUMN IF NOT EXISTS, so ignore duplicate-column errors.
    for stmt in [
        "ALTER TABLE providers ADD COLUMN icon TEXT",
        "ALTER TABLE providers ADD COLUMN icon_color TEXT",
+        // `api_key_env` (kimi-code 2.0.0+): the name of the environment
+        // variable holding the credential, an alternative to `api_key`.
+        "ALTER TABLE providers ADD COLUMN api_key_env TEXT",
    ] {
        if let Err(e) = conn.execute(stmt, []) {
            let msg = e.to_string();
@@ -132,28 +166,29 @@ pub fn load_config(agent: &Agent) -> DbResult<Config> {
    let mut providers = IndexMap::new();
    {
        let mut stmt = tx.prepare(
-            "SELECT name, provider_type, base_url, api_key, env, note, official_url, managed, enabled, active, icon, icon_color, raw_other
+            "SELECT name, provider_type, base_url, api_key, api_key_env, env, note, official_url, managed, enabled, active, icon, icon_color, raw_other
             FROM providers WHERE agent = ?1 ORDER BY id",
        )?;
        let provider_rows = stmt.query_map(params![agent.as_str()], |row| {
            let provider_type: String = row.get(1)?;
-            let env_json: Option<String> = row.get(4)?;
-            let raw_json: Option<String> = row.get(12)?;
+            let env_json: Option<String> = row.get(5)?;
+            let raw_json: Option<String> = row.get(13)?;
            Ok(Provider {
                name: row.get(0)?,
                provider_type: provider_type_for_str(&provider_type),
                base_url: row.get(2)?,
                api_key: row.get(3)?,
+                api_key_env: row.get(4)?,
                env: env_json
                    .and_then(|s| serde_json::from_str(&s).ok())
                    .unwrap_or_default(),
-                note: row.get(5)?,
-                official_url: row.get(6)?,
-                managed: row.get::<_, i32>(7)? != 0,
-                enabled: row.get::<_, i32>(8)? != 0,
-                active: row.get::<_, i32>(9)? != 0,
-                icon: row.get(10)?,
-                icon_color: row.get(11)?,
+                note: row.get(6)?,
+                official_url: row.get(7)?,
+                managed: row.get::<_, i32>(8)? != 0,
+                enabled: row.get::<_, i32>(9)? != 0,
+                active: row.get::<_, i32>(10)? != 0,
+                icon: row.get(11)?,
+                icon_color: row.get(12)?,
                raw_other: raw_json
                    .and_then(|s| serde_json::from_str(&s).ok())
                    .unwrap_or(Value::Null),
@@ -223,8 +258,8 @@ pub fn save_config(agent: &Agent, config: &Config) -> DbResult<()> {
    {
        let mut insert_provider = tx.prepare(
            "INSERT INTO providers
-             (agent, name, provider_type, base_url, api_key, env, note, official_url, managed, enabled, active, icon, icon_color, raw_other)
-             VALUES (?1, ?2, ?3, ?4, ?5, ?6, ?7, ?8, ?9, ?10, ?11, ?12, ?13, ?14)",
+             (agent, name, provider_type, base_url, api_key, api_key_env, env, note, official_url, managed, enabled, active, icon, icon_color, raw_other)
+             VALUES (?1, ?2, ?3, ?4, ?5, ?6, ?7, ?8, ?9, ?10, ?11, ?12, ?13, ?14, ?15)",
        )?;

        for provider in config.providers.values() {
@@ -234,6 +269,7 @@ pub fn save_config(agent: &Agent, config: &Config) -> DbResult<()> {
                provider.provider_type.as_str(),
                provider.base_url,
                provider.api_key,
+                provider.api_key_env,
                serde_json::to_string(&provider.env).ok(),
                provider.note,
                provider.official_url,
@@ -330,6 +366,104 @@ pub fn delete_setting_pub(key: &str) -> DbResult<()> {
    Ok(())
 }

+// ---------------------------------------------------------------------------
+// Archived-session usage snapshots
+// ---------------------------------------------------------------------------
+
+/// Per-model usage totals for one calendar day of an archived session.
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(rename_all = "camelCase")]
+pub struct DayModelStat {
+    /// Raw model key as written to wire.jsonl (`__secondary__` kept verbatim).
+    pub model: String,
+    pub requests: u64,
+    pub input_other: u64,
+    pub output: u64,
+    pub input_cache_read: u64,
+    pub input_cache_creation: u64,
+    pub cost_usd: f64,
+}
+
+/// One day's per-model usage, stored as JSON in `archived_sessions.day_stats`.
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(rename_all = "camelCase")]
+pub struct DayStat {
+    /// Local calendar day, `YYYY-MM-DD`.
+    pub day: String,
+    /// Local midnight of `day`, epoch ms — used as the synthesized record time.
+    pub day_start_ms: u64,
+    pub models: Vec<DayModelStat>,
+}
+
+/// Usage snapshot of one session, taken at the moment it is archived.
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(rename_all = "camelCase")]
+pub struct ArchivedSessionSnapshot {
+    pub session_id: String,
+    pub workspace_id: String,
+    pub title: Option<String>,
+    pub archived_at_ms: u64,
+    pub updated_at_ms: Option<u64>,
+    pub created_at_ms: Option<u64>,
+    pub total_tokens: u64,
+    pub total_cost_usd: f64,
+    pub day_stats: Vec<DayStat>,
+}
+
+/// Insert or replace one archived-session snapshot (keyed by session id, which
+/// is a UUID and never reused).
+pub fn upsert_archived_session(row: &ArchivedSessionSnapshot) -> DbResult<()> {
+    let conn = init_db()?;
+    let day_stats = serde_json::to_string(&row.day_stats)?;
+    conn.execute(
+        "INSERT OR REPLACE INTO archived_sessions
+         (session_id, workspace_id, title, archived_at_ms, updated_at_ms, created_at_ms, total_tokens, total_cost_usd, day_stats)
+         VALUES (?1, ?2, ?3, ?4, ?5, ?6, ?7, ?8, ?9)",
+        params![
+            row.session_id,
+            row.workspace_id,
+            row.title,
+            row.archived_at_ms as i64,
+            row.updated_at_ms.map(|v| v as i64),
+            row.created_at_ms.map(|v| v as i64),
+            row.total_tokens as i64,
+            row.total_cost_usd,
+            day_stats,
+        ],
+    )?;
+    Ok(())
+}
+
+/// Every stored archived-session snapshot. Unreadable `day_stats` JSON degrades
+/// to an empty breakdown instead of failing the whole list.
+pub fn list_archived_sessions() -> DbResult<Vec<ArchivedSessionSnapshot>> {
+    let conn = init_db()?;
+    let mut stmt = conn.prepare(
+        "SELECT session_id, workspace_id, title, archived_at_ms, updated_at_ms, created_at_ms,
+                total_tokens, total_cost_usd, day_stats
+         FROM archived_sessions",
+    )?;
+    let rows = stmt.query_map([], |row| {
+        let day_stats_json: String = row.get(8)?;
+        Ok(ArchivedSessionSnapshot {
+            session_id: row.get(0)?,
+            workspace_id: row.get(1)?,
+            title: row.get(2)?,
+            archived_at_ms: row.get::<_, i64>(3)? as u64,
+            updated_at_ms: row.get::<_, Option<i64>>(4)?.map(|v| v as u64),
+            created_at_ms: row.get::<_, Option<i64>>(5)?.map(|v| v as u64),
+            total_tokens: row.get::<_, i64>(6)? as u64,
+            total_cost_usd: row.get(7)?,
+            day_stats: serde_json::from_str(&day_stats_json).unwrap_or_default(),
+        })
+    })?;
+    let mut out = Vec::new();
+    for row in rows {
+        out.push(row?);
+    }
+    Ok(out)
+}
+
 fn provider_type_for_str(s: &str) -> ProviderType {
    ProviderType::from_kimi_type(s)
 }
@@ -153,6 +153,14 @@ pub fn kimi_code_to_config(value: &TomlValue) -> Config {

            let base_url = table.get("base_url").and_then(|v| v.as_str()).map(|s| s.to_string());
            let api_key = table.get("api_key").and_then(|v| v.as_str()).map(|s| s.to_string());
+            // kimi-code 2.0.0+: `api_key_env` names the environment variable
+            // holding the credential; it is mutually exclusive with `api_key`.
+            // Captured as its own field (not left in raw_other) so export can
+            // enforce the exclusivity and the UI can edit it.
+            let api_key_env = table
+                .get("api_key_env")
+                .and_then(|v| v.as_str())
+                .map(|s| s.to_string());
            let managed = table.contains_key("oauth")
                || table.get("managed").and_then(|v| v.as_bool()).unwrap_or(false);
            let enabled = table.get("enabled").and_then(|v| v.as_bool()).unwrap_or(true);
@@ -181,6 +189,7 @@ pub fn kimi_code_to_config(value: &TomlValue) -> Config {
                rest.remove("type");
                rest.remove("base_url");
                rest.remove("api_key");
+                rest.remove("api_key_env");
                rest.remove("managed");
                rest.remove("enabled");
                rest.remove("env");
@@ -200,6 +209,7 @@ pub fn kimi_code_to_config(value: &TomlValue) -> Config {
                    provider_type,
                    base_url: base_url.filter(|s| !s.is_empty()),
                    api_key: api_key.filter(|s| !s.is_empty()),
+                    api_key_env: api_key_env.filter(|s| !s.is_empty()),
                    env,
                    note: None,
                    official_url: None,
@@ -289,11 +299,30 @@ pub fn config_to_kimi_code(config: &Config, existing: Option<&TomlValue>) -> Tom
        if let Some(base_url) = provider.base_url.clone().filter(|s| !s.is_empty()) {
            pt.insert("base_url".to_string(), TomlValue::String(base_url));
        }
+        // `api_key_env` (kimi-code 2.0.0+) and `api_key` are mutually
+        // exclusive upstream: emitting both makes the CLI reject the provider
+        // block, so the env var name alone is written whenever the provider is
+        // configured that way. A legacy row that still carries the key in
+        // raw_other (imports before the field existed) is used as the source
+        // too, so the setting survives the upgrade.
+        let api_key_env = provider
+            .api_key_env
+            .clone()
+            .or_else(|| {
+                provider
+                    .raw_other
+                    .get("api_key_env")
+                    .and_then(|v| v.as_str())
+                    .map(|s| s.to_string())
+            })
+            .filter(|s| !s.is_empty());
        // Managed (OAuth) providers persist an empty api_key line so the block
        // matches the official CLI's provisioned shape (apiKey: ''). A
        // user-set api_key on a managed provider (api_key outranks OAuth in
        // the CLI's credential priority) must survive the round-trip.
-        if provider.managed
+        if let Some(env_key) = api_key_env {
+            pt.insert("api_key_env".to_string(), TomlValue::String(env_key));
+        } else if provider.managed
            && provider.api_key.as_deref().map_or(true, |s| s.is_empty())
        {
            pt.insert("api_key".to_string(), TomlValue::String("".to_string()));
@@ -328,10 +357,12 @@ pub fn config_to_kimi_code(config: &Config, existing: Option<&TomlValue>) -> Tom
                });
            pt.insert("oauth".to_string(), oauth);
        }
-        // Merge remaining raw fields (oauth and env are handled explicitly above).
+        // Merge remaining raw fields (oauth, env and api_key_env are handled
+        // explicitly above).
        if let TomlValue::Table(mut extra) = json_to_toml(&provider.raw_other).unwrap_or(TomlValue::Table(Table::new())) {
            extra.remove("oauth");
            extra.remove("env");
+            extra.remove("api_key_env");
            // Strip Kimi-Switch-private field: the remembered per-provider
            // default model is stored in raw_other.default_model and must NOT
            // leak into the agent's config.toml.
@@ -562,6 +593,7 @@ api_key = ""
                provider_type: ProviderType::Anthropic,
                base_url: Some("https://fast.cdks.work".to_string()),
                api_key: Some("sk-test".to_string()),
+                api_key_env: None,
                env: IndexMap::new(),
                note: None,
                official_url: None,
@@ -623,6 +655,7 @@ api_key = ""
            provider_type: ProviderType::Openai,
            base_url: Some(base_url.to_string()),
            api_key: Some("sk-test".to_string()),
+            api_key_env: None,
            env: IndexMap::new(),
            note: None,
            official_url: None,
@@ -760,6 +793,7 @@ default_effort = "low"
                provider_type: ProviderType::Anthropic,
                base_url: Some("https://example.com".to_string()),
                api_key: Some("sk-test".to_string()),
+                api_key_env: None,
                env: IndexMap::new(),
                note: None,
                official_url: None,
@@ -868,6 +902,7 @@ max_context_size = 1048576
                provider_type: ProviderType::Anthropic,
                base_url: Some("https://a.example.com".to_string()),
                api_key: Some("sk-a".to_string()),
+                api_key_env: None,
                env: IndexMap::new(),
                note: None,
                official_url: None,
@@ -888,6 +923,7 @@ max_context_size = 1048576
                provider_type: ProviderType::Openai,
                base_url: Some("https://b.example.com".to_string()),
                api_key: Some("sk-b".to_string()),
+                api_key_env: None,
                env: IndexMap::new(),
                note: None,
                official_url: None,
@@ -929,6 +965,7 @@ max_context_size = 1048576
                provider_type: ProviderType::Anthropic,
                base_url: None,
                api_key: Some("sk-x".to_string()),
+                api_key_env: None,
                env: IndexMap::new(),
                note: None,
                official_url: None,
@@ -974,6 +1011,7 @@ max_context_size = 1048576
            provider_type: ProviderType::Kimi,
            base_url: Some("https://api.kimi.com/coding/v1".to_string()),
            api_key: api_key.map(String::from),
+            api_key_env: None,
            env: IndexMap::new(),
            note: None,
            official_url: None,
@@ -1153,4 +1191,112 @@ enabled = true
            "user-removed section must be dropped on export"
        );
    }
+
+    #[test]
+    fn kimi_code_import_extracts_api_key_env() {
+        // kimi-code 2.0.0+: `api_key_env` names the environment variable that
+        // holds the credential. It must become its own Provider field and stop
+        // living in raw_other, so export can enforce the api_key exclusivity.
+        let toml_str = r#"
+default_model = "p1/m1"
+
+[providers.p1]
+type = "openai"
+base_url = "https://proxy.example.com/v1"
+api_key_env = "MY_PROXY_KEY"
+
+[models."p1/m1"]
+provider = "p1"
+model = "m1"
+max_context_size = 128000
+"#;
+        let value: TomlValue = toml_str.parse().unwrap();
+        let config = kimi_code_to_config(&value);
+
+        let provider = config.providers.get("p1").unwrap();
+        assert_eq!(provider.api_key_env.as_deref(), Some("MY_PROXY_KEY"));
+        assert_eq!(provider.api_key, None);
+        assert!(
+            provider.raw_other.get("api_key_env").is_none(),
+            "api_key_env must not stay in raw_other"
+        );
+
+        let exported = config_to_kimi_code(&config, Some(&value));
+        let exported_provider = exported
+            .as_table().unwrap()
+            .get("providers").unwrap()
+            .as_table().unwrap()
+            .get("p1").unwrap()
+            .as_table().unwrap();
+        assert_eq!(
+            exported_provider.get("api_key_env").and_then(|v| v.as_str()),
+            Some("MY_PROXY_KEY")
+        );
+        assert!(
+            !exported_provider.contains_key("api_key"),
+            "api_key and api_key_env are mutually exclusive upstream"
+        );
+    }
+
+    #[test]
+    fn kimi_code_export_api_key_env_wins_over_stored_api_key() {
+        // A provider switched to env-var credentials in the UI must export
+        // ONLY `api_key_env`: writing both keys makes the CLI reject the block.
+        // A legacy raw_other copy of the key acts as the source when the field
+        // is empty (rows saved before the field existed).
+        let make_provider = |api_key: Option<&str>, api_key_env: Option<&str>, raw: Value| Provider {
+            name: "p".to_string(),
+            provider_type: ProviderType::Openai,
+            base_url: Some("https://proxy.example.com/v1".to_string()),
+            api_key: api_key.map(String::from),
+            api_key_env: api_key_env.map(String::from),
+            env: IndexMap::new(),
+            note: None,
+            official_url: None,
+            managed: false,
+            enabled: true,
+            active: false,
+            icon: None,
+            icon_color: None,
+            raw_other: raw,
+            usage_kinds: None,
+            usage_config: None,
+        };
+        let export = |p: Provider| {
+            let config = Config {
+                default_model: None,
+                providers: IndexMap::from([("p".to_string(), p)]),
+                models: IndexMap::new(),
+                raw_other: Value::Null,
+                imported_section_keys: Vec::new(),
+            };
+            config_to_kimi_code(&config, None)
+                .as_table().unwrap()
+                .get("providers").unwrap()
+                .as_table().unwrap()
+                .get("p").unwrap()
+                .as_table()
+                .unwrap()
+                .clone()
+        };
+
+        // Both fields set (a DB row saved before the UI cleared the key).
+        let pt = export(make_provider(Some("sk-stale"), Some("MY_PROXY_KEY"), Value::Null));
+        assert_eq!(pt.get("api_key_env").and_then(|v| v.as_str()), Some("MY_PROXY_KEY"));
+        assert!(!pt.contains_key("api_key"), "the stored key must be dropped");
+
+        // Legacy row: only raw_other carries the key.
+        let pt = export(make_provider(
+            Some("sk-stale"),
+            None,
+            serde_json::json!({"api_key_env": "LEGACY_KEY"}),
+        ));
+        assert_eq!(pt.get("api_key_env").and_then(|v| v.as_str()), Some("LEGACY_KEY"));
+        assert!(!pt.contains_key("api_key"));
+
+        // No env reference → the stored key is written as before.
+        let pt = export(make_provider(Some("sk-keep"), None, Value::Null));
+        assert_eq!(pt.get("api_key").and_then(|v| v.as_str()), Some("sk-keep"));
+        assert!(!pt.contains_key("api_key_env"));
+    }
 }
@@ -4,6 +4,7 @@ pub mod dashboard;
 pub mod db;
 pub mod kimi_code_io;
 pub mod models;
+pub mod models_dev;
 pub mod oauth;
 pub mod pi_io;
 pub mod plugins;
@@ -103,10 +104,14 @@ pub fn run() {
            commands::list_provider_models,
            commands::test_connectivity,
            commands::query_provider_usage,
+            commands::health_check_all,
            commands::debug_log,
            commands::get_app_setting,
            commands::set_app_setting,
            commands::check_for_update,
+            commands::get_kimi_code_version,
+            commands::get_kimi_code_latest,
+            commands::get_kimi_code_version_info,
            commands::download_update,
            commands::open_installer,
            commands::open_external_url,
@@ -115,12 +120,17 @@ pub fn run() {
            commands::kimi_oauth_start,
            commands::kimi_oauth_poll,
            commands::get_experimental_env_status,
+            models_dev::sync_models_dev,
+            models_dev::get_models_dev_status,
+            models_dev::get_models_dev_snapshot,
+            models_dev::reset_models_dev,
            dashboard::get_paths,
            dashboard::get_prices,
            dashboard::get_summary,
            dashboard::get_day_detail,
            dashboard::list_sessions,
            dashboard::archive_session,
+            dashboard::archive_sessions_before,
            dashboard::unarchive_session,
            dashboard::delete_session,
            dashboard::delete_workspace,
@@ -107,6 +107,12 @@ pub struct Provider {
    pub provider_type: ProviderType,
    pub base_url: Option<String>,
    pub api_key: Option<String>,
+    /// kimi-code 2.0.0+: name of the process environment variable holding the
+    /// credential, as an alternative to a stored `api_key` (the two are
+    /// mutually exclusive upstream — exporting both makes the CLI reject the
+    /// provider block). Only the variable *name* is stored, never its value.
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub api_key_env: Option<String>,
    #[serde(default)]
    pub env: IndexMap<String, String>,
    #[serde(default, skip_serializing_if = "Option::is_none")]
@@ -149,6 +155,7 @@ impl PartialEq for Provider {
            && self.provider_type == other.provider_type
            && self.base_url == other.base_url
            && self.api_key == other.api_key
+            && self.api_key_env == other.api_key_env
            && self.env == other.env
            && self.note == other.note
            && self.official_url == other.official_url
@@ -172,6 +179,7 @@ impl std::fmt::Debug for Provider {
            .field("provider_type", &self.provider_type)
            .field("base_url", &self.base_url)
            .field("api_key", &"<redacted>")
+            .field("api_key_env", &self.api_key_env)
            .field("env", &"<redacted>")
            .field("managed", &self.managed)
            .field("enabled", &self.enabled)
@@ -285,6 +293,11 @@ pub struct UsageConfig {
    /// Per-request timeout in seconds; 0/None = default (8s).
    #[serde(default, skip_serializing_if = "Option::is_none")]
    pub timeout_seconds: Option<u64>,
+    /// 用量预警阈值（百分比，0-100）；None/0 = 关闭。判定在前端
+    /// src/hooks/useUsageQuery.ts，Rust 只负责透传——缺了此字段，serde 会在
+    /// 保存时把前端写下的 threshold 静默丢掉。
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub threshold: Option<u32>,
 }

 impl UsageConfig {
@@ -0,0 +1,352 @@
+//! models.dev reference-data sync.
+//!
+//! The app carries a compiled-in models.dev snapshot (generated by
+//! scripts/fetch-models-dev.mjs before every build). This module lets the
+//! user refresh it at runtime from https://models.dev/api.json without
+//! rebuilding: the synced copy lives at `~/.kimi-switch/models-dev.json` and
+//! takes precedence whenever it exists and parses. The dashboard price index
+//! (see dashboard::models_dev_cost_index) re-checks that file's mtime so new
+//! prices take effect without an app restart.
+
+use std::collections::HashSet;
+use std::fs;
+use std::path::PathBuf;
+use std::time::Duration;
+
+use chrono::Utc;
+use serde::Serialize;
+use serde_json::Value;
+
+use crate::db::kimi_switch_data_dir;
+
+/// Compiled-in snapshot — the same file the dashboard price index used to
+/// embed directly. Always available; the runtime-synced copy (if any) wins.
+const BUILTIN_SNAPSHOT: &str = include_str!("../../src/lib/models-dev.json");
+
+const SOURCE_URL: &str = "https://models.dev/api.json";
+const SYNCED_FILE: &str = "models-dev.json";
+
+/// Where the runtime-synced snapshot lives. The dashboard stats this path's
+/// mtime to decide when to rebuild its price index.
+pub fn synced_path() -> PathBuf {
+    kimi_switch_data_dir().join(SYNCED_FILE)
+}
+
+/// Read + validate the synced copy. Anything unparsable or missing
+/// `last_updated` is treated as absent so the built-in snapshot takes over.
+fn read_synced() -> Option<String> {
+    let raw = fs::read_to_string(synced_path()).ok()?;
+    let v: Value = serde_json::from_str(&raw).ok()?;
+    if v.get("last_updated").and_then(|x| x.as_str()).is_some() {
+        Some(raw)
+    } else {
+        None
+    }
+}
+
+/// The snapshot the app should use right now: the synced copy when present
+/// and valid, otherwise the compiled-in one.
+pub fn effective_snapshot() -> String {
+    read_synced().unwrap_or_else(|| BUILTIN_SNAPSHOT.to_string())
+}
+
+#[derive(Debug, Serialize, Clone)]
+#[serde(rename_all = "camelCase")]
+pub struct ModelsDevStatus {
+    /// "synced" = runtime online copy; "builtin" = compiled at build time.
+    pub source: &'static str,
+    pub last_updated: String,
+    pub model_count: u64,
+    pub provider_count: u64,
+}
+
+/// Count models / providers and read the `last_updated` stamp from a
+/// snapshot string ("<provider>/<model>" keys, plus the `last_updated` meta).
+fn status_of(source: &'static str, raw: &str) -> ModelsDevStatus {
+    let v: Value = serde_json::from_str(raw).unwrap_or(Value::Null);
+    let mut model_count = 0u64;
+    let mut providers = HashSet::new();
+    if let Some(obj) = v.as_object() {
+        for (k, entry) in obj {
+            if k == "last_updated" || !entry.is_object() {
+                continue;
+            }
+            model_count += 1;
+            if let Some((p, _)) = k.split_once('/') {
+                providers.insert(p.to_string());
+            }
+        }
+    }
+    ModelsDevStatus {
+        source,
+        last_updated: v
+            .get("last_updated")
+            .and_then(|x| x.as_str())
+            .unwrap_or("unknown")
+            .to_string(),
+        model_count,
+        provider_count: providers.len() as u64,
+    }
+}
+
+fn status(source: &'static str) -> ModelsDevStatus {
+    match source {
+        "synced" => {
+            let raw = read_synced().unwrap_or_default();
+            status_of("synced", &raw)
+        }
+        _ => status_of("builtin", BUILTIN_SNAPSHOT),
+    }
+}
+
+/// Numbers serialize as integers when they are whole (matches the mjs
+/// script's JSON output, e.g. context 1000000 not 1000000.0).
+fn num(v: f64) -> Value {
+    if v.fract() == 0.0 && v.abs() < 9e15 {
+        Value::from(v as i64)
+    } else {
+        Value::from(v)
+    }
+}
+
+/// Transform models.dev api.json into the compact snapshot shape produced by
+/// scripts/fetch-models-dev.mjs (the canonical field-picking logic lives
+/// there; this is the runtime port). Pure — unit-tested below.
+fn build_snapshot(raw: &Value) -> Result<Value, String> {
+    let providers = raw.as_object().ok_or("api.json is not a JSON object")?;
+    let mut snapshot = serde_json::Map::new();
+    snapshot.insert(
+        "last_updated".into(),
+        Value::String(Utc::now().format("%Y-%m-%d").to_string()),
+    );
+
+    for (provider_id, provider) in providers {
+        let Some(models) = provider.get("models").and_then(|m| m.as_object()) else {
+            continue;
+        };
+        for (model_id, m) in models {
+            let mut entry = serde_json::Map::new();
+            if let Some(name) = m.get("name").and_then(|x| x.as_str()) {
+                entry.insert("name".into(), Value::String(name.to_string()));
+            }
+            // Context 0 means "not applicable" (image/audio models) — treat as
+            // missing so callers fall back to defaults.
+            if let Some(ctx) = m
+                .pointer("/limit/context")
+                .and_then(|x| x.as_f64())
+                .filter(|c| *c > 0.0)
+            {
+                entry.insert("context".into(), num(ctx));
+            }
+            for flag in ["reasoning", "tool_call", "structured_output"] {
+                if m.get(flag).and_then(|x| x.as_bool()) == Some(true) {
+                    entry.insert(flag.into(), Value::Bool(true));
+                }
+            }
+            if let Some(input) = m.pointer("/modalities/input").and_then(|x| x.as_array()) {
+                let has = |s: &str| input.iter().any(|v| v.as_str() == Some(s));
+                if has("image") {
+                    entry.insert("image".into(), Value::Bool(true));
+                }
+                if has("video") {
+                    entry.insert("video".into(), Value::Bool(true));
+                }
+            }
+            if let Some(cost) = m.get("cost").and_then(|c| c.as_object()) {
+                let mut c = serde_json::Map::new();
+                for k in ["input", "output", "cache_read", "cache_write"] {
+                    if let Some(n) = cost.get(k).and_then(|x| x.as_f64()) {
+                        c.insert(k.into(), num(n));
+                    }
+                }
+                if !c.is_empty() {
+                    entry.insert("cost".into(), Value::Object(c));
+                }
+            }
+            snapshot.insert(format!("{provider_id}/{model_id}"), Value::Object(entry));
+        }
+    }
+    Ok(Value::Object(snapshot))
+}
+
+/// reqwest is built without the default system-proxy feature, so proxies must
+/// be attached explicitly. Mirror fetch-models-dev.mjs: env vars first, then
+/// git's http.proxy.
+fn proxy_url() -> Option<String> {
+    let env = ["HTTPS_PROXY", "https_proxy", "HTTP_PROXY", "http_proxy"]
+        .iter()
+        .find_map(|k| std::env::var(k).ok().filter(|p| !p.trim().is_empty()));
+    env.or_else(git_http_proxy)
+}
+
+fn git_http_proxy() -> Option<String> {
+    let out = std::process::Command::new("git")
+        .args(["config", "--get", "http.proxy"])
+        .output()
+        .ok()?;
+    let s = String::from_utf8_lossy(&out.stdout).trim().to_string();
+    (!s.is_empty()).then_some(s)
+}
+
+fn http_client() -> Result<reqwest::Client, String> {
+    let mut b = reqwest::Client::builder()
+        .user_agent(concat!("KimiSwitch/", env!("CARGO_PKG_VERSION")))
+        .timeout(Duration::from_secs(60));
+    if let Some(p) = proxy_url() {
+        b = b.proxy(reqwest::Proxy::all(&p).map_err(|e| format!("invalid proxy '{p}': {e}"))?);
+    }
+    b.build().map_err(|e| format!("failed to build HTTP client: {e}"))
+}
+
+/// Download api.json, transform it and atomically replace the synced copy.
+/// On any failure the previous snapshot (synced or built-in) stays untouched.
+async fn sync_from_remote() -> Result<ModelsDevStatus, String> {
+    let client = http_client()?;
+    let raw: Value = client
+        .get(SOURCE_URL)
+        .header("Accept", "application/json")
+        .send()
+        .await
+        .map_err(|e| format!("request failed: {e}"))?
+        .error_for_status()
+        .map_err(|e| format!("models.dev returned {e}"))?
+        .json()
+        .await
+        .map_err(|e| format!("failed to parse api.json: {e}"))?;
+
+    let snapshot = build_snapshot(&raw)?;
+    let json = serde_json::to_string_pretty(&snapshot).map_err(|e| e.to_string())? + "\n";
+
+    let path = synced_path();
+    if let Some(dir) = path.parent() {
+        fs::create_dir_all(dir).map_err(|e| format!("failed to create {}: {e}", dir.display()))?;
+    }
+    let tmp = path.with_extension("json.tmp");
+    fs::write(&tmp, &json).map_err(|e| format!("failed to write {}: {e}", tmp.display()))?;
+    // Windows rename replaces an existing destination (MOVEFILE_REPLACE_EXISTING).
+    fs::rename(&tmp, &path).map_err(|e| {
+        let _ = fs::remove_file(&tmp);
+        format!("failed to replace {}: {e}", path.display())
+    })?;
+    Ok(status_of("synced", &json))
+}
+
+/// Refresh the models.dev snapshot from the network.
+#[tauri::command]
+pub async fn sync_models_dev() -> Result<ModelsDevStatus, String> {
+    sync_from_remote().await
+}
+
+/// Where the current reference data comes from (for the settings card).
+#[tauri::command]
+pub fn get_models_dev_status() -> Result<ModelsDevStatus, String> {
+    Ok(if read_synced().is_some() {
+        status("synced")
+    } else {
+        status("builtin")
+    })
+}
+
+/// The synced snapshot's raw JSON, or None when the frontend should keep
+/// using the bundled static asset.
+#[tauri::command]
+pub fn get_models_dev_snapshot() -> Result<Option<String>, String> {
+    Ok(read_synced())
+}
+
+/// Drop the synced copy and fall back to the compiled-in snapshot.
+#[tauri::command]
+pub fn reset_models_dev() -> Result<ModelsDevStatus, String> {
+    match fs::remove_file(synced_path()) {
+        Ok(()) => {}
+        Err(e) if e.kind() == std::io::ErrorKind::NotFound => {}
+        Err(e) => return Err(format!("failed to remove synced snapshot: {e}")),
+    }
+    Ok(status("builtin"))
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use serde_json::json;
+
+    #[test]
+    fn build_snapshot_picks_reference_fields() {
+        let raw = json!({
+            "moonshotai": {
+                "models": {
+                    "kimi-k3": {
+                        "name": "Kimi K3",
+                        "limit": { "context": 262144, "output": 16384 },
+                        "reasoning": true,
+                        "tool_call": true,
+                        "structured_output": true,
+                        "modalities": { "input": ["text", "image"], "output": ["text"] },
+                        "cost": { "input": 3.0, "output": 15.0, "cache_read": 0.3, "cache_write": 0.0 }
+                    },
+                    "kimi-image": {
+                        "name": "Kimi Image",
+                        "limit": { "context": 0 },
+                        "modalities": { "input": ["text", "image", "video"] }
+                    }
+                }
+            },
+            "zhipuai": {
+                "models": {
+                    "glm-5.3": {
+                        "name": "GLM-5.3",
+                        "reasoning": false,
+                        "cost": { "input": 0.6, "output": 2.2 }
+                    }
+                }
+            }
+        });
+
+        let snap = build_snapshot(&raw).unwrap();
+        let obj = snap.as_object().unwrap();
+        assert!(obj.contains_key("last_updated"));
+
+        let k3 = &obj["moonshotai/kimi-k3"];
+        assert_eq!(k3["name"], "Kimi K3");
+        assert_eq!(k3["context"], 262144);
+        assert_eq!(k3["reasoning"], true);
+        assert_eq!(k3["tool_call"], true);
+        assert_eq!(k3["structured_output"], true);
+        assert_eq!(k3["image"], true);
+        assert!(k3.get("video").is_none());
+        assert_eq!(k3["cost"]["input"], 3.0);
+        assert_eq!(k3["cost"]["cache_read"], 0.3);
+        assert_eq!(k3["cost"]["cache_write"], 0.0);
+
+        // Context 0 → dropped; video modality → flagged.
+        let img = &obj["moonshotai/kimi-image"];
+        assert!(img.get("context").is_none());
+        assert_eq!(img["video"], true);
+        assert!(img.get("cost").is_none());
+
+        // Explicit false is omitted (same as the mjs script).
+        let glm = &obj["zhipuai/glm-5.3"];
+        assert!(glm.get("reasoning").is_none());
+        assert_eq!(glm["cost"]["input"], 0.6);
+    }
+
+    #[test]
+    fn status_counts_models_and_providers() {
+        let raw = r#"{
+            "last_updated": "2026-09-21",
+            "moonshotai/kimi-k3": { "cost": { "input": 3.0 } },
+            "moonshotai/kimi-k2": { "name": "K2" },
+            "zhipuai/glm-5.3": {}
+        }"#;
+        let st = status_of("builtin", raw);
+        assert_eq!(st.last_updated, "2026-09-21");
+        assert_eq!(st.model_count, 3);
+        assert_eq!(st.provider_count, 2);
+    }
+
+    #[test]
+    fn num_serializes_whole_numbers_as_integers() {
+        assert_eq!(num(262144.0), json!(262144));
+        assert_eq!(num(0.3), json!(0.3));
+    }
+}
@@ -659,6 +659,7 @@ fn provision_managed_provider(region: KimiRegion, oauth_key: &str) -> Result<(),
            provider_type: crate::models::ProviderType::Kimi,
            base_url: None,
            api_key: None,
+            api_key_env: None,
            env: IndexMap::new(),
            note: None,
            official_url: None,
@@ -332,6 +332,9 @@ pub fn pi_file_to_config(file: &PiModelsFile) -> Config {
            provider_type,
            base_url: pi_provider.base_url.clone().filter(|s| !s.is_empty()),
            api_key: pi_provider.api_key.clone().filter(|s| !s.is_empty()),
+            // Pi's provider file has no env-var credential reference; the
+            // field stays empty so a Kimi-Code-only setting never leaks here.
+            api_key_env: None,
            env: IndexMap::new(),
            note: None,
            official_url: None,
@@ -9,6 +9,9 @@
 //! "weekly_limit" / "monthly_limit"），`used` = 已用百分比（0-100），`total` = 100，
 //! `remaining` = 剩余百分比，`resets_at` 为 ISO 8601 字符串。
 //!
+//! 例外：Kimi 的加力钱包（`PLAN_BOOSTER_WALLET`）不是百分比窗口而是金额行，
+//! `used` / `total` / `remaining` 均为货币金额（分换算而来），`unit` 为币种。
+//!
 //! 错误通道语义与 balance.rs 一致（Err = 瞬时，Ok(success:false) = 确定性）。

 use super::balance::{get_json, get_json_with_ua, AuthStyle, Fetched};
@@ -16,11 +19,17 @@ use super::usage_types::{UsageData, UsageResult};
 use std::time::Duration;

 // 套餐类 tier id 的唯一来源：所有套餐供应商（Kimi/智谱/MiniMax/OpenCode Go 及
-// 未来新增）都只用这三个 id。前端 src/lib/usage-display.ts 的 planLabel() 依赖
+// 未来新增）都只用这四个 id。前端 src/lib/usage-display.ts 的 planLabel() 依赖
 // 此约定做本地化映射——新增 tier id 时必须同步加映射。
 const TIER_FIVE_HOUR: &str = "five_hour";
 const TIER_WEEKLY_LIMIT: &str = "weekly_limit";
 const TIER_MONTHLY_LIMIT: &str = "monthly_limit";
+/// Kimi quota 模型的月度代码窗口（kimi-code #3787 后 `limit_month_code`）。
+const TIER_MONTH_CODE: &str = "month_code";
+/// Kimi quota 响应里的加力钱包（boosterWallet）。**不是百分比 tier**：这条是
+/// 金额行，前端 src/lib/usage-display.ts 的 planLabel() 负责本地化，
+/// UsageFooter 按「金额行」而非配额行渲染（used/total 是钱数，不是配额比例）。
+const PLAN_BOOSTER_WALLET: &str = "booster_wallet";

 /// 套餐条目的统一构造：按百分比表示用量。
 fn percent_tier(name: &str, used_percent: f64, resets_at: Option<String>) -> UsageData {
@@ -70,8 +79,14 @@ fn parse_f64(value: &serde_json::Value) -> Option<f64> {
 // GET {base_url}/usages
 //   默认 https://api.kimi.com/coding/v1/usages
 //   global: https://api.kimi.ai/coding/v1/usages
-// Response: { limits: [{ detail: { limit, remaining, resetTime } }],
-//             usage: { limit, remaining, resetTime } }
+// Response（kimi-code #3787 / 0.43.1 起服务端切换为 quota 模型）:
+//   { usages: { limit_5h, limit_7d, limit_month_total, limit_month_code:
+//               { used_ratio: 0-1, reset_time?: ISO 8601 } },
+//     boosterWallet: { balanceCents, totalCents, monthlyChargeLimitEnabled,
+//                      monthlyChargeLimitCents, monthlyUsedCents, currency },
+//     goods_version }
+// 旧结构（兼容保留）: { limits: [{ detail: { limit, remaining, resetTime } }],
+//                     usage: { limit, remaining, resetTime } }

 /// 由 base_url 拼接 usages 查询 URL；base_url 为空/空白时回退大陆默认。
 /// 纯函数，便于单测。
@@ -95,8 +110,9 @@ pub async fn query_kimi_coding(
        Fetched::Body(body) => {
            let tiers = parse_kimi_coding(&body);
            if tiers.is_empty() {
-                // 响应里没有可解析的套餐档位（limits/usage 缺失或字段变了）。
-                // 把原始响应（仅用量数字，无密钥）透出，方便对照接口结构修复。
+                // 响应里没有可解析的套餐档位（quota/usages 与旧 limits/usage
+                // 均缺失或字段变了）。把原始响应（仅用量数字，无密钥）透出，
+                // 方便对照接口结构修复。
                let preview = serde_json::to_string(&body)
                    .unwrap_or_else(|_| "<unserializable body>".into());
                let trimmed: String = preview.chars().take(400).collect();
@@ -111,6 +127,11 @@ pub async fn query_kimi_coding(
 }

 fn parse_kimi_coding(body: &serde_json::Value) -> Vec<UsageData> {
+    // 新 quota 模型（kimi-code #3787 / 0.43.1 起服务端下发）：优先解析。
+    if let Some(tiers) = parse_kimi_quota(body) {
+        return tiers;
+    }
+
    let mut tiers = Vec::new();

    // 5 小时窗口限额（优先显示）
@@ -130,6 +151,67 @@ fn parse_kimi_coding(body: &serde_json::Value) -> Vec<UsageData> {
    tiers
 }

+/// 解析 quota 模型的 `usages` 窗口映射（`limit_5h` / `limit_7d` /
+/// `limit_month_total` / `limit_month_code`）+ 加力钱包。窗口与钱包都缺失或
+/// 无可解析条目时返回 None，让调用方回退旧结构。
+fn parse_kimi_quota(body: &serde_json::Value) -> Option<Vec<UsageData>> {
+    let usages = body.get("usages")?;
+    let mut tiers = Vec::new();
+    for (key, tier) in [
+        ("limit_5h", TIER_FIVE_HOUR),
+        ("limit_7d", TIER_WEEKLY_LIMIT),
+        ("limit_month_total", TIER_MONTHLY_LIMIT),
+        ("limit_month_code", TIER_MONTH_CODE),
+    ] {
+        let Some(entry) = usages.get(key) else { continue };
+        let Some(ratio) = entry.get("used_ratio").and_then(parse_f64) else {
+            continue;
+        };
+        // used_ratio 规范为 0-1 比率；>1 时视为已是百分数，原样使用。
+        let used_percent = if ratio <= 1.0 { ratio * 100.0 } else { ratio };
+        let resets_at = entry.get("reset_time").and_then(extract_reset_time);
+        tiers.push(percent_tier(tier, used_percent, resets_at));
+    }
+    // 加力钱包排在四个窗口之后（钱包是余额，不是滚动窗口）。
+    if let Some(wallet) = body.get("boosterWallet").and_then(booster_wallet_tier) {
+        tiers.push(wallet);
+    }
+    if tiers.is_empty() { None } else { Some(tiers) }
+}
+
+/// 解析加力钱包（boosterWallet）为金额行。字段单位为「分」，统一换算成元/美元。
+///
+/// 只在响应含新版 `*Cents` 字段时产出条目：kimi-code #3787 之前的
+/// `boosterWallet` 只有 `balance`（无币种/精度语义），产出全 0 的伪行只会误导。
+fn booster_wallet_tier(wallet: &serde_json::Value) -> Option<UsageData> {
+    let cents = |key: &str| wallet.get(key).and_then(parse_f64);
+    let (balance, total, used) = (
+        cents("balanceCents"),
+        cents("totalCents"),
+        cents("monthlyUsedCents"),
+    );
+    if balance.is_none() && total.is_none() && used.is_none() {
+        return None;
+    }
+    // 币种缺失时不猜（前端 formatAmount 会退化为纯数字展示）。
+    let unit = wallet
+        .get("currency")
+        .and_then(|v| v.as_str())
+        .map(str::trim)
+        .filter(|s| !s.is_empty())
+        .map(str::to_string);
+    Some(UsageData {
+        plan_name: Some(PLAN_BOOSTER_WALLET.to_string()),
+        remaining: balance.map(|c| c / 100.0),
+        total: total.map(|c| c / 100.0),
+        used: used.map(|c| c / 100.0),
+        unit,
+        is_valid: Some(true),
+        // 钱包没有窗口重置时间。
+        resets_at: None,
+    })
+}
+
 fn kimi_limit_tier(name: &str, detail: &serde_json::Value) -> UsageData {
    let limit = detail.get("limit").and_then(parse_f64).unwrap_or(1.0);
    let remaining = detail.get("remaining").and_then(parse_f64).unwrap_or(0.0);
@@ -459,6 +541,97 @@ mod tests {
        );
    }

+    #[test]
+    fn kimi_coding_quota_model_four_windows() {
+        // kimi-code #3787 / 0.43.1 起服务端下发的 quota 模型：used_ratio 为
+        // 0-1 比率，reset_time 为 ISO 8601 字符串。
+        let body = json!({
+            "goods_version": 3,
+            "usages": {
+                "limit_5h": { "used_ratio": 0.6, "reset_time": "2026-09-15T17:00:00Z" },
+                "limit_7d": { "used_ratio": 0.25 },
+                "limit_month_total": { "used_ratio": 0.1, "reset_time": "2026-10-01T00:00:00Z" },
+                "limit_month_code": { "used_ratio": 0.88 }
+            },
+            "boosterWallet": { "balance": 0 }
+        });
+        let tiers = parse_kimi_coding(&body);
+        // 旧版 boosterWallet（只有 balance，无 *Cents）不产出钱包行。
+        assert_eq!(tiers.len(), 4);
+        assert_eq!(tiers[0].plan_name.as_deref(), Some("five_hour"));
+        assert_eq!(tiers[0].used, Some(60.0));
+        assert_eq!(tiers[0].remaining, Some(40.0));
+        assert_eq!(tiers[0].total, Some(100.0));
+        assert_eq!(
+            tiers[0].resets_at.as_deref(),
+            Some("2026-09-15T17:00:00Z")
+        );
+        assert_eq!(tiers[1].plan_name.as_deref(), Some("weekly_limit"));
+        assert_eq!(tiers[1].used, Some(25.0));
+        assert!(tiers[1].resets_at.is_none());
+        assert_eq!(tiers[2].plan_name.as_deref(), Some("monthly_limit"));
+        assert_eq!(tiers[2].used, Some(10.0));
+        assert_eq!(tiers[3].plan_name.as_deref(), Some("month_code"));
+        assert_eq!(tiers[3].used, Some(88.0));
+        assert_eq!(tiers[3].remaining, Some(12.0));
+    }
+
+    #[test]
+    fn kimi_coding_quota_appends_booster_wallet_row() {
+        // 加力钱包是金额行（分 → 元/美元），追加在四个窗口之后。
+        let body = json!({
+            "usages": {
+                "limit_5h": { "used_ratio": 0.6 },
+                "limit_7d": { "used_ratio": 0.25 },
+                "limit_month_total": { "used_ratio": 0.1 },
+                "limit_month_code": { "used_ratio": 0.88 }
+            },
+            "boosterWallet": {
+                "balanceCents": 1234,
+                "totalCents": 5000,
+                "monthlyChargeLimitEnabled": true,
+                "monthlyChargeLimitCents": 2000,
+                "monthlyUsedCents": 800,
+                "currency": "USD"
+            }
+        });
+        let tiers = parse_kimi_coding(&body);
+        assert_eq!(tiers.len(), 5);
+        assert_eq!(tiers[0].plan_name.as_deref(), Some("five_hour"));
+        let wallet = &tiers[4];
+        assert_eq!(wallet.plan_name.as_deref(), Some("booster_wallet"));
+        assert_eq!(wallet.used, Some(8.0));
+        assert_eq!(wallet.total, Some(50.0));
+        assert_eq!(wallet.remaining, Some(12.34));
+        assert_eq!(wallet.unit.as_deref(), Some("USD"));
+        assert_eq!(wallet.is_valid, Some(true));
+        assert!(wallet.resets_at.is_none());
+    }
+
+    #[test]
+    fn kimi_coding_quota_ratio_above_one_is_percent() {
+        // 防御：>1 的 used_ratio 视为已是百分数，不再乘 100。
+        let body = json!({
+            "usages": { "limit_5h": { "used_ratio": 42.0 } }
+        });
+        let tiers = parse_kimi_coding(&body);
+        assert_eq!(tiers.len(), 1);
+        assert_eq!(tiers[0].used, Some(42.0));
+    }
+
+    #[test]
+    fn kimi_coding_quota_missing_windows_falls_back_to_legacy() {
+        // usages 存在但没有任何可解析条目 → 回退旧结构解析。
+        let body = json!({
+            "usages": { "limit_5h": {} },
+            "usage": { "limit": 1000, "remaining": 900, "resetTime": "2026-08-01T00:00:00Z" }
+        });
+        let tiers = parse_kimi_coding(&body);
+        assert_eq!(tiers.len(), 1);
+        assert_eq!(tiers[0].plan_name.as_deref(), Some("weekly_limit"));
+        assert_eq!(tiers[0].used, Some(10.0));
+    }
+
    #[test]
    fn kimi_coding_reset_time_seconds_vs_millis() {
        // 秒级时间戳自动 ×1000
@@ -10,6 +10,7 @@

 pub mod balance;
 pub mod coding_plan;
+pub mod probe;
 pub mod usage_types;

 pub use usage_types::{UsageData, UsageResult};
@@ -336,6 +337,7 @@ mod tests {
            user_id: uid.map(str::to_string),
            auto_query_interval_minutes: None,
            timeout_seconds: None,
+            threshold: None,
        };
        // 缺 config / 缺任一项 / 空字符串 → None（确定性配置错误）
        assert_eq!(newapi_creds(None), None);
@@ -0,0 +1,155 @@
+//! 供应商推理端点探活（「一键体检」的第二步）。
+//!
+//! GET `{base_url}/v1/models`，带 `Authorization: Bearer <api_key>`。
+//! 2xx 且响应体是合法 JSON 即算通过（不校验内容结构，只验证端点能应答）；
+//! 其余状态码 / 网络错误归类成确定性错误文案。
+//!
+//! 错误文案与 balance.rs 一致用英文短句，前端 localizeHealthError 负责
+//! 双语展示（见 src/hooks/useHealthCheck.ts）。
+
+use std::time::Duration;
+
+/// 探活请求超时。体检要快速出结果、不卡住 UI，比账单查询的 8s 更短。
+pub const PROBE_TIMEOUT: Duration = Duration::from_secs(5);
+
+/// 由 base_url 拼出 `/v1/models` 探测地址。
+///
+/// 供应商配置里的 base_url 多数已经带 `/v1` 后缀（如
+/// `https://api.kimi.com/coding/v1`），直接再拼 `/v1/models` 会得到
+/// `/v1/v1/models`，所以先剥掉重复后缀；结尾斜杠与首尾空白一并归一。
+/// 空/纯空白返回 None（调用方直接判失败，不发请求）。纯函数，便于单测。
+fn models_url(base_url: &str) -> Option<String> {
+    let base = base_url.trim().trim_end_matches('/');
+    if base.is_empty() {
+        return None;
+    }
+    let root = base.strip_suffix("/v1").unwrap_or(base);
+    Some(format!("{root}/v1/models"))
+}
+
+/// 探活前置校验：返回最终请求 URL。base_url 缺失、api_key 为空都是确定性
+/// 配置错误，不发请求即失败（没有 key 时必然 401，直接给出更可操作的结论）。
+fn validate_probe_input(base_url: &str, api_key: &str) -> Result<String, String> {
+    let Some(url) = models_url(base_url) else {
+        return Err("missing base_url".to_string());
+    };
+    if api_key.trim().is_empty() {
+        return Err("no API key configured".to_string());
+    }
+    Ok(url)
+}
+
+/// 探活一个供应商的推理端点。`Ok(())` = 端点正常应答且返回合法 JSON。
+pub async fn probe_provider(
+    base_url: &str,
+    api_key: &str,
+    timeout: Duration,
+) -> Result<(), String> {
+    let url = validate_probe_input(base_url, api_key)?;
+
+    let client = reqwest::Client::builder()
+        .timeout(timeout)
+        .build()
+        .map_err(|e| format!("failed to build HTTP client: {e}"))?;
+
+    // 注意：api_key 只允许进请求头，严禁拼进 URL / 日志 / 错误信息。
+    let resp = client
+        .get(&url)
+        .header("Accept", "application/json")
+        .header("Authorization", format!("Bearer {api_key}"))
+        .send()
+        .await
+        .map_err(|e| {
+            if e.is_timeout() {
+                "request timed out".to_string()
+            } else if e.is_connect() {
+                "connection failed (DNS or refused)".to_string()
+            } else {
+                format!("network error: {e}")
+            }
+        })?;
+
+    let status = resp.status();
+    if status.is_success() {
+        // 先取完整响应体再解析：读体超时/中断是瞬时错误，与「返回的不是
+        // JSON」区分开（reqwest 的 .json() 会把两者都包成 decode）。
+        let raw = resp
+            .bytes()
+            .await
+            .map_err(|e| format!("failed to read response: {e}"))?;
+        return match serde_json::from_slice::<serde_json::Value>(&raw) {
+            Ok(_) => Ok(()),
+            Err(_) => Err("response is not valid JSON".to_string()),
+        };
+    }
+
+    match status.as_u16() {
+        401 | 403 => Err(format!("Authentication failed (HTTP {status})")),
+        404 => Err(format!("endpoint not found (HTTP {status})")),
+        code => Err(format!("API error (HTTP {code})")),
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn models_url_dedupes_v1_suffix() {
+        assert_eq!(
+            models_url("https://api.kimi.com/coding/v1").as_deref(),
+            Some("https://api.kimi.com/coding/v1/models")
+        );
+        // 不带 /v1 的裸域补上 /v1
+        assert_eq!(
+            models_url("https://api.deepseek.com").as_deref(),
+            Some("https://api.deepseek.com/v1/models")
+        );
+    }
+
+    #[test]
+    fn models_url_normalizes_slashes_and_whitespace() {
+        // 结尾斜杠（含多个）与首尾空白都要归一，且不得拼出 /v1/v1
+        assert_eq!(
+            models_url("https://api.kimi.com/coding/v1/").as_deref(),
+            Some("https://api.kimi.com/coding/v1/models")
+        );
+        assert_eq!(
+            models_url("  https://api.deepseek.com/v1///  ").as_deref(),
+            Some("https://api.deepseek.com/v1/models")
+        );
+        assert_eq!(
+            models_url("https://api.deepseek.com/").as_deref(),
+            Some("https://api.deepseek.com/v1/models")
+        );
+        // 非 /v1 的路径后缀原样保留（只有 /v1 需要去重）
+        assert_eq!(
+            models_url("https://api.novita.ai/v3").as_deref(),
+            Some("https://api.novita.ai/v3/v1/models")
+        );
+    }
+
+    #[test]
+    fn models_url_rejects_empty_base() {
+        assert_eq!(models_url(""), None);
+        assert_eq!(models_url("   "), None);
+        assert_eq!(models_url("/"), None);
+    }
+
+    #[test]
+    fn probe_input_validation_rejects_empty_base_and_missing_key() {
+        // 两者都是确定性配置错误，不发请求即返回。
+        assert_eq!(
+            validate_probe_input("  ", "sk-x").unwrap_err(),
+            "missing base_url"
+        );
+        assert_eq!(
+            validate_probe_input("https://example.com", "   ").unwrap_err(),
+            "no API key configured"
+        );
+        assert_eq!(
+            validate_probe_input("https://example.com/v1", "sk-x").unwrap(),
+            "https://example.com/v1/models"
+        );
+    }
+}
@@ -1,6 +1,6 @@
 {
  "productName": "Kimi Switch",
-  "version": "0.7.17",
+  "version": "0.7.22",
  "identifier": "com.kimiswitch.app",
  "build": {
    "beforeDevCommand": "npm run dev",
@@ -78,6 +78,10 @@ export function AgentSettingsPanel({ rawOther, onChange }: AgentSettingsPanelPro
    update({ background: { ...settings.background, ...patch } });
  };

+  const updateWatch = (patch: Partial<AgentSettings["watch"]>) => {
+    update({ watch: { ...settings.watch, ...patch } });
+  };
+
  const setRules = (rules: PermissionRule[]) => {
    update({ permission: { rules } });
  };
@@ -141,6 +145,15 @@ export function AgentSettingsPanel({ rawOther, onChange }: AgentSettingsPanelPro
          value={settings.loop_control?.reserved_context_size ?? 50000}
          onChange={(v) => updateLoopControl({ reserved_context_size: v })}
        />
+        {/* kimi-code 0.43.0+ `loop_control.compaction_max_attempts`. The
+            upstream default is 5; an empty field or 0 removes the key instead
+            of writing a literal 0 (see setAgentSettings). */}
+        <NumberField
+          label={t("compactionMaxAttempts")}
+          value={settings.loop_control?.compaction_max_attempts ?? 5}
+          onChange={(v) => updateLoopControl({ compaction_max_attempts: v })}
+        />
+        <p className="text-xs text-content-muted">{t("compactionMaxAttemptsHint")}</p>
      </Card>

      <Card title={t("backgroundSettings")}>
@@ -156,6 +169,20 @@ export function AgentSettingsPanel({ rawOther, onChange }: AgentSettingsPanelPro
        />
      </Card>

+      <Card title={t("watchSettings")}>
+        {/* Top-level `[watch] enabled` (kimi-code 2.0.1+; off by default since
+            2.0.2). KIMI_CODE_WATCH outranks this config at runtime — the env
+            var is probed by get_experimental_env_status as a non-flag entry
+            but deliberately locks nothing here, the toggle just documents the
+            config value that applies when the env var is unset. */}
+        <Checkbox
+          label={t("watchEnabled")}
+          checked={settings.watch?.enabled ?? false}
+          onChange={(checked) => updateWatch({ enabled: checked })}
+        />
+        <p className="text-xs text-content-muted">{t("watchEnabledDesc")}</p>
+      </Card>
+
      <Card title={t("permissionRules")}>
        {/* kimi-code `[permission] dangerous_command_guard`. The env
            var KIMI_CODE_DANGEROUS_COMMAND_GUARD (literal "true"/"false")
@@ -11,6 +11,7 @@ import { AgentSettingsPanel } from "./AgentSettingsPanel";
 import { KimiOAuthDialog } from "./KimiOAuthDialog";
 import { ProviderIcon } from "./ProviderIcon";
 import { IconPicker } from "./IconPicker";
+import { Segmented } from "./ui/controls";
 import type { Agent, DiscoveredModel, Model, Provider, ProviderType } from "../types";

 const KNOWN_CAPABILITIES = [
@@ -127,6 +128,19 @@ export function ProviderEdit({
  // Preset this provider was created from (if any) — provides the
  // "Get API Key" / referral links shown under the key input.
  const preset = findPresetForProvider(provider);
+  /**
+   * Credential source. `api_key_env` (kimi-code 2.0.0+) names an environment
+   * variable instead of storing the key in config.toml; the two are mutually
+   * exclusive upstream, so the UI is a mode switch and each mode clears the
+   * other field. Derived from the provider itself (no local state) so the
+   * switch follows provider switches and reloads; an empty-but-present value
+   * ("env mode picked, name not typed yet") still reads as env mode.
+   * Only Kimi Code understands `api_key_env` — Pi's provider file has no
+   * equivalent, so the switch and the env input are both gated on the agent
+   * (a Pi provider that somehow carries the key keeps its plain key input).
+   */
+  const apiKeyEnvSupported = agent === "kimi_code";
+  const apiKeyEnvMode = apiKeyEnvSupported && provider.api_key_env != null;

  useEffect(() => {
    const def = defaultBaseUrl(agent, provider.provider_type);
@@ -322,27 +336,73 @@ export function ProviderEdit({

                {!provider.managed && (
                  <div>
+                    {apiKeyEnvSupported && (
+                      <div className="flex items-center gap-3 mb-2">
+                        <span className="text-sm text-content-muted">
+                          {t("apiKeySource")}
+                        </span>
+                        <Segmented
+                          options={[
+                            { key: "direct", label: t("apiKeySourceDirect") },
+                            { key: "env", label: t("apiKeySourceEnv") },
+                          ]}
+                          value={apiKeyEnvMode ? "env" : "direct"}
+                          onChange={(mode) => {
+                            // Mutual exclusion: picking a source clears the
+                            // other one so export never writes both keys.
+                            if (mode === "env") {
+                              onChange({
+                                ...provider,
+                                api_key: null,
+                                api_key_env: provider.api_key_env ?? "",
+                              });
+                            } else {
+                              onChange({ ...provider, api_key_env: null });
+                            }
+                          }}
+                        />
+                      </div>
+                    )}
                    <label htmlFor={apiKeyId} className="block text-sm text-content-muted mb-1.5">
-                      {t("apiKey")} <span className="text-red-500">*</span>
+                      {apiKeyEnvMode ? t("apiKeyEnvName") : t("apiKey")}{" "}
+                      <span className="text-red-500">*</span>
                    </label>
-                    <div className="relative">
+                    {apiKeyEnvMode ? (
                      <input
                        id={apiKeyId}
-                        type={showApiKey ? "text" : "password"}
-                        className="w-full bg-input border border-border rounded-lg px-3 py-2 pr-10 text-sm focus:ring-2 focus:ring-blue-500 focus:outline-none"
-                        value={provider.api_key || ""}
+                        type="text"
+                        className="w-full bg-input border border-border rounded-lg px-3 py-2 text-sm font-mono focus:ring-2 focus:ring-blue-500 focus:outline-none"
+                        value={provider.api_key_env ?? ""}
+                        placeholder="KIMI_API_KEY"
                        onChange={(e) =>
-                          onChange({ ...provider, api_key: e.target.value || null })
+                          onChange({ ...provider, api_key_env: e.target.value })
                        }
                      />
-                      <button
-                        type="button"
-                        onClick={() => setShowApiKey((v) => !v)}
-                        className="absolute right-2 top-1/2 -translate-y-1/2 text-content-muted hover:text-content-primary text-xs px-1"
-                      >
-                        {showApiKey ? t("hide") : t("show")}
-                      </button>
-                    </div>
+                    ) : (
+                      <div className="relative">
+                        <input
+                          id={apiKeyId}
+                          type={showApiKey ? "text" : "password"}
+                          className="w-full bg-input border border-border rounded-lg px-3 py-2 pr-10 text-sm focus:ring-2 focus:ring-blue-500 focus:outline-none"
+                          value={provider.api_key || ""}
+                          onChange={(e) =>
+                            onChange({ ...provider, api_key: e.target.value || null })
+                          }
+                        />
+                        <button
+                          type="button"
+                          onClick={() => setShowApiKey((v) => !v)}
+                          className="absolute right-2 top-1/2 -translate-y-1/2 text-content-muted hover:text-content-primary text-xs px-1"
+                        >
+                          {showApiKey ? t("hide") : t("show")}
+                        </button>
+                      </div>
+                    )}
+                    {apiKeyEnvMode && (
+                      <p className="mt-1.5 text-xs text-content-muted">
+                        {t("apiKeyEnvHint")}
+                      </p>
+                    )}
                    {(() => {
                      const linkUrl = preset?.apiKeyUrl ?? provider.official_url ?? null;
                      if (!linkUrl) return null;
@@ -1,10 +1,26 @@
 import { useState } from "react";
 import { invoke } from "@tauri-apps/api/core";
-import { Pencil, Copy, Activity, Loader2, Trash2, BarChart3 } from "lucide-react";
+import {
+  Pencil,
+  Copy,
+  Activity,
+  Loader2,
+  Trash2,
+  BarChart3,
+  Stethoscope,
+} from "lucide-react";
 import { useTranslation } from "../i18n";
 import { ProviderIcon } from "./ProviderIcon";
 import { UsageFooter } from "./UsageFooter";
 import { useUsageQuery } from "../hooks/useUsageQuery";
+import {
+  useHealthCheck,
+  localizeHealthError,
+  probeLine,
+  usageLine,
+  healthDotClass,
+  type HealthResult,
+} from "../hooks/useHealthCheck";
 import type { Agent, Model, Provider } from "../types";

 interface ConnectivityResult {
@@ -40,6 +56,12 @@ interface ProviderCardProps {
  models: Record<string, Model>;
  /** Inline connectivity-test state for this provider, or undefined when idle. */
  ts: TestState | undefined;
+  /** 一键体检结论；undefined = 本轮未出结果（检测中或未检测）。 */
+  health: HealthResult | undefined;
+  /** 体检请求进行中。 */
+  healthRunning: boolean;
+  /** 体检至少跑过一轮（决定是否渲染灰色「未检测」点）。 */
+  healthRan: boolean;
  onEdit: (name: string) => void;
  onDelete: (name: string) => void;
  onDuplicate: (name: string) => void;
@@ -54,6 +76,9 @@ function ProviderCard({
  defaultModel,
  models,
  ts,
+  health,
+  healthRunning,
+  healthRan,
  onEdit,
  onDelete,
  onDuplicate,
@@ -70,7 +95,10 @@ function ProviderCard({
    provider.name,
    provider.usageKinds,
    provider.usageConfig?.autoQueryIntervalMinutes,
-    provider.usageConfig?.enabled === false
+    provider.usageConfig?.enabled === false,
+    provider.usageConfig?.threshold,
+    provider.usageConfig?.templateType === "newapi" ||
+      provider.usageConfig?.templateType === "sub2api"
  );
  const providerModels = Object.values(models).filter(
    (m) => m.provider === provider.name
@@ -80,6 +108,17 @@ function ProviderCard({
    : null;
  const isActive = provider.active === true;

+  // 一键体检状态点 tooltip：两项检测各自如实展示——探活可能是「不支持模型
+  // 列表」（中性），账单可能是「失败: xxx」，都不掩盖。
+  const healthTip = health
+    ? [
+        health.probe ? probeLine(health.probe, health.latencyMs, t) : null,
+        health.usage ? usageLine(health.usage, t) : null,
+      ]
+        .filter((s): s is string => s !== null)
+        .join(" · ")
+    : "";
+
  return (
    <div
      className={`group relative flex flex-col gap-3 p-4 rounded-xl border transition-colors cursor-pointer w-full ${
@@ -102,6 +141,34 @@ function ProviderCard({
          <h3 className="font-semibold text-content-primary truncate">
            {provider.name}
          </h3>
+          {/* 一键体检状态点：绿=正常、红=失败、灰=未判定（未检测 / 检测中） */}
+          {health ? (
+            <span
+              role="img"
+              className={`w-2.5 h-2.5 shrink-0 rounded-full ${healthDotClass(
+                health.verdict
+              )}`}
+              title={healthTip}
+              aria-label={healthTip}
+            />
+          ) : healthRunning ? (
+            <span
+              className="shrink-0 flex items-center"
+              title={t("healthCheckRunning")}
+              aria-label={t("healthCheckRunning")}
+            >
+              <Loader2 className="w-3.5 h-3.5 animate-spin text-content-muted" />
+            </span>
+          ) : healthRan ? (
+            <span
+              role="img"
+              className={`w-2.5 h-2.5 shrink-0 rounded-full ${healthDotClass(
+                "neutral"
+              )}`}
+              title={t("healthNotChecked")}
+              aria-label={t("healthNotChecked")}
+            />
+          ) : null}
          {isActive && (
            <span className="text-xs px-2 py-0.5 rounded-full bg-green-100 dark:bg-green-900/30 text-green-600 dark:text-green-400 border border-green-300 dark:border-green-500/30">
              {t("inUse")}
@@ -156,7 +223,7 @@ function ProviderCard({
        )}
        {/* compact usage summary — sits left of the switch button,
            mirroring cc-switch's card layout (usage → action buttons) */}
-        {(provider.usageKinds?.length ?? 0) > 0 && (
+        {usage.supported && (
          <UsageFooter usage={usage} variant="compact" />
        )}
        <button
@@ -273,6 +340,7 @@ export function ProviderList({
 }: ProviderListProps) {
  const { t } = useTranslation();
  const [testState, setTestState] = useState<Record<string, TestState>>({});
+  const health = useHealthCheck(agent);

  // managed（OAuth 登录托管）供应商排在最前；其余保持原有顺序（稳定排序）。
  const sortedProviders = [...providers].sort(
@@ -312,28 +380,69 @@ export function ProviderList({

  return (
    <div className="h-full flex flex-col">
-      <div className="flex items-center justify-between px-4 py-3 border-b border-border">
-        <h2 className="font-medium text-content-primary">{t("providers")}</h2>
-        <button
-          type="button"
-          onClick={onAdd}
-          title={t("addProvider")}
-          className="w-9 h-9 flex items-center justify-center rounded-full bg-orange-500 hover:bg-orange-600 text-white shadow-lg focus:ring-2 focus:ring-orange-400 focus:outline-none transition-colors"
-        >
-          <svg
-            width="20"
-            height="20"
-            viewBox="0 0 24 24"
-            fill="none"
-            stroke="currentColor"
-            strokeWidth="2.5"
-            strokeLinecap="round"
-            strokeLinejoin="round"
-            aria-hidden="true"
+      <div className="flex items-center justify-between gap-3 px-4 py-3 border-b border-border">
+        <div className="flex items-center gap-3 min-w-0">
+          <h2 className="font-medium text-content-primary shrink-0">
+            {t("providers")}
+          </h2>
+          {/* 体检汇总：检测中 / 请求失败 / x 正常 / y 失败 */}
+          {(health.running || health.ran || health.error) && (
+            <span
+              className={`text-xs tabular-nums truncate ${
+                health.error ? "text-red-500 dark:text-red-400" : "text-content-muted"
+              }`}
+              title={health.error ? localizeHealthError(health.error, t) : undefined}
+            >
+              {health.error
+                ? t("healthRequestFailed", {
+                    error: localizeHealthError(health.error, t),
+                  })
+                : health.running
+                  ? t("healthCheckRunning")
+                  : t("healthSummary", {
+                      ok: health.okCount,
+                      fail: health.failCount,
+                    })}
+            </span>
+          )}
+        </div>
+        <div className="flex items-center gap-2 shrink-0">
+          <button
+            type="button"
+            onClick={health.run}
+            disabled={health.running}
+            title={t("healthCheckHint")}
+            aria-label={t("healthCheck")}
+            className="flex items-center gap-1.5 h-9 px-3 rounded-full border border-border text-sm text-content-muted hover:text-content-primary hover:bg-hover-2 disabled:opacity-50 focus:ring-2 focus:ring-blue-500 focus:outline-none transition-colors whitespace-nowrap"
          >
-            <path d="M12 5v14M5 12h14" />
-          </svg>
-        </button>
+            {health.running ? (
+              <Loader2 className="w-4 h-4 animate-spin" />
+            ) : (
+              <Stethoscope className="w-4 h-4" />
+            )}
+            {t("healthCheck")}
+          </button>
+          <button
+            type="button"
+            onClick={onAdd}
+            title={t("addProvider")}
+            className="w-9 h-9 flex items-center justify-center rounded-full bg-orange-500 hover:bg-orange-600 text-white shadow-lg focus:ring-2 focus:ring-orange-400 focus:outline-none transition-colors"
+          >
+            <svg
+              width="20"
+              height="20"
+              viewBox="0 0 24 24"
+              fill="none"
+              stroke="currentColor"
+              strokeWidth="2.5"
+              strokeLinecap="round"
+              strokeLinejoin="round"
+              aria-hidden="true"
+            >
+              <path d="M12 5v14M5 12h14" />
+            </svg>
+          </button>
+        </div>
      </div>

      <div className="flex-1 overflow-auto p-4">
@@ -371,6 +480,9 @@ export function ProviderList({
                defaultModel={defaultModel}
                models={models}
                ts={testState[provider.name]}
+                health={health.results[provider.name]}
+                healthRunning={health.running}
+                healthRan={health.ran}
                onEdit={onEdit}
                onDelete={onDelete}
                onDuplicate={onDuplicate}
@@ -5,12 +5,20 @@ import { listen } from "@tauri-apps/api/event";
 import { useTranslation, type Language } from "../i18n";
 import { useTheme, type Theme } from "../hooks/useTheme";
 import type { UpdateInfo } from "../hooks/useUpdateCheck";
+import {
+  useKimiCodeVersion,
+  isNewerVersion,
+  type KimiCodeStatus,
+} from "../hooks/useKimiCodeVersion";

 function formatVersion(version: string): string {
  const v = version.trim();
  return v.startsWith("v") || v.startsWith("V") ? v : `v${v}`;
 }

+/** 上游 kimi-code releases 页面。 */
+const KIMI_CODE_RELEASES_URL = "https://github.com/MoonshotAI/kimi-code/releases";
+
 interface SettingsModalProps {
  open: boolean;
  onClose: () => void;
@@ -30,6 +38,11 @@ export function SettingsModal({
 }: SettingsModalProps) {
  const { t, lang, setLang } = useTranslation();
  const { theme, setTheme } = useTheme();
+  const {
+    info: kimiCodeInfo,
+    checking: kimiCodeChecking,
+    checkNow: checkKimiCode,
+  } = useKimiCodeVersion();
  const [appVersion, setAppVersion] = useState("");
  const [downloadProgress, setDownloadProgress] = useState<number | null>(null);
  const [downloadedPath, setDownloadedPath] = useState<string | null>(null);
@@ -158,6 +171,47 @@ export function SettingsModal({
    return d.toLocaleString(lang === "zh" ? "zh-CN" : "en-US");
  };

+  // 状态徽标：绿=已适配 / 黄=建议升级 KimiSwitch / 橙=部分兼容 / 灰=未知或未安装
+  const kimiCodeBadge = (status: KimiCodeStatus) => {
+    switch (status) {
+      case "current":
+        return {
+          label: t("kimiCodeStatusCurrent"),
+          className:
+            "bg-green-100 dark:bg-green-900/30 text-green-600 dark:text-green-400 border-green-300 dark:border-green-500/30",
+        };
+      case "outdated":
+        return {
+          label: t("kimiCodeStatusOutdated"),
+          className:
+            "bg-yellow-100 dark:bg-yellow-900/30 text-yellow-700 dark:text-yellow-400 border-yellow-300 dark:border-yellow-500/30",
+        };
+      case "partial":
+        return {
+          label: t("kimiCodeStatusPartial"),
+          className:
+            "bg-orange-100 dark:bg-orange-900/30 text-orange-600 dark:text-orange-400 border-orange-300 dark:border-orange-500/30",
+        };
+      case "unknown":
+        return {
+          label: t("kimiCodeStatusUnknown"),
+          className: "bg-hover-2 text-content-muted border-border",
+        };
+      default:
+        return {
+          label: t("kimiCodeStatusNotInstalled"),
+          className: "bg-hover-2 text-content-muted border-border",
+        };
+    }
+  };
+
+  const kimiCodeUpdateAvailable =
+    !!kimiCodeInfo?.installed &&
+    !!kimiCodeInfo?.latest &&
+    isNewerVersion(kimiCodeInfo.latest, kimiCodeInfo.installed);
+
+  const kimiCodeBadgeInfo = kimiCodeInfo ? kimiCodeBadge(kimiCodeInfo.status) : null;
+
  return createPortal(
    <div
      className="fixed inset-0 z-[9999] flex items-center justify-center bg-black/60 backdrop-blur-sm p-4"
@@ -362,6 +416,72 @@ export function SettingsModal({
            </div>
          </section>

+          {/* Kimi Code CLI version & compatibility */}
+          <section>
+            <h3 className="mb-3 text-[11px] uppercase tracking-wider text-content-muted">
+              {t("kimiCodeSection")}
+            </h3>
+            <div className="space-y-2 rounded-lg border border-border bg-input p-3">
+              <div className="flex items-center justify-between text-sm">
+                <span className="text-content-muted">{t("kimiCodeInstalled")}</span>
+                <span className="font-medium text-content-primary tabular-nums">
+                  {kimiCodeInfo?.installed ? formatVersion(kimiCodeInfo.installed) : "—"}
+                </span>
+              </div>
+              <div className="flex items-center justify-between text-sm">
+                <span className="text-content-muted">{t("kimiCodeLatest")}</span>
+                <span className="font-medium text-content-primary tabular-nums">
+                  {kimiCodeInfo?.latest ? formatVersion(kimiCodeInfo.latest) : "—"}
+                </span>
+              </div>
+              <div className="flex items-center justify-between text-sm">
+                <span className="text-content-muted">{t("kimiCodeStatusLabel")}</span>
+                {kimiCodeBadgeInfo ? (
+                  <span
+                    className={`text-xs px-2 py-0.5 rounded-full border ${kimiCodeBadgeInfo.className}`}
+                  >
+                    {kimiCodeBadgeInfo.label}
+                  </span>
+                ) : (
+                  <span className="text-xs text-content-muted">—</span>
+                )}
+              </div>
+              <div className="pt-2 flex items-center gap-2">
+                <button
+                  type="button"
+                  onClick={checkKimiCode}
+                  disabled={kimiCodeChecking}
+                  className="px-3 py-1.5 text-sm rounded border border-border hover:bg-hover-2 disabled:opacity-50 transition-colors flex items-center gap-1.5"
+                >
+                  <svg
+                    width="14"
+                    height="14"
+                    viewBox="0 0 24 24"
+                    fill="none"
+                    stroke="currentColor"
+                    strokeWidth="2"
+                    strokeLinecap="round"
+                    strokeLinejoin="round"
+                    className={kimiCodeChecking ? "animate-spin" : ""}
+                  >
+                    <path d="M21 12a9 9 0 1 1-2.64-6.36" />
+                    <polyline points="21 3 21 9 15 9" />
+                  </svg>
+                  {kimiCodeChecking ? t("checking") : t("kimiCodeCheck")}
+                </button>
+                {kimiCodeUpdateAvailable && (
+                  <button
+                    type="button"
+                    onClick={() => openExternal(KIMI_CODE_RELEASES_URL)}
+                    className="px-3 py-1.5 text-sm rounded border border-border hover:bg-hover-2 transition-colors"
+                  >
+                    {t("kimiCodeOpenReleases")}
+                  </button>
+                )}
+              </div>
+            </div>
+          </section>
+
          {/* About */}
          <section>
            <h3 className="mb-3 text-[11px] uppercase tracking-wider text-content-muted">
@@ -24,6 +24,15 @@ import {
  type ValidationErrorKey,
 } from "../lib/subagent-settings";
 import { Card, Toggle } from "./ui/controls";
+import { applyModelsDevSnapshot, reloadModelsDev } from "../lib/models-dev";
+
+/** Mirror of models_dev::ModelsDevStatus (camelCase over IPC). */
+interface ModelsDevStatus {
+  source: "synced" | "builtin";
+  lastUpdated: string;
+  modelCount: number;
+  providerCount: number;
+}

 interface SubagentSettingsPageProps {
  /** config.raw_other — hosts the `[experimental]` and `[secondary_model]` sections. */
@@ -48,10 +57,6 @@ const FLAG_LABELS: Record<string, { name: TranslationKey; desc: TranslationKey }
  notify_user: { name: "flagNotifyUser", desc: "flagNotifyUserDesc" },
  tower: { name: "flagTower", desc: "flagTowerDesc" },
  subagent_fork: { name: "flagSubagentFork", desc: "flagSubagentForkDesc" },
-  auto_session_title: {
-    name: "flagAutoSessionTitle",
-    desc: "flagAutoSessionTitleDesc",
-  },
 };

 /** Validation-error i18n key per engine rule, for the pre-write self-check. */
@@ -159,13 +164,65 @@ export function SubagentSettingsPage({
  const [addSelection, setAddSelection] = useState("");
  /** WebUI-open button in flight; disables both buttons while non-null. */
  const [webuiBusy, setWebuiBusy] = useState<"embedded" | "browser" | null>(null);
+  /** models.dev snapshot provenance (synced copy vs bundled asset). */
+  const [modelsDevStatus, setModelsDevStatus] = useState<ModelsDevStatus | null>(null);
+  /** Sync/restore button in flight; blocks the card's buttons. */
+  const [modelsDevBusy, setModelsDevBusy] = useState(false);
+  /** Last sync error (inline) and success flag (auto-cleared on next action). */
+  const [syncError, setSyncError] = useState<string | null>(null);
+  const [syncOk, setSyncOk] = useState(false);

  useEffect(() => {
    invoke<ExperimentalEnvStatus>("get_experimental_env_status")
      .then(setEnv)
      .catch(() => setEnv({}));
+    invoke<ModelsDevStatus>("get_models_dev_status")
+      .then(setModelsDevStatus)
+      .catch(() => setModelsDevStatus(null));
  }, []);

+  const refreshModelsDevStatus = () => {
+    invoke<ModelsDevStatus>("get_models_dev_status")
+      .then(setModelsDevStatus)
+      .catch(() => {});
+  };
+
+  const handleModelsDevSync = async () => {
+    setModelsDevBusy(true);
+    setSyncError(null);
+    setSyncOk(false);
+    try {
+      const st = await invoke<ModelsDevStatus>("sync_models_dev");
+      setModelsDevStatus(st);
+      // Hot-swap the frontend index; the dashboard price index rebuilds on
+      // its next query (mtime-triggered on the Rust side).
+      const raw = await invoke<string | null>("get_models_dev_snapshot");
+      if (raw) applyModelsDevSnapshot(JSON.parse(raw) as Record<string, unknown>);
+      setSyncOk(true);
+    } catch (err) {
+      setSyncError(err instanceof Error ? err.message : String(err));
+    } finally {
+      setModelsDevBusy(false);
+    }
+  };
+
+  const handleModelsDevRestore = async () => {
+    if (!confirm(t("modelsDataRestoreConfirm"))) return;
+    setModelsDevBusy(true);
+    setSyncError(null);
+    setSyncOk(false);
+    try {
+      const st = await invoke<ModelsDevStatus>("reset_models_dev");
+      setModelsDevStatus(st);
+      await reloadModelsDev();
+    } catch (err) {
+      setSyncError(err instanceof Error ? err.message : String(err));
+      refreshModelsDevStatus();
+    } finally {
+      setModelsDevBusy(false);
+    }
+  };
+
  const flags = getExperimentalFlags(rawOther);
  const pool = getSubagentModelPool(rawOther);
  const poolKeys = pool ? Object.keys(pool.models) : [];
@@ -402,6 +459,66 @@ export function SubagentSettingsPage({
          </div>
        </Card>

+        {/* models.dev reference data: online sync / restore bundled */}
+        <Card title={t("modelsDataSection")}>
+          <p className="text-xs text-content-muted">{t("modelsDataDesc")}</p>
+          {modelsDevStatus && (
+            <div className="flex flex-wrap items-center gap-2 text-sm">
+              <span
+                className={`shrink-0 rounded border px-1.5 py-0.5 text-xs ${
+                  modelsDevStatus.source === "synced"
+                    ? "border-blue-500/30 text-blue-600 dark:text-blue-400"
+                    : "border-border text-content-muted"
+                }`}
+              >
+                {modelsDevStatus.source === "synced"
+                  ? t("modelsDataSourceSynced")
+                  : t("modelsDataSourceBuiltin")}
+              </span>
+              <span className="text-xs text-content-muted">
+                {t("modelsDataStats", {
+                  date: modelsDevStatus.lastUpdated,
+                  models: modelsDevStatus.modelCount.toLocaleString(),
+                  providers: modelsDevStatus.providerCount.toLocaleString(),
+                })}
+              </span>
+            </div>
+          )}
+          {syncOk && (
+            <div className="bg-green-100 dark:bg-green-900/20 border border-green-300 dark:border-green-500/30 rounded-lg px-3 py-2 text-xs text-green-700 dark:text-green-400">
+              {t("modelsDataSyncOk")}
+            </div>
+          )}
+          {syncError && (
+            <div
+              role="alert"
+              className="bg-red-100 dark:bg-red-900/20 border border-red-300 dark:border-red-500/30 rounded-lg px-3 py-2 text-xs text-red-700 dark:text-red-400"
+            >
+              {t("modelsDataSyncFailed")}: {syncError}
+            </div>
+          )}
+          <div className="flex flex-wrap gap-2">
+            <button
+              type="button"
+              disabled={modelsDevBusy}
+              onClick={handleModelsDevSync}
+              className="px-3 py-1.5 text-sm rounded bg-blue-600 text-white hover:bg-blue-500 focus:ring-2 focus:ring-blue-500 focus:outline-none disabled:opacity-50 disabled:cursor-not-allowed"
+            >
+              {modelsDevBusy ? t("modelsDataSyncing") : t("modelsDataSyncNow")}
+            </button>
+            {modelsDevStatus?.source === "synced" && (
+              <button
+                type="button"
+                disabled={modelsDevBusy}
+                onClick={handleModelsDevRestore}
+                className="px-3 py-1.5 text-sm border border-border rounded hover:bg-hover-2 focus:ring-2 focus:ring-blue-500 focus:outline-none disabled:opacity-50 disabled:cursor-not-allowed"
+              >
+                {t("modelsDataRestore")}
+              </button>
+            )}
+          </div>
+        </Card>
+
        {/* Subagent: secondary model picker / pool + inherited settings */}
        <Card title={t("secondaryModelSection")}>
          {isPoolView && pool ? (
@@ -364,7 +364,7 @@ export function UsageConfigModal({ open, agent, provider, onClose, onSave }: Usa
                )}
              </div>

-              {/* timeout + auto query interval */}
+              {/* timeout + auto query interval + alert threshold */}
              <div className="rounded-lg border border-border px-4 py-4 grid grid-cols-2 gap-6">
                <div>
                  <label className="block text-sm text-content-muted mb-1.5">{t("usageTimeout")}</label>
@@ -394,6 +394,31 @@ export function UsageConfigModal({ open, agent, provider, onClose, onSave }: Usa
                  />
                  <p className="mt-1 text-xs text-content-muted">{t("usageAutoIntervalHint")}</p>
                </div>
+                <div>
+                  <label className="block text-sm text-content-muted mb-1.5">{t("usageThreshold")} %</label>
+                  <input
+                    type="number"
+                    min={0}
+                    max={100}
+                    className="w-32 bg-input border border-border rounded-lg px-3 py-2 text-sm focus:ring-2 focus:ring-blue-500 focus:outline-none"
+                    value={cfg.threshold ?? ""}
+                    onChange={(e) => {
+                      const raw = e.target.value.trim();
+                      // 清空输入框写回 undefined（关闭），而不是 0：两者判定等价，
+                      // 但 undefined 会被 serde 的 skip_serializing_if 从 SQLite
+                      // JSON 里省掉，不留冗余字段。
+                      if (raw === "") {
+                        setCfg({ ...cfg, threshold: undefined });
+                        return;
+                      }
+                      const n = parseInt(raw, 10);
+                      // 非数字（如 "1e" / "-"）不落库，保留上一次的有效值。
+                      if (Number.isNaN(n)) return;
+                      setCfg({ ...cfg, threshold: Math.min(100, Math.max(0, n)) });
+                    }}
+                  />
+                  <p className="mt-1 text-xs text-content-muted">{t("usageThresholdHint")}</p>
+                </div>
              </div>
            </>
          )}
@@ -1,7 +1,13 @@
 import { useEffect, useState } from "react";
 import { Loader2, RefreshCw } from "lucide-react";
 import { useTranslation } from "../i18n";
-import { formatAmount, localizeUsageError, planLabel } from "../lib/usage-display";
+import {
+  BOOSTER_WALLET_PLAN,
+  formatAmount,
+  localizeUsageError,
+  planLabel,
+} from "../lib/usage-display";
+import { isAlertTier } from "../hooks/useUsageQuery";
 import type { UsageData, UsageQueryState } from "../hooks/useUsageQuery";

 interface UsageFooterProps {
@@ -14,6 +20,20 @@ interface UsageFooterProps {
  variant?: "detail" | "compact";
 }

+/**
+ * 金额行 vs 配额行：NewAPI 与 Kimi 加力钱包（booster_wallet）只报金额，其
+ * used/total 是钱数而不是配额比例，必须走 formatAmount 渲染路径。判定只能看
+ * planName —— sub2api 的 USD 行 unit 同样不是 "%"，但语义仍是「额度用了百分之
+ * 多少」，按 unit 判定会把它误渲染成余额。
+ */
+function isMoneyRow(d: UsageData): boolean {
+  return d.planName === "NewAPI" || d.planName === BOOSTER_WALLET_PLAN;
+}
+
+function isPlanRow(d: UsageData): boolean {
+  return !!d.planName && !isMoneyRow(d) && (d.total != null || d.used != null);
+}
+
 export function UsageFooter({ usage, variant = "detail" }: UsageFooterProps) {
  const { t } = useTranslation();
  const { status, data, error, updatedAt, supported, refresh } = usage;
@@ -85,10 +105,7 @@ export function UsageFooter({ usage, variant = "detail" }: UsageFooterProps) {
    // A "plan" is a quota tier (used/total, e.g. 5h / weekly limit). Balance
    // entries carry only remaining + unit — their plan_name is a currency /
    // brand label and must render as an amount, not a percentage.
-    const isPlan =
-      !!d.planName &&
-      d.planName !== "NewAPI" &&
-      (d.total != null || d.used != null);
+    const isPlan = isPlanRow(d);
    if (d.isValid === false) {
      return { icon: "⚡", text: t("usageInvalidKey"), color: "text-red-500 dark:text-red-400" };
    }
@@ -207,13 +224,14 @@ export function UsageFooter({ usage, variant = "detail" }: UsageFooterProps) {
      {data.map((d, i) => {
        // Same plan-vs-balance rule as compactSummary: balance rows have no
        // quota fields (total/used), so they render as an amount.
-        const isPlan =
-          !!d.planName &&
-          d.planName !== "NewAPI" &&
-          (d.total != null || d.used != null);
+        const isPlan = isPlanRow(d);
        const pct = isPlan ? percentOf(d) : null;
-        const color =
-          pct == null
+        // 预警高亮：usage.alert 已按「查询成功」门控，失败时保留的上次数据
+        // 不会被误标红（此处是唯一的高亮来源，随每次查询结果重算）。
+        const rowAlert = usage.alert && isAlertTier(d, usage.threshold);
+        const color = rowAlert
+          ? "text-red-500 dark:text-red-400"
+          : pct == null
            ? "text-content-muted"
            : pct < 70
              ? "text-green-600 dark:text-green-400"
@@ -248,6 +266,15 @@ export function UsageFooter({ usage, variant = "detail" }: UsageFooterProps) {
            >
              {main}
            </span>
+            {rowAlert && (
+              <span
+                className="shrink-0 text-red-500 dark:text-red-400"
+                title={t("usageThresholdAlert")}
+                aria-label={t("usageThresholdAlert")}
+              >
+                ⚠
+              </span>
+            )}
            {resetText && (
              <span className="text-content-muted shrink-0">{resetText}</span>
            )}
@@ -1,8 +1,8 @@
-import { useMemo, useState } from "react";
+import { useEffect, useMemo, useState } from "react";
 import { useSessions } from "../../hooks/useSessions";
 import { fmtInt, fmtTime } from "../../lib/dashboard-format";
 import { useTranslation } from "../../i18n";
-import type { SessionRow, WorkspaceRow } from "../../types/sessions";
+import type { BulkArchiveResult, SessionRow, WorkspaceRow } from "../../types/sessions";

 function fmtBytes(n: number): string {
  const v = Number(n) || 0;
@@ -23,6 +23,44 @@ function rowKey(row: SessionRow): string {
  return `${row.workspaceId}/${row.id}`;
 }

+type ArchiveAllMode = "month" | "half" | "week" | "custom";
+
+type ArchiveAllState = {
+  mode: ArchiveAllMode;
+  /** YYYY-MM-DD, used when mode is "custom". */
+  date: string;
+  busy: boolean;
+  result: BulkArchiveResult | null;
+  error: string | null;
+};
+
+function localDateInput(d: Date): string {
+  const m = String(d.getMonth() + 1).padStart(2, "0");
+  const day = String(d.getDate()).padStart(2, "0");
+  return `${d.getFullYear()}-${m}-${day}`;
+}
+
+/** Today minus one month — the default value of the custom-date picker. */
+function defaultCustomDate(): string {
+  const d = new Date();
+  d.setMonth(d.getMonth() - 1);
+  return localDateInput(d);
+}
+
+/** Cutoff epoch ms for the selected option (custom dates start at local midnight). */
+function cutoffFor(state: ArchiveAllState): number {
+  const now = new Date();
+  if (state.mode === "month" || state.mode === "half" || state.mode === "week") {
+    const d = new Date(now);
+    if (state.mode === "month") d.setMonth(d.getMonth() - 1);
+    if (state.mode === "half") d.setDate(d.getDate() - 15);
+    if (state.mode === "week") d.setDate(d.getDate() - 7);
+    return d.getTime();
+  }
+  const [y, m, day] = state.date.split("-").map(Number);
+  return new Date(y || now.getFullYear(), (m || 1) - 1, day || 1, 0, 0, 0, 0).getTime();
+}
+
 type ConfirmState =
  | { mode: "one"; row: SessionRow }
  | { mode: "bulk"; rows: SessionRow[] }
@@ -52,6 +90,7 @@ export function SessionsPage() {
    refresh,
    archiveSession,
    unarchiveSession,
+    archiveBefore,
    deleteSession,
    deleteWorkspace,
    getPreview,
@@ -62,6 +101,7 @@ export function SessionsPage() {
  const [confirm, setConfirm] = useState<ConfirmState>(null);
  const [confirmBusy, setConfirmBusy] = useState(false);
  const [preview, setPreview] = useState<PreviewState>(null);
+  const [archiveAll, setArchiveAll] = useState<ArchiveAllState | null>(null);

  const workspaces = data?.workspaces ?? [];
  const sessions = data?.sessions ?? [];
@@ -199,6 +239,30 @@ export function SessionsPage() {
    0
  );

+  useEffect(() => {
+    if (!archiveAll) return;
+    const onKey = (e: KeyboardEvent) => {
+      if (e.key === "Escape") setArchiveAll((s) => (s && s.busy ? s : null));
+    };
+    window.addEventListener("keydown", onKey);
+    return () => window.removeEventListener("keydown", onKey);
+  }, [archiveAll]);
+
+  const runArchiveAll = async () => {
+    if (!archiveAll || archiveAll.busy) return;
+    const cutoffMs = cutoffFor(archiveAll);
+    setArchiveAll((s) => (s ? { ...s, busy: true, result: null, error: null } : s));
+    try {
+      const res = await archiveBefore(cutoffMs);
+      setArchiveAll((s) => (s ? { ...s, busy: false, result: res } : s));
+      setSelected(new Set());
+    } catch (e) {
+      setArchiveAll((s) =>
+        s ? { ...s, busy: false, error: e instanceof Error ? e.message : String(e) } : s
+      );
+    }
+  };
+
  return (
    <div className="flex h-full flex-col overflow-hidden p-4 space-y-3">
      <div className="flex shrink-0 flex-col gap-3 sm:flex-row sm:items-end sm:justify-between">
@@ -376,6 +440,23 @@ export function SessionsPage() {
                  </button>
                </>
              )}
+              {status !== "archived" && (
+                <button
+                  type="button"
+                  onClick={() =>
+                    setArchiveAll({
+                      mode: "month",
+                      date: defaultCustomDate(),
+                      busy: false,
+                      result: null,
+                      error: null,
+                    })
+                  }
+                  className="px-2 py-1 text-xs rounded border border-border hover:bg-hover-2"
+                >
+                  {t("sessionsArchiveAll")}
+                </button>
+              )}
              <select
                value={workspace}
                onChange={(e) => setWorkspace(e.target.value)}
@@ -548,6 +629,112 @@ export function SessionsPage() {
        {data?.home && <>{data.home}</>}
      </div>

+      {/* Bulk archive dialog */}
+      {archiveAll && (
+        <div className="fixed inset-0 z-50 flex items-center justify-center bg-black/60 p-4">
+          <div className="w-full max-w-md rounded-xl border border-border bg-panel shadow-xl">
+            <div className="space-y-3 p-5">
+              <h3 className="text-base font-semibold text-content-primary">
+                {t("archiveAllTitle")}
+              </h3>
+              <p className="text-sm text-content-muted">{t("archiveAllDesc")}</p>
+              <div className="space-y-0.5">
+                {(
+                  [
+                    ["month", t("archiveOptionMonth")],
+                    ["half", t("archiveOptionHalfMonth")],
+                    ["week", t("archiveOptionWeek")],
+                    ["custom", t("archiveOptionCustom")],
+                  ] as const
+                ).map(([mode, label]) => (
+                  <label
+                    key={mode}
+                    className="flex items-center gap-2 rounded-md px-2 py-1.5 text-sm text-content-primary hover:bg-input"
+                  >
+                    <input
+                      type="radio"
+                      name="archiveAllMode"
+                      className="h-4 w-4 accent-blue-600"
+                      checked={archiveAll.mode === mode}
+                      disabled={archiveAll.busy}
+                      onChange={() =>
+                        setArchiveAll((s) =>
+                          s ? { ...s, mode, result: null, error: null } : s
+                        )
+                      }
+                    />
+                    <span>{label}</span>
+                  </label>
+                ))}
+              </div>
+              {archiveAll.mode === "custom" && (
+                <input
+                  type="date"
+                  value={archiveAll.date}
+                  disabled={archiveAll.busy}
+                  onChange={(e) =>
+                    setArchiveAll((s) => (s ? { ...s, date: e.target.value } : s))
+                  }
+                  className="bg-input border border-border rounded px-2 py-1 text-sm focus:ring-2 focus:ring-blue-500 focus:outline-none"
+                />
+              )}
+              <div className="rounded-md border border-border bg-input px-3 py-2.5 text-sm text-content-muted">
+                {t("archiveAllConfirm", {
+                  date: new Date(cutoffFor(archiveAll)).toLocaleDateString(
+                    locale === "zh" ? "zh-CN" : "en-US"
+                  ),
+                })}
+              </div>
+              {archiveAll.error && (
+                <div className="rounded-md border border-red-500/40 bg-red-900/20 px-3 py-2 text-sm text-red-300">
+                  {t("archiveAllFailed")}: {archiveAll.error}
+                </div>
+              )}
+              {archiveAll.result && archiveAll.result.archived === 0 && (
+                <div className="rounded-md border border-border bg-input px-3 py-2 text-sm text-content-muted">
+                  {t("archiveAllEmpty")}
+                </div>
+              )}
+              {archiveAll.result && archiveAll.result.archived > 0 && (
+                <div className="rounded-md border border-emerald-500/40 bg-emerald-900/20 px-3 py-2 text-sm text-emerald-300">
+                  {t("archiveAllDone", {
+                    archived: archiveAll.result.archived,
+                    skipped: archiveAll.result.skipped,
+                  })}
+                </div>
+              )}
+              {archiveAll.result && archiveAll.result.errors.length > 0 && (
+                <div className="space-y-1 rounded-md border border-red-500/40 bg-red-900/20 px-3 py-2 text-xs text-red-300">
+                  {archiveAll.result.errors.slice(0, 3).map((err, i) => (
+                    <div key={i} className="break-all">
+                      {err}
+                    </div>
+                  ))}
+                </div>
+              )}
+            </div>
+            <div className="flex justify-end gap-2 border-t border-border bg-footer px-5 py-3">
+              <button
+                type="button"
+                disabled={archiveAll.busy}
+                onClick={() => setArchiveAll(null)}
+                className="px-3 py-1.5 text-sm rounded border border-border hover:bg-hover-2 disabled:opacity-50"
+              >
+                {archiveAll.result ? t("close") : t("cancel")}
+              </button>
+              <button
+                type="button"
+                disabled={archiveAll.busy}
+                onClick={runArchiveAll}
+                className="px-3 py-1.5 text-sm rounded bg-blue-600 text-white hover:bg-blue-500 disabled:opacity-50"
+              >
+                {archiveAll.busy ? t("archiveAllRunning") : t("sessionsArchiveAll")}
+              </button>
+            </div>
+          </div>
+        </div>
+      )}
+
      {/* Confirm dialog */}
      {confirm && (
        <div className="fixed inset-0 z-50 flex items-center justify-center bg-black/60 p-4">
@@ -0,0 +1,125 @@
+import { describe, expect, it } from "vitest";
+import {
+  localizeHealthError,
+  probeLine,
+  usageLine,
+  healthDotClass,
+  type CheckResult,
+} from "./useHealthCheck";
+
+/**
+ * 只回显 key，便于断言选中的是哪条文案（真实插值由 i18n Provider 负责）。
+ */
+const t = ((key: string) => `<${key}>`) as unknown as Parameters<
+  typeof localizeHealthError
+>[1];
+
+/** 带插值的假 t：`<key|{a}=1>`，用于断言参数透传。 */
+const tWithVars = ((key: string, vars?: Record<string, string | number>) =>
+  vars
+    ? `<${key}|${Object.entries(vars)
+        .map(([k, v]) => `${k}=${v}`)
+        .join(",")}>`
+    : `<${key}>`) as unknown as Parameters<typeof localizeHealthError>[1];
+
+describe("localizeHealthError", () => {
+  it("探活错误按前缀映射到对应文案", () => {
+    expect(localizeHealthError("missing base_url", t)).toBe(
+      "<healthErrNoBaseUrl>"
+    );
+    expect(localizeHealthError("no API key configured", t)).toBe(
+      "<healthErrNoKey>"
+    );
+    expect(localizeHealthError("Authentication failed (HTTP 401)", t)).toBe(
+      "<healthErrAuth>"
+    );
+    expect(localizeHealthError("endpoint not found (HTTP 404)", t)).toBe(
+      "<healthErrNotFound>"
+    );
+    expect(localizeHealthError("request timed out", t)).toBe(
+      "<healthErrTimeout>"
+    );
+    expect(
+      localizeHealthError("connection failed (DNS or refused)", t)
+    ).toBe("<healthErrUnreachable>");
+    expect(localizeHealthError("response is not valid JSON", t)).toBe(
+      "<healthErrBadJson>"
+    );
+  });
+
+  it("未知错误原样透出（HTTP 状态码等诊断细节）", () => {
+    expect(localizeHealthError("API error (HTTP 502)", t)).toBe(
+      "API error (HTTP 502)"
+    );
+  });
+
+  it("探活 + 账单两段拼接时逐段映射", () => {
+    expect(
+      localizeHealthError(
+        "endpoint not found (HTTP 404); usage query disabled in config panel",
+        t
+      )
+    ).toBe("<healthErrNotFound>; <usageErrDisabled>");
+  });
+
+  it("账单链路的报错复用用量错误映射", () => {
+    // commands.rs 的 kinds 循环会把单条错误包成 `{kind}: {msg}`。
+    expect(
+      localizeHealthError(
+        "Authentication failed (HTTP 401); balance:deepseek: Authentication failed (HTTP 401)",
+        t
+      )
+    ).toBe("<healthErrAuth>; <usageInvalidKey>");
+  });
+});
+
+describe("probeLine", () => {
+  it("通过时带延迟", () => {
+    expect(probeLine({ state: "pass" }, 123, tWithVars)).toBe(
+      "<healthProbeOk|latency=123>"
+    );
+  });
+
+  it("中性（404）说明端点不支持模型列表，而不是报失败", () => {
+    expect(
+      probeLine(
+        { state: "neutral", error: "endpoint not found (HTTP 404)" },
+        0,
+        tWithVars
+      )
+    ).toBe("<healthProbeUnsupported>");
+  });
+
+  it("失败时透出本地化后的原因", () => {
+    expect(
+      probeLine({ state: "fail", error: "request timed out" }, 0, tWithVars)
+    ).toBe("<healthProbeFail|error=<healthErrTimeout>>");
+  });
+});
+
+describe("usageLine", () => {
+  it("通过 / 失败 / 兜底三态", () => {
+    const pass: CheckResult = { state: "pass" };
+    expect(usageLine(pass, tWithVars)).toBe("<healthUsageOk>");
+
+    expect(
+      usageLine(
+        { state: "fail", error: "Authentication failed (HTTP 401)" },
+        tWithVars
+      )
+    ).toBe("<healthUsageFail|error=<healthErrAuth>>");
+
+    // 账单链路不会产生中性结论，兜底按「未判定」展示
+    expect(usageLine({ state: "neutral" }, tWithVars)).toBe(
+      "<healthNotChecked>"
+    );
+  });
+});
+
+describe("healthDotClass", () => {
+  it("绿=通过、红=失败、灰=未判定（中性）", () => {
+    expect(healthDotClass("pass")).toBe("bg-green-500");
+    expect(healthDotClass("fail")).toBe("bg-red-500");
+    expect(healthDotClass("neutral")).toBe("bg-gray-400 dark:bg-gray-500");
+  });
+});
@@ -0,0 +1,175 @@
+import { useCallback, useEffect, useRef, useState } from "react";
+import { invoke } from "@tauri-apps/api/core";
+import type { Agent } from "../types";
+import type { TranslationKey } from "../i18n/zh";
+import { localizeUsageError } from "../lib/usage-display";
+
+type TranslateFn = (
+  key: TranslationKey,
+  vars?: Record<string, string | number>
+) => string;
+
+/** 单项检测的三态结论（mirrors Rust `HealthCheck`）。 */
+export type HealthCheck = "pass" | "fail" | "neutral";
+
+/** 单项检测结论（探活 / 账单各一份，mirrors Rust `CheckResult`）。 */
+export interface CheckResult {
+  state: HealthCheck;
+  /** 失败 / 中性原因的原始文案；通过为 null。 */
+  error?: string | null;
+}
+
+/** Mirrors the Rust `HealthResult` (camelCase via serde). */
+export interface HealthResult {
+  providerName: string;
+  /** 总体判定（Rust 侧是唯一事实来源）：pass=绿 / fail=红 / neutral=灰。 */
+  verdict: HealthCheck;
+  /** verdict === "pass" 的便捷布尔。 */
+  ok: boolean;
+  /** 失败原因汇总（仅 fail 项，`; ` 分隔）；无失败为 null。 */
+  error?: string | null;
+  /** 账单/用量结论；null/undefined = 未执行。 */
+  usage?: CheckResult | null;
+  /** 推理端点探活结论；探活是必做项，恒为 Some。 */
+  probe?: CheckResult | null;
+  /** 探活耗时（毫秒）。 */
+  latencyMs: number;
+}
+
+/**
+ * 单条 Rust 错误文案 → 本地化文案。未知错误原样透出（HTTP 状态码等诊断
+ * 细节）。前缀与 src-tauri/src/services/probe.rs、commands.rs 保持一致。
+ */
+function localizeHealthPart(error: string, t: TranslateFn): string {
+  if (error.startsWith("missing base_url")) return t("healthErrNoBaseUrl");
+  if (error.startsWith("no API key configured")) return t("healthErrNoKey");
+  if (error.startsWith("Authentication failed")) return t("healthErrAuth");
+  if (error.startsWith("endpoint not found")) return t("healthErrNotFound");
+  if (error.startsWith("request timed out")) return t("healthErrTimeout");
+  if (error.startsWith("connection failed")) return t("healthErrUnreachable");
+  if (error.startsWith("response is not valid JSON")) return t("healthErrBadJson");
+  // 账单链路的报错复用已有的用量错误映射（禁用 / OAuth 过期 / NewAPI 凭据…）。
+  return localizeUsageError(error, t);
+}
+
+/**
+ * 体检结论里的错误文案本地化。探活与账单两条链路的错误由 Rust 用 `; `
+ * 拼进同一个字段，逐段映射后再拼回，保证两段都能翻译。
+ */
+export function localizeHealthError(error: string, t: TranslateFn): string {
+  return error
+    .split("; ")
+    .map((part) => localizeHealthPart(part, t))
+    .join("; ");
+}
+
+/** 探活一项的 tooltip 文案。 */
+export function probeLine(
+  probe: CheckResult,
+  latencyMs: number,
+  t: TranslateFn
+): string {
+  if (probe.state === "pass") return t("healthProbeOk", { latency: latencyMs });
+  // 中性 = 端点未实现 /v1/models（404），如实说明「不支持」而不是报失败。
+  if (probe.state === "neutral") return t("healthProbeUnsupported");
+  return t("healthProbeFail", {
+    error: probe.error ? localizeHealthError(probe.error, t) : "",
+  });
+}
+
+/** 账单一项的 tooltip 文案。 */
+export function usageLine(usage: CheckResult, t: TranslateFn): string {
+  if (usage.state === "pass") return t("healthUsageOk");
+  if (usage.state === "fail") {
+    return t("healthUsageFail", {
+      error: usage.error ? localizeHealthError(usage.error, t) : "",
+    });
+  }
+  // 账单链路不产生中性结论（只有探活 404 会），兜底按「未判定」展示。
+  return t("healthNotChecked");
+}
+
+/**
+ * 状态点颜色：绿=正常、红=失败、灰=未判定（未检测 / 检测中 / 全部中性）。
+ * 判定本身由 Rust 的 overall_verdict 给出，这里只做映射。
+ */
+export function healthDotClass(verdict: HealthCheck): string {
+  switch (verdict) {
+    case "pass":
+      return "bg-green-500";
+    case "fail":
+      return "bg-red-500";
+    default:
+      return "bg-gray-400 dark:bg-gray-500";
+  }
+}
+
+export interface HealthCheckState {
+  /** 体检进行中：按钮转圈、尚未出结果的行显示灰色检测中指示。 */
+  running: boolean;
+  /** 至少跑过一轮（决定是否渲染状态点）。 */
+  ran: boolean;
+  /** provider name → 结果。 */
+  results: Record<string, HealthResult>;
+  okCount: number;
+  failCount: number;
+  /** 整体请求失败（配置读取失败 / 总超时）的文案；null = 无错误。 */
+  error: string | null;
+  run: () => void;
+}
+
+/**
+ * 一键体检：invoke `health_check_all`，把结果按 provider 名索引。
+ * 结果保留到下次点击或组件卸载（页面刷新），不在后台轮询。
+ */
+export function useHealthCheck(agent: Agent): HealthCheckState {
+  const [running, setRunning] = useState(false);
+  const [ran, setRan] = useState(false);
+  const [results, setResults] = useState<Record<string, HealthResult>>({});
+  const [error, setError] = useState<string | null>(null);
+  // Generation counter: 重跑/卸载后到达的旧响应直接丢弃，不写回状态。
+  const genRef = useRef(0);
+
+  useEffect(
+    () => () => {
+      genRef.current += 1;
+    },
+    []
+  );
+
+  const run = useCallback(async () => {
+    const gen = ++genRef.current;
+    setRunning(true);
+    setError(null);
+    // 清空上一轮结果：检测中的行显示灰色转圈，而不是留着旧结论误导。
+    setResults({});
+    try {
+      const list = await invoke<HealthResult[]>("health_check_all", { agent });
+      if (genRef.current !== gen) return;
+      const map: Record<string, HealthResult> = {};
+      for (const r of list) map[r.providerName] = r;
+      setResults(map);
+    } catch (e) {
+      if (genRef.current !== gen) return;
+      // invoke 拒绝时 Rust 侧 Err(String) 是字符串，不是 Error 实例。
+      setError(typeof e === "string" ? e : e instanceof Error ? e.message : String(e));
+    } finally {
+      if (genRef.current === gen) {
+        setRunning(false);
+        setRan(true);
+      }
+    }
+  }, [agent]);
+
+  const values = Object.values(results);
+  return {
+    running,
+    ran,
+    results,
+    // 中性（灰）既不算正常也不算失败，不计入汇总。
+    okCount: values.filter((r) => r.verdict === "pass").length,
+    failCount: values.filter((r) => r.verdict === "fail").length,
+    error,
+    run: () => void run(),
+  };
+}
@@ -0,0 +1,28 @@
+import { describe, expect, it } from "vitest";
+import { isNewerVersion } from "./useKimiCodeVersion";
+
+describe("isNewerVersion", () => {
+  it("同版本不算有更新", () => {
+    expect(isNewerVersion("2.0.2", "2.0.2")).toBe(false);
+    expect(isNewerVersion("v2.0.2", "2.0.2")).toBe(false);
+  });
+
+  it("按 major/minor/patch 逐段比较", () => {
+    expect(isNewerVersion("2.0.2", "2.0.1")).toBe(true);
+    expect(isNewerVersion("2.1.0", "2.0.9")).toBe(true);
+    expect(isNewerVersion("3.0.0", "2.99.99")).toBe(true);
+    expect(isNewerVersion("2.0.1", "2.0.2")).toBe(false);
+    expect(isNewerVersion("1.9.9", "2.0.0")).toBe(false);
+  });
+
+  it("忽略 v 前缀与预发布后缀", () => {
+    expect(isNewerVersion("v0.44.0-beta.1", "0.43.1")).toBe(true);
+    expect(isNewerVersion("0.43.1", "0.43.1-rc.2")).toBe(false);
+  });
+
+  it("缺位补 0，无法解析时不显示更新", () => {
+    expect(isNewerVersion("2.1", "2.0.9")).toBe(true);
+    expect(isNewerVersion("abc", "2.0.1")).toBe(false);
+    expect(isNewerVersion("2.0.2", "abc")).toBe(false);
+  });
+});
@@ -0,0 +1,101 @@
+import { useCallback, useEffect, useRef, useState } from "react";
+import { invoke } from "@tauri-apps/api/core";
+
+/** kimi-code 与本应用的适配状态（Rust 侧基线表决定）。 */
+export type KimiCodeStatus =
+  | "current"
+  | "outdated"
+  | "partial"
+  | "unknown"
+  | "not_installed";
+
+export interface KimiCodeVersionInfo {
+  /** 本机安装的 kimi-code 版本；未安装 / 检测失败为 null。 */
+  installed: string | null;
+  /** 上游最新 release tag（已去掉 v 前缀）；查询失败为 null。 */
+  latest: string | null;
+  status: KimiCodeStatus;
+}
+
+const THROTTLE_KEY = "kimi-switch-last-kc-version-check";
+// 与 useUpdateCheck 一致：节流窗口与定时复查同为 8h，长时间挂着的窗口也能
+// 自动拿到新版本，同时避免短时间反复触发后端调用。
+const CHECK_INTERVAL_MS = 8 * 60 * 60 * 1000; // 8 hours
+
+/** 解析 major.minor.patch 三元组；缺位补 0，无法解析返回 null。 */
+function versionTriple(version: string): [number, number, number] | null {
+  const core = version.trim().replace(/^[vV]/, "").split(/[-+]/)[0];
+  const nums = core.split(".").slice(0, 3).map((p) => Number.parseInt(p.trim(), 10));
+  if (nums.length === 0 || nums.some((n) => !Number.isFinite(n) || n < 0)) return null;
+  while (nums.length < 3) nums.push(0);
+  return [nums[0], nums[1], nums[2]];
+}
+
+/** latest 是否严格高于 installed（用于决定是否显示 releases 链接）。 */
+export function isNewerVersion(latest: string, installed: string): boolean {
+  const l = versionTriple(latest);
+  const i = versionTriple(installed);
+  if (!l || !i) return false;
+  for (let k = 0; k < 3; k += 1) {
+    if (l[k] !== i[k]) return l[k] > i[k];
+  }
+  return false;
+}
+
+/**
+ * 读取「本机 kimi-code 版本 / 上游最新版 / 适配状态」，模式与 useUpdateCheck
+ * 相同：挂载时按 8h 节流自动查一次，之后定时复查，checkNow 强制刷新。
+ */
+export function useKimiCodeVersion() {
+  const [info, setInfo] = useState<KimiCodeVersionInfo | null>(null);
+  const [checking, setChecking] = useState(false);
+  const [lastChecked, setLastChecked] = useState<number | null>(() => {
+    try {
+      const v = localStorage.getItem(THROTTLE_KEY);
+      return v ? Number(v) : null;
+    } catch {
+      return null;
+    }
+  });
+
+  const doCheck = useCallback(
+    async (force: boolean) => {
+      if (!force) {
+        if (lastChecked && Date.now() - lastChecked < CHECK_INTERVAL_MS) return;
+      }
+      setChecking(true);
+      try {
+        const next = await invoke<KimiCodeVersionInfo>("get_kimi_code_version_info");
+        setInfo(next);
+        const now = Date.now();
+        setLastChecked(now);
+        try {
+          localStorage.setItem(THROTTLE_KEY, String(now));
+        } catch {
+          // ignore
+        }
+      } catch {
+        // 静默失败：检测不到 kimi / 网络不通都不该打扰用户，手动检查仍可用。
+      } finally {
+        setChecking(false);
+      }
+    },
+    [lastChecked]
+  );
+
+  // 定时器不随 lastChecked 重建，所以用 ref 持有最新的 doCheck。
+  const doCheckRef = useRef(doCheck);
+  useEffect(() => {
+    doCheckRef.current = doCheck;
+  }, [doCheck]);
+
+  useEffect(() => {
+    doCheckRef.current(false);
+    const interval = setInterval(() => doCheckRef.current(false), CHECK_INTERVAL_MS);
+    return () => clearInterval(interval);
+  }, []);
+
+  const checkNow = useCallback(() => doCheck(true), [doCheck]);
+
+  return { info, checking, checkNow, lastChecked };
+}
@@ -2,6 +2,7 @@ import { useCallback, useEffect, useState } from "react";
 import { invoke } from "@tauri-apps/api/core";
 import type {
  ActionResponse,
+  BulkArchiveResult,
  PreviewResult,
  SessionStatusFilter,
  SessionsResult,
@@ -54,6 +55,17 @@ export function useSessions() {
    [refresh]
  );

+  const archiveBefore = useCallback(
+    async (cutoffMs: number) => {
+      const result = await invoke<BulkArchiveResult>("archive_sessions_before", {
+        cutoffMs,
+      });
+      await refresh();
+      return result;
+    },
+    [refresh]
+  );
+
  const deleteSession = useCallback(
    async (workspaceId: string, sessionId: string, sessionStatus?: string) => {
      await invoke<ActionResponse>("delete_session", {
@@ -100,6 +112,7 @@ export function useSessions() {
    refresh,
    archiveSession,
    unarchiveSession,
+    archiveBefore,
    deleteSession,
    deleteWorkspace,
    getPreview,
@@ -0,0 +1,63 @@
+import { describe, expect, it } from "vitest";
+import { isAlertTier, usageQuerySupported, type UsageData } from "./useUsageQuery";
+
+/** 百分比窗口行（five_hour 等）；套餐类 tier 的 unit 恒为 "%"。 */
+function percentTier(used: number): UsageData {
+  return { planName: "five_hour", used, total: 100, remaining: 100 - used, unit: "%" };
+}
+
+describe("isAlertTier", () => {
+  it("阈值缺省或 0（关闭）时不告警", () => {
+    expect(isAlertTier(percentTier(100), undefined)).toBe(false);
+    expect(isAlertTier(percentTier(100), null)).toBe(false);
+    expect(isAlertTier(percentTier(100), 0)).toBe(false);
+  });
+
+  it("低于阈值不告警", () => {
+    expect(isAlertTier(percentTier(79), 80)).toBe(false);
+    expect(isAlertTier(percentTier(79.9), 80)).toBe(false);
+  });
+
+  it("等于或超过阈值告警", () => {
+    expect(isAlertTier(percentTier(80), 80)).toBe(true);
+    expect(isAlertTier(percentTier(80.5), 80)).toBe(true);
+    expect(isAlertTier(percentTier(100), 80)).toBe(true);
+  });
+
+  it("非 % unit 的行不告警（金额行的 used 是钱数）", () => {
+    // Kimi 加力钱包：used = monthlyUsedCents/100，与百分比阈值不可比。
+    expect(
+      isAlertTier(
+        { planName: "booster_wallet", used: 8, total: 50, remaining: 12.34, unit: "USD" },
+        5
+      )
+    ).toBe(false);
+    // sub2api 的 USD 配额行同理，不参与百分比预警。
+    expect(
+      isAlertTier({ planName: "daily_limit", used: 90, total: 100, unit: "USD" }, 50)
+    ).toBe(false);
+  });
+
+  it("used 缺失时不告警", () => {
+    expect(isAlertTier({ planName: "five_hour", unit: "%", remaining: 20 }, 50)).toBe(false);
+  });
+});
+
+describe("usageQuerySupported", () => {
+  it("有识别类型即可查询", () => {
+    expect(usageQuerySupported(["balance:deepseek"])).toBe(true);
+    expect(usageQuerySupported([], true)).toBe(true);
+  });
+
+  it("显式模板（newapi/sub2api）绕过 usage_kinds——自定义供应商也该查", () => {
+    expect(usageQuerySupported(undefined, true)).toBe(true);
+  });
+
+  it("无识别类型且非显式模板时不查询", () => {
+    expect(usageQuerySupported(undefined)).toBe(false);
+    expect(usageQuerySupported([])).toBe(false);
+    // auto 模板没有识别到类型 = 不支持（与弹窗短路口径一致）
+    expect(usageQuerySupported([], false)).toBe(false);
+    expect(usageQuerySupported(undefined, undefined)).toBe(false);
+  });
+});
@@ -56,6 +56,34 @@ export interface UsageQueryState {
  updatedAt: number | null;
  refresh: (force: boolean) => void;
  supported: boolean;
+  /** 生效的预警阈值（0/缺失已归一为 undefined）；footer 用它判定单行高亮。 */
+  threshold?: number;
+  /** 查询成功且任一百分比行 used ≥ threshold。失败/加载中恒为 false， */
+  /** 所以失败时保留的上次数据不会被误高亮。 */
+  alert: boolean;
+}
+
+/**
+ * 单行是否触发预警。只判定百分比行（unit === "%"）：余额/钱包等金额行的
+ * used 是钱数，与百分比阈值不可比。threshold 缺失或 ≤0（关闭）时恒为 false。
+ */
+export function isAlertTier(
+  d: UsageData,
+  threshold?: number | null
+): boolean {
+  if (threshold == null || threshold <= 0) return false;
+  return d.unit === "%" && d.used != null && d.used >= threshold;
+}
+
+/**
+ * 是否值得为该供应商发起用量查询：有预设识别类型，或后端按显式模板
+ * （newapi / sub2api，二者绕过 usage_kinds）查询。
+ */
+export function usageQuerySupported(
+  usageKinds?: string[],
+  explicitTemplate?: boolean
+): boolean {
+  return (usageKinds?.length ?? 0) > 0 || explicitTemplate === true;
 }

 export function useUsageQuery(
@@ -63,9 +91,14 @@ export function useUsageQuery(
  providerName: string,
  usageKinds?: string[],
  autoIntervalMinutes?: number,
-  disabled?: boolean
+  disabled?: boolean,
+  threshold?: number,
+  /** Backend resolves the query from an explicit template (newapi / sub2api)
+   * regardless of usageKinds — those bypass kinds entirely, so the row is
+   * supported even when the provider has no preset-detected kinds. */
+  explicitTemplate?: boolean
 ): UsageQueryState {
-  const supported = (usageKinds?.length ?? 0) > 0;
+  const supported = usageQuerySupported(usageKinds, explicitTemplate);
  // Cache key includes the agent so a Kimi Code provider and a Pi provider
  // with the same name do not clobber each other's cached result.
  const cacheKey = `${agent}:${providerName}`;
@@ -175,6 +208,15 @@ export function useUsageQuery(
    return () => clearInterval(id);
  }, [supported, autoIntervalMinutes, runQuery, disabled]);

+  // 阈值 0/负数/非有限值一律视为关闭（与配置面板「0 或空 = 关闭」一致）。
+  // 每次查询结果写入 data 后此处重新求值，因此刷新后预警自动出现/消除。
+  const alertThreshold =
+    typeof threshold === "number" && Number.isFinite(threshold) && threshold > 0
+      ? threshold
+      : undefined;
+  const alert =
+    status === "success" && data.some((d) => isAlertTier(d, alertThreshold));
+
  return {
    status,
    data,
@@ -182,5 +224,7 @@ export function useUsageQuery(
    updatedAt,
    refresh: runQuery,
    supported,
+    threshold: alertThreshold,
+    alert,
  };
 }
@@ -43,6 +43,26 @@ export const enTranslations: Record<TranslationKey, string> = {
  connectivityOk: "{name} reachable ({latency}ms)",
  connectivitySlow: "{name} reachable but slow ({latency}ms)",
  connectivityFail: "{name} unreachable: {error}",
+  // Health check (ProviderList toolbar button + per-row status dot)
+  healthCheck: "Health check",
+  healthCheckHint:
+    "Probe every enabled provider's inference endpoint and billing query in parallel",
+  healthCheckRunning: "Checking…",
+  healthSummary: "{ok} OK / {fail} failed",
+  healthProbeOk: "Endpoint OK ({latency}ms)",
+  healthProbeUnsupported: "Endpoint exposes no model list (404)",
+  healthProbeFail: "Endpoint failed: {error}",
+  healthUsageOk: "Billing OK",
+  healthUsageFail: "Billing failed: {error}",
+  healthNotChecked: "Not checked",
+  healthRequestFailed: "Health check request failed: {error}",
+  healthErrNoBaseUrl: "Missing base URL",
+  healthErrNoKey: "No API key configured",
+  healthErrAuth: "Authentication failed",
+  healthErrNotFound: "Endpoint not found",
+  healthErrTimeout: "Connection timed out",
+  healthErrUnreachable: "Cannot connect (DNS failure or connection refused)",
+  healthErrBadJson: "Response is not valid JSON",
  copiedProvider: "Duplicated as {name}",

  // Provider edit
@@ -78,6 +98,12 @@ export const enTranslations: Record<TranslationKey, string> = {
  authField: "Auth Field",
  apiKey: "API Key",
  getApiKeyLink: "Get API Key",
+  apiKeySource: "Credential source",
+  apiKeySourceDirect: "Enter key",
+  apiKeySourceEnv: "Environment variable",
+  apiKeyEnvName: "Variable name",
+  apiKeyEnvHint:
+    "Only the variable name (api_key_env) is written; Kimi Code reads the secret from its environment. The two sources are mutually exclusive, so switching clears the other.",
  show: "Show",
  hide: "Hide",
  requestUrl: "Request URL",
@@ -142,9 +168,16 @@ export const enTranslations: Record<TranslationKey, string> = {
  loopControlSettings: "Loop Control",
  maxAttemptsPerStep: "Max attempts per step",
  reservedContextSize: "Reserved context size",
+  compactionMaxAttempts: "Max compaction attempts",
+  compactionMaxAttemptsHint:
+    "Retry limit after a failed context compaction. Empty or 0 leaves the key out of the config (upstream default 5).",
  backgroundSettings: "Background Tasks",
  maxRunningTasks: "Max running tasks",
  keepAliveOnExit: "Keep alive on exit",
+  watchSettings: "File Watching",
+  watchEnabled: "Enable file watching",
+  watchEnabledDesc:
+    "Watch config files and the workspace for changes (configurable since kimi-code 2.0.1, off by default since 2.0.2). The KIMI_CODE_WATCH environment variable overrides this setting.",
  permissionRules: "Permission Rules",
  permissionDecision: "Decision",
  permissionPattern: "Pattern",
@@ -225,6 +258,19 @@ export const enTranslations: Record<TranslationKey, string> = {
  roleUser: "User",
  roleAssistant: "Assistant",
  roleSystem: "System",
+  sessionsArchiveAll: "Archive old sessions",
+  archiveAllTitle: "Archive old sessions",
+  archiveAllDesc:
+    "Archives every active session whose last activity is older than the selected date. Files are kept and usage stats stay intact.",
+  archiveOptionMonth: "Older than 1 month",
+  archiveOptionHalfMonth: "Older than 15 days",
+  archiveOptionWeek: "Older than 1 week",
+  archiveOptionCustom: "Specific date",
+  archiveAllConfirm: "Archive every active session last active before {date}. Continue?",
+  archiveAllRunning: "Archiving…",
+  archiveAllDone: "Archived {archived} sessions, skipped {skipped}",
+  archiveAllFailed: "Archive failed",
+  archiveAllEmpty: "No sessions match",

  // Heatmap
  noData: "No data",
@@ -270,6 +316,18 @@ export const enTranslations: Record<TranslationKey, string> = {
  install: "Install",
  updateCheckFailed: "Update check failed",
  lastChecked: "Last checked",
+  // Kimi Code version tracking
+  kimiCodeSection: "Kimi Code version",
+  kimiCodeInstalled: "Installed",
+  kimiCodeLatest: "Latest",
+  kimiCodeStatusLabel: "Compatibility",
+  kimiCodeStatusCurrent: "Supported",
+  kimiCodeStatusOutdated: "Update KimiSwitch recommended",
+  kimiCodeStatusPartial: "Partially compatible",
+  kimiCodeStatusUnknown: "Unknown version",
+  kimiCodeStatusNotInstalled: "Not installed",
+  kimiCodeCheck: "Check",
+  kimiCodeOpenReleases: "Open releases",
  appDescription: "Windows desktop LLM provider config manager",
  settingsWebsite: "Official website",
  kimiEyesSite: "Kimi Eyes plugin site",
@@ -436,6 +494,11 @@ export const enTranslations: Record<TranslationKey, string> = {
  usageTierDaily: "Daily",
  usageTierWeekly: "Weekly",
  usageTierMonthly: "Monthly",
+  usageTierMonthCode: "Monthly (code)",
+  usageTierBoosterWallet: "Booster",
+  usageThreshold: "Alert threshold",
+  usageThresholdHint: "Highlight any window that reaches this percentage; 0 = off",
+  usageThresholdAlert: "Threshold reached",
  usageErrNoKey: "No API key configured",
  usageErrDisabled: "Usage query is disabled in the config panel",
  usageErrLoginExpired: "Kimi Code login expired and auto-refresh failed; run `kimi login` again",
@@ -477,8 +540,6 @@ export const enTranslations: Record<TranslationKey, string> = {
  flagSubagentForkDesc: "Agent/AgentSwarm tools can start a subagent with a snapshot of the calling agent's history (fork parameter)",
  flagWaitFor: "WaitFor Tool",
  flagWaitForDesc: "The agent can wait for background tasks to finish within the current turn",
-  flagAutoSessionTitle: "Auto Session Title",
-  flagAutoSessionTitleDesc: "Generate session titles automatically via the managed chat_title tool",
  flagNotifyUser: "NotifyUser tool",
  flagNotifyUserDesc: "Experimental Updates panel: paginated progress messages from the main agent and subagents (off by default)",
  flagDefaultOn: "On by default",
@@ -533,6 +594,21 @@ export const enTranslations: Record<TranslationKey, string> = {
  openWebUIBrowser: "Open in Browser",
  webuiOpening: "Opening...",

+  // models.dev reference data (context / capabilities / pricing)
+  modelsDataSection: "Model Reference Data (models.dev)",
+  modelsDataDesc:
+    "Model context limits, capability flags, and pricing come from an online models.dev snapshot. Syncing applies new models immediately — no app update needed.",
+  modelsDataSourceSynced: "Online sync",
+  modelsDataSourceBuiltin: "Bundled with release",
+  modelsDataStats: "{date} · {models} models / {providers} providers",
+  modelsDataSyncNow: "Sync Now",
+  modelsDataSyncing: "Syncing...",
+  modelsDataSyncOk: "Synced — model parameters and pricing refreshed",
+  modelsDataSyncFailed: "Sync failed",
+  modelsDataRestore: "Restore Built-in Data",
+  modelsDataRestoreConfirm:
+    "Delete the local synced copy and fall back to the snapshot bundled with this release? You can sync again at any time.",
+
  // Plugin marketplace
  pluginMarketplace: "Plugin Marketplace",
  pluginMarketplaceSubtitle:
@@ -41,6 +41,25 @@ export const zhTranslations = {
  connectivityOk: "{name} 连通正常（{latency}ms）",
  connectivitySlow: "{name} 连通较慢（{latency}ms）",
  connectivityFail: "{name} 无法连通：{error}",
+  // 一键体检（ProviderList 工具栏按钮 + 行内状态点）
+  healthCheck: "一键体检",
+  healthCheckHint: "并行检测所有启用供应商的推理端点与账单查询",
+  healthCheckRunning: "体检中…",
+  healthSummary: "{ok} 正常 / {fail} 失败",
+  healthProbeOk: "端点正常（{latency}ms）",
+  healthProbeUnsupported: "端点不支持模型列表（404）",
+  healthProbeFail: "端点失败：{error}",
+  healthUsageOk: "账单正常",
+  healthUsageFail: "账单失败：{error}",
+  healthNotChecked: "未检测",
+  healthRequestFailed: "体检请求失败：{error}",
+  healthErrNoBaseUrl: "缺少 base_url",
+  healthErrNoKey: "未配置 API Key",
+  healthErrAuth: "鉴权失败",
+  healthErrNotFound: "端点不存在",
+  healthErrTimeout: "连接超时",
+  healthErrUnreachable: "无法连接（DNS 解析失败或连接被拒绝）",
+  healthErrBadJson: "响应不是合法 JSON",
  copiedProvider: "已复制为 {name}",

  // Provider edit
@@ -76,6 +95,12 @@ export const zhTranslations = {
  authField: "认证字段",
  apiKey: "API Key",
  getApiKeyLink: "获取 API Key",
+  apiKeySource: "凭证来源",
+  apiKeySourceDirect: "直接填写",
+  apiKeySourceEnv: "环境变量",
+  apiKeyEnvName: "环境变量名",
+  apiKeyEnvHint:
+    "只写入变量名（api_key_env），密钥由 Kimi Code 从运行环境读取；两者互斥，切换后会清空另一种。",
  show: "显示",
  hide: "隐藏",
  requestUrl: "请求地址",
@@ -140,9 +165,15 @@ export const zhTranslations = {
  loopControlSettings: "循环控制",
  maxAttemptsPerStep: "单步最大尝试次数",
  reservedContextSize: "上下文预留大小",
+  compactionMaxAttempts: "压缩最大尝试次数",
+  compactionMaxAttemptsHint: "上下文压缩失败后的重试上限，留空或 0 表示不写入该项（上游默认 5）。",
  backgroundSettings: "后台任务",
  maxRunningTasks: "最大并发数",
  keepAliveOnExit: "退出时保持运行",
+  watchSettings: "文件监听",
+  watchEnabled: "启用文件监听",
+  watchEnabledDesc:
+    "监听配置文件与工作目录变化（kimi-code 2.0.1 起可配置，2.0.2 起默认关闭）。环境变量 KIMI_CODE_WATCH 会覆盖本配置。",
  permissionRules: "权限规则",
  permissionDecision: "处置",
  permissionPattern: "模式",
@@ -220,6 +251,18 @@ export const zhTranslations = {
  roleUser: "用户",
  roleAssistant: "助手",
  roleSystem: "系统",
+  sessionsArchiveAll: "一键归档",
+  archiveAllTitle: "一键归档会话",
+  archiveAllDesc: "将归档最后活跃时间早于所选日期的全部活跃会话；文件保留，统计不受影响。",
+  archiveOptionMonth: "1 个月前",
+  archiveOptionHalfMonth: "半个月前",
+  archiveOptionWeek: "1 星期前",
+  archiveOptionCustom: "指定日期",
+  archiveAllConfirm: "将归档最后活跃时间早于 {date} 的全部活跃会话。确认？",
+  archiveAllRunning: "正在归档…",
+  archiveAllDone: "已归档 {archived} 个会话，跳过 {skipped} 个",
+  archiveAllFailed: "归档失败",
+  archiveAllEmpty: "没有符合条件的会话",

  // Heatmap
  noData: "暂无数据",
@@ -265,6 +308,18 @@ export const zhTranslations = {
  install: "安装",
  updateCheckFailed: "检查更新失败",
  lastChecked: "上次检查",
+  // Kimi Code 版本跟踪
+  kimiCodeSection: "Kimi Code 版本",
+  kimiCodeInstalled: "已安装版本",
+  kimiCodeLatest: "最新版本",
+  kimiCodeStatusLabel: "适配状态",
+  kimiCodeStatusCurrent: "已适配",
+  kimiCodeStatusOutdated: "建议升级 KimiSwitch",
+  kimiCodeStatusPartial: "部分兼容",
+  kimiCodeStatusUnknown: "未知版本",
+  kimiCodeStatusNotInstalled: "未安装",
+  kimiCodeCheck: "检查",
+  kimiCodeOpenReleases: "打开 Releases",
  appDescription: "Windows 桌面端 LLM 供应商配置管理器",
  settingsWebsite: "官方网站",
  kimiEyesSite: "Kimi Eyes 插件站",
@@ -431,6 +486,11 @@ export const zhTranslations = {
  usageTierDaily: "每日",
  usageTierWeekly: "7天",
  usageTierMonthly: "30天",
+  usageTierMonthCode: "30天（代码）",
+  usageTierBoosterWallet: "加力钱包",
+  usageThreshold: "预警阈值",
+  usageThresholdHint: "任一用量窗口达到该百分比时高亮提醒，0 为关闭",
+  usageThresholdAlert: "已达预警阈值",
  usageErrNoKey: "未配置 API Key",
  usageErrDisabled: "用量查询已在配置面板中停用",
  usageErrLoginExpired: "Kimi Code 登录已过期且自动续期失败，请重新运行 `kimi login`",
@@ -472,8 +532,6 @@ export const zhTranslations = {
  flagSubagentForkDesc: "Agent/AgentSwarm 工具可携带主代理会话快照启动子代理（fork 参数）",
  flagWaitFor: "WaitFor 工具",
  flagWaitForDesc: "代理可在当前回合内等待后台任务完成，而无需结束回合再被唤醒",
-  flagAutoSessionTitle: "自动会话标题",
-  flagAutoSessionTitleDesc: "通过托管 chat_title 工具自动生成会话标题",
  flagNotifyUser: "NotifyUser 工具",
  flagNotifyUserDesc: "实验性 Updates 面板：分页展示主代理与子代理的进度消息（默认关闭）",
  flagDefaultOn: "默认开启",
@@ -525,6 +583,21 @@ export const zhTranslations = {
  openWebUIBrowser: "在浏览器打开",
  webuiOpening: "打开中...",

+  // models.dev reference data (context / capabilities / pricing)
+  modelsDataSection: "模型参考数据（models.dev）",
+  modelsDataDesc:
+    "模型上下文长度、能力标记与单价参考来自 models.dev 在线快照。同步后新模型的参数与价格立即生效，无需等待新版本。",
+  modelsDataSourceSynced: "在线同步",
+  modelsDataSourceBuiltin: "随版本内置",
+  modelsDataStats: "{date} · {models} 模型 / {providers} 供应商",
+  modelsDataSyncNow: "立即同步",
+  modelsDataSyncing: "同步中...",
+  modelsDataSyncOk: "已同步，模型参数与单价已刷新",
+  modelsDataSyncFailed: "同步失败",
+  modelsDataRestore: "恢复内置数据",
+  modelsDataRestoreConfirm:
+    "删除本地同步副本并回退到随版本内置的快照数据？之后可随时重新同步。",
+
  // Plugin marketplace
  pluginMarketplace: "插件市场",
  pluginMarketplaceSubtitle:
@@ -134,10 +134,104 @@ describe("loop_control — v1/v2 key handling", () => {
 });

 // ---------------------------------------------------------------------------
-// [permission] dangerous_command_guard (kimi-code 0.40.1) — absent key means
-// upstream default ON; an explicit false must survive even with no rules
+// loop_control.compaction_max_attempts (kimi-code 0.43.0) — read via the loop
+// spread, written only when set, and 0/empty removes the key (upstream
+// default 5 applies)
 // ---------------------------------------------------------------------------

+describe("loop_control.compaction_max_attempts", () => {
+  it("reads an existing value through the loop spread", () => {
+    expect(
+      getAgentSettings({ loop_control: { compaction_max_attempts: 7 } })
+        .loop_control?.compaction_max_attempts
+    ).toBe(7);
+  });
+
+  it("reads as undefined when the key is absent (UI falls back to 5)", () => {
+    expect(
+      getAgentSettings({ loop_control: {} }).loop_control?.compaction_max_attempts
+    ).toBeUndefined();
+  });
+
+  it("writes the value when the patch sets it", () => {
+    const next = setAgentSettings({}, {
+      loop_control: { compaction_max_attempts: 8 },
+    });
+    expect(loopOf(next).compaction_max_attempts).toBe(8);
+  });
+
+  it("drops the key for 0 and for a cleared (undefined) field", () => {
+    const raw = { loop_control: { compaction_max_attempts: 6 } };
+    expect(loopOf(setAgentSettings(raw, {
+      loop_control: { compaction_max_attempts: 0 },
+    }))).not.toHaveProperty("compaction_max_attempts");
+    expect(loopOf(setAgentSettings(raw, {
+      loop_control: { compaction_max_attempts: undefined },
+    }))).not.toHaveProperty("compaction_max_attempts");
+  });
+
+  it("does not materialize the upstream default on an unrelated save", () => {
+    // An absent key must stay absent — a plain save must not pin `5`.
+    const next = setAgentSettings({ thinking: { effort: "high" } }, {});
+    expect(loopOf(next)).not.toHaveProperty("compaction_max_attempts");
+    // An existing value survives a save that does not touch it.
+    const kept = setAgentSettings(
+      { loop_control: { compaction_max_attempts: 4 } },
+      { thinking: { effort: "low" } }
+    );
+    expect(loopOf(kept).compaction_max_attempts).toBe(4);
+  });
+});
+
+// ---------------------------------------------------------------------------
+// [watch] enabled (kimi-code 2.0.1+) — top-level section, off by default since
+// 2.0.2; mirrors the [background] handling
+// ---------------------------------------------------------------------------
+
+function watchOf(raw: unknown): Record<string, unknown> {
+  const w = (raw as { watch?: unknown })?.watch;
+  return w && typeof w === "object" && !Array.isArray(w)
+    ? (w as Record<string, unknown>)
+    : {};
+}
+
+describe("[watch] enabled", () => {
+  it("reads as false when the section or key is absent (2.0.2 default)", () => {
+    expect(getAgentSettings({}).watch?.enabled).toBe(false);
+    expect(getAgentSettings({ watch: {} }).watch?.enabled).toBe(false);
+  });
+
+  it("reads an explicit value", () => {
+    expect(getAgentSettings({ watch: { enabled: true } }).watch?.enabled).toBe(true);
+    expect(getAgentSettings({ watch: { enabled: false } }).watch?.enabled).toBe(false);
+  });
+
+  it("writes the top-level [watch] section", () => {
+    const next = setAgentSettings({}, { watch: { enabled: true } }) as {
+      watch?: Record<string, unknown>;
+    };
+    expect(next.watch).toEqual({ enabled: true });
+  });
+
+  it("keeps the section on saves that do not touch it", () => {
+    const next = setAgentSettings(
+      { watch: { enabled: true } },
+      { thinking: { effort: "high" } }
+    ) as { watch?: Record<string, unknown> };
+    expect(next.watch).toEqual({ enabled: true });
+  });
+
+  it("does not disturb the other managed sections", () => {
+    const next = setAgentSettings(
+      { background: { max_running_tasks: 2 } },
+      { watch: { enabled: true } }
+    ) as { background?: Record<string, unknown> };
+    expect(next.background?.max_running_tasks).toBe(2);
+    expect(watchOf(next).enabled).toBe(true);
+  });
+});
+
+
 describe("permission.dangerous_command_guard", () => {
  it("reads as undefined when the key is absent (default on)", () => {
    expect(
@@ -13,6 +13,11 @@ const DEFAULT_SETTINGS: AgentSettings = {
  background: {
    keep_alive_on_exit: false,
  },
+  // `[watch] enabled` — kimi-code 2.0.1 added the key, 2.0.2 flipped the
+  // default to off, so an absent key reads as "off".
+  watch: {
+    enabled: false,
+  },
  permission: { rules: [] },
  hooks: [],
 };
@@ -76,6 +81,10 @@ export function getAgentSettings(rawOther: unknown): AgentSettings {
      ...DEFAULT_SETTINGS.background,
      ...getSection<AgentSettings["background"]>(rawOther, "background"),
    },
+    watch: {
+      ...DEFAULT_SETTINGS.watch,
+      ...getSection<AgentSettings["watch"]>(rawOther, "watch"),
+    },
    permission: {
      rules: sectionPermission?.rules ?? [],
      // `dangerous_command_guard` stays undefined unless the config carries
@@ -111,6 +120,7 @@ export function setAgentSettings(
    thinking: { ...current.thinking, ...patch.thinking },
    loop_control: { ...current.loop_control, ...patch.loop_control },
    background: { ...current.background, ...patch.background },
+    watch: { ...current.watch, ...patch.watch },
    permission,
    hooks: patch.hooks ?? current.hooks ?? [],
  };
@@ -129,9 +139,16 @@ export function setAgentSettings(
  if (!opts?.legacyV1) {
    delete next.loop_control?.max_retries_per_step;
  }
+  // `compaction_max_attempts` (kimi-code 0.43.0+): 0 / undefined means "leave
+  // the key out" — the config then falls back to the upstream default (5)
+  // instead of pinning a literal 0.
+  if (!next.loop_control?.compaction_max_attempts) {
+    delete next.loop_control?.compaction_max_attempts;
+  }
  if (next.thinking) root.thinking = next.thinking;
  if (next.loop_control) root.loop_control = next.loop_control;
  if (next.background) root.background = next.background;
+  if (next.watch) root.watch = next.watch;
  // Keep `[permission]` when it carries rules *or* an explicit
  // dangerous_command_guard — dropping the section would silently lose a
  // guard=false write (absent key = upstream default on). An empty rules
@@ -3,13 +3,18 @@
 * (see scripts/fetch-models-dev.mjs).
 *
 * The snapshot (~1.3 MB JSON) is NOT bundled into the main chunk anymore:
- * parsing a 1.3 MB JSON literal at startup blocks first paint. It is served
- * as a static asset (`/models-dev.json`) and loaded once in the background.
- * `getModelRef` stays synchronous and returns undefined until the index is
- * ready — callers already fall back to defaults — and `modelsDevReady()`
- * lets the app re-render once the index arrives.
+ * parsing a 1.3 MB JSON literal at startup blocks first paint. Loading order
+ * (see modelsDevReady): the runtime-synced copy under ~/.kimi-switch/
+ * (via models_dev.rs, written by the 高级设置 sync button) when present,
+ * otherwise the bundled static asset (`/models-dev.json`). `getModelRef`
+ * stays synchronous and returns undefined until the index is ready — callers
+ * already fall back to defaults — and `modelsDevReady()` lets the app
+ * re-render once the index arrives. `applyModelsDevSnapshot` hot-swaps the
+ * index after an online sync without a restart.
 */

+import { invoke } from "@tauri-apps/api/core";
+
 export interface ModelCost {
  input?: number;
  output?: number;
@@ -49,32 +54,64 @@ function buildIndex(raw: Record<string, unknown>) {
  return lower;
 }

+/**
+ * Swap in a new snapshot (initial load, online sync, or restore-to-builtin)
+ * and re-notify listeners so context/capability/price columns re-render.
+ * Listeners stay registered across reloads by design.
+ */
+function applySnapshot(raw: Record<string, unknown>) {
+  byLowerKey = buildIndex(raw);
+  for (const cb of readyListeners) cb();
+}
+
+async function fetchBundled(): Promise<Record<string, unknown>> {
+  const r = await fetch(`${import.meta.env.BASE_URL}models-dev.json`);
+  if (!r.ok) throw new Error(`models-dev.json HTTP ${r.status}`);
+  return (await r.json()) as Record<string, unknown>;
+}
+
 let loadPromise: Promise<void> | null = null;

 export function modelsDevReady(): Promise<void> {
  if (!loadPromise) {
-    loadPromise = fetch(`${import.meta.env.BASE_URL}models-dev.json`)
-      .then((r) => {
-        if (!r.ok) throw new Error(`models-dev.json HTTP ${r.status}`);
-        return r.json() as Promise<Record<string, unknown>>;
-      })
-      .then((raw) => {
-        byLowerKey = buildIndex(raw);
-        const listeners = readyListeners;
-        readyListeners = [];
-        for (const cb of listeners) cb();
-      })
-      .catch((err) => {
-        // A failed load is permanent for this session: fall back to defaults.
-        byLowerKey = {};
-        loadPromise = null; // allow one retry next time
-        console.warn("models-dev.json load failed:", err);
-      });
+    // Prefer the runtime-synced copy (~/.kimi-switch/models-dev.json, via
+    // models_dev.rs); outside Tauri (or without one) fall back to the bundled
+    // static asset. A failed load is permanent for this session.
+    loadPromise = (async () => {
+      let raw: Record<string, unknown> | null = null;
+      try {
+        const synced = await invoke<string | null>("get_models_dev_snapshot");
+        if (synced) raw = JSON.parse(synced) as Record<string, unknown>;
+      } catch {
+        /* not running inside Tauri — use the bundled asset */
+      }
+      applySnapshot(raw ?? (await fetchBundled()));
+    })().catch((err) => {
+      byLowerKey = {};
+      loadPromise = null; // allow one retry next time
+      console.warn("models-dev.json load failed:", err);
+    });
  }
  return loadPromise;
 }

-/** Register a callback invoked once the models.dev index is ready. */
+/**
+ * Apply a freshly synced snapshot (already fetched by the caller) and notify
+ * listeners so the UI refreshes without a restart.
+ */
+export function applyModelsDevSnapshot(raw: Record<string, unknown>): void {
+  applySnapshot(raw);
+}
+
+/**
+ * Reload from the bundled static asset (after "restore built-in data" —
+ * the synced copy was already deleted on the Rust side).
+ */
+export function reloadModelsDev(): Promise<void> {
+  return fetchBundled().then(applySnapshot);
+}
+
+/** Register a callback invoked whenever the models.dev index changes. */
 export function onModelsDevReady(cb: () => void): void {
  if (byLowerKey) {
    cb();
@@ -461,22 +461,23 @@ describe("validateSubagentPool", () => {
 // ---------------------------------------------------------------------------

 describe("experimental flag registry", () => {
-  it("mirrors the 0.42.0 v2 registry (6 flags)", () => {
+  it("mirrors the 0.43.0 v2 registry (5 flags)", () => {
    expect(EXPERIMENTAL_FLAGS.map((f) => f.id)).toEqual([
      "wait_for",
      "tool-select",
      "notify_user",
      "tower",
      "subagent_fork",
-      "auto_session_title",
    ]);
    // Removed upstream: file_history (0.41.0, turn-level file history is
-    // always on) plus secondary-model / persistence_minidb_readmodel /
-    // remote-control / search_worker (0.42.0 — promoted or moved to
-    // dedicated config, the minidb read-model now lives under [database]),
-    // so the adapter must not mirror them either.
+    // always on), auto_session_title (0.43.0, graduated — AI session titles
+    // are always on, #3749), plus secondary-model /
+    // persistence_minidb_readmodel / remote-control / search_worker
+    // (0.42.0 — promoted or moved to dedicated config, the minidb read-model
+    // now lives under [database]), so the adapter must not mirror them either.
    for (const id of [
      "file_history",
+      "auto_session_title",
      "secondary-model",
      "persistence_minidb_readmodel",
      "remote-control",
@@ -33,10 +33,11 @@ export interface ExperimentalFlagDef {
  defaultEnabled?: boolean;
 }

-/** Known experimental flags — mirrors the kimi-code 0.42.0 v2 flag registry
+/** Known experimental flags — mirrors the kimi-code 0.43.0 v2 flag registry
 * (the per-feature flag.ts files under packages/agent-core-v2/src):
- * wait_for, tool-select, notify_user, tower, subagent_fork,
- * auto_session_title. Removed in 0.42.0: secondary-model, remote-control
+ * wait_for, tool-select, notify_user, tower, subagent_fork.
+ * Removed in 0.43.0: auto_session_title (graduated — AI session titles are
+ * always on, #3749). Removed in 0.42.0: secondary-model, remote-control
 * and search_worker (promoted or dropped upstream — no flag left), and
 * persistence_minidb_readmodel (now controlled by the `[database]` config
 * section). `acp-v2` was removed upstream earlier; `file_history` was
@@ -53,10 +54,6 @@ export const EXPERIMENTAL_FLAGS: ExperimentalFlagDef[] = [
  { id: "notify_user", envVar: "KIMI_CODE_EXPERIMENTAL_NOTIFY_USER" },
  { id: "tower", envVar: "KIMI_CODE_EXPERIMENTAL_TOWER" },
  { id: "subagent_fork", envVar: "KIMI_CODE_EXPERIMENTAL_SUBAGENT_FORK" },
-  {
-    id: "auto_session_title",
-    envVar: "KIMI_CODE_EXPERIMENTAL_AUTO_SESSION_TITLE",
-  },
 ];

 export const EXPERIMENTAL_MASTER_ENV = "KIMI_CODE_EXPERIMENTAL_FLAG";
@@ -59,6 +59,7 @@ describe("planLabel", () => {
    expect(planLabel("five_hour", t)).toBe("usageTier5h");
    expect(planLabel("weekly_limit", t)).toBe("usageTierWeekly");
    expect(planLabel("monthly_limit", t)).toBe("usageTierMonthly");
+    expect(planLabel("booster_wallet", t)).toBe("usageTierBoosterWallet");
    expect(planLabel("NewAPI", t)).toBe("NewAPI");
    expect(planLabel("¥", t)).toBe("¥");
  });
@@ -5,15 +5,23 @@ type TranslateFn = (
  vars?: Record<string, string | number>
 ) => string;

+/**
+ * Kimi quota 的加力钱包行 id（src-tauri/src/services/coding_plan.rs 的
+ * PLAN_BOOSTER_WALLET）。它是金额行而非配额百分比行，planLabel() 与
+ * UsageFooter 的金额行判定共用此常量，避免字面量散落。
+ */
+export const BOOSTER_WALLET_PLAN = "booster_wallet";
+
 /**
 * Map a raw Rust plan/tier name to a localized label. Unknown names (proper
 * nouns like "OpenRouter", "NewAPI", currency codes) pass through unchanged.
 *
 * Coverage note: ALL plan-type (套餐) queries flow through
 * src-tauri/src/services/coding_plan.rs::percent_tier, whose names come only
- * from TIER_FIVE_HOUR / TIER_WEEKLY_LIMIT / TIER_MONTHLY_LIMIT — so this
- * mapping covers every current and future plan provider automatically. If a
- * new tier id is ever added there, add a case here too.
+ * from TIER_FIVE_HOUR / TIER_WEEKLY_LIMIT / TIER_MONTHLY_LIMIT /
+ * TIER_MONTH_CODE — so this mapping covers every current and future plan
+ * provider automatically. If a new tier id is ever added there, add a case
+ * here too.
 */
 export function planLabel(name: string, t: TranslateFn): string {
  switch (name) {
@@ -25,6 +33,10 @@ export function planLabel(name: string, t: TranslateFn): string {
      return t("usageTierWeekly");
    case "monthly_limit":
      return t("usageTierMonthly");
+    case "month_code":
+      return t("usageTierMonthCode");
+    case BOOSTER_WALLET_PLAN:
+      return t("usageTierBoosterWallet");
    default:
      return name;
  }
@@ -50,7 +50,10 @@ export function validateProviders(
    const reasons: string[] = [];

    const apiKey = provider.api_key?.trim();
-    if (!apiKey && !provider.managed) {
+    // kimi-code 2.0.0+: the credential may come from an environment variable
+    // instead of a stored key — a provider configured that way is complete.
+    const apiKeyEnv = provider.api_key_env?.trim();
+    if (!apiKey && !apiKeyEnv && !provider.managed) {
      reasons.push(t("validationNoApiKey"));
    }

@@ -15,6 +15,12 @@ export interface Provider {
  provider_type: ProviderType;
  base_url: string | null;
  api_key: string | null;
+  /**
+   * Name of the process environment variable holding the credential
+   * (kimi-code 2.0.0+). Mutually exclusive with `api_key` upstream — the
+   * export writes only one of the two. Only the variable *name* is stored.
+   */
+  api_key_env?: string | null;
  env: Record<string, string>;
  /** Optional note / remark for the provider. */
  note?: string | null;
@@ -90,6 +96,9 @@ export interface UsageConfig {
  autoQueryIntervalMinutes?: number;
  /** Per-request timeout in seconds; 0/undefined = default (8s). */
  timeoutSeconds?: number;
+  /** 用量预警阈值（百分比 0-100）；0/undefined = 关闭预警。
+   *  判定见 src/hooks/useUsageQuery.ts 的 isAlertTier()。 */
+  threshold?: number;
 }

 /** A model discovered from a provider's API endpoint. */
@@ -125,6 +134,9 @@ export interface LoopControlConfig {
   *  KIMI_CODE_LEGACY_FLAG=1 (v1 engine compat). v2 saves strip it. */
  max_retries_per_step?: number;
  reserved_context_size?: number;
+  /** Retry budget for context compaction (kimi-code 0.43.0+). Absent/0 = the
+   *  key is not written and upstream's default (5) applies. */
+  compaction_max_attempts?: number;
 }

 export interface BackgroundConfig {
@@ -132,6 +144,13 @@ export interface BackgroundConfig {
  keep_alive_on_exit?: boolean;
 }

+/** Top-level `[watch]` section (kimi-code 2.0.1+). The key was added in
+ *  2.0.1 and its default flipped to off in 2.0.2, so an absent key reads as
+ *  disabled. `KIMI_CODE_WATCH` overrides the config value at runtime. */
+export interface WatchConfig {
+  enabled?: boolean;
+}
+
 export interface PermissionRule {
  decision?: "allow" | "deny" | "ask";
  scope?: string;
@@ -150,6 +169,8 @@ export interface AgentSettings {
  thinking?: ThinkingConfig;
  loop_control?: LoopControlConfig;
  background?: BackgroundConfig;
+  /** Top-level `[watch]` section (kimi-code 2.0.1+). */
+  watch?: WatchConfig;
  permission?: {
    rules?: PermissionRule[];
    /** kimi-code `[permission]` key. Default true when the key is
@@ -39,6 +39,12 @@ export interface ActionResponse {
  deleted?: boolean | null;
 }

+export interface BulkArchiveResult {
+  archived: number;
+  skipped: number;
+  errors: string[];
+}
+
 export interface PreviewMessage {
  role: string;
  time: number | null;
Author	SHA1	Message	Date
KimiSwitch Dev	13e15657af	fix(usage): 列表支持显式模板查询（Sub2API/NewAPI）——对齐后端绕过 usage_kinds 的口径 — bump to v0.7.22 Release / Version consistency (push) Canceled after 0s Details Release / Build (macos-latest) (push) Canceled after 0s Details Release / Build (ubuntu-latest) (push) Canceled after 0s Details Release / Build (windows-latest) (push) Canceled after 0s Details Release / Attach macOS install script (push) Canceled after 0s Details	2026-09-29 10:42:13 +08:00
KimiSwitch Dev	11286419b1	feat(models-dev): 高级设置新增 models.dev 在线同步 — 上下文/能力/单价即时生效 + 一键恢复内置；修复 public 快照构建漂移；确认适配 kimi-code 2.1.0 — bump to v0.7.21 Release / Version consistency (push) Canceled after 0s Details Release / Build (macos-latest) (push) Canceled after 0s Details Release / Build (ubuntu-latest) (push) Canceled after 0s Details Release / Build (windows-latest) (push) Canceled after 0s Details Release / Attach macOS install script (push) Canceled after 0s Details	2026-09-24 00:19:10 +08:00
KimiSwitch Dev	26e783c2ba	chore(data): 同步 models-dev 快照至 2026-09-20（7870 模型 / 222 供应商）	2026-09-20 10:29:40 +08:00
KimiSwitch Dev	11f87fd9af	chore(docs): release notes 归档 docs/release-notes/ + 根目录汇总 CHANGELOG.md - 20 份 release-notes-v*.md 移入 docs/release-notes/（git mv 保留历史） - 新增 CHANGELOG.md：v0.7.1~v0.7.20 倒序浓缩汇总（日期取各文件原始添加提交） - publish-gitea.sh 与 release.yml 改为 docs/release-notes/<tag>.md 优先、根目录回退	2026-09-20 09:58:01 +08:00
KimiSwitch Dev	fcd08eae32	feat: 阈值预警/加力钱包/api_key_env/版本跟踪器/一键体检 — bump to v0.7.20 Release / Version consistency (push) Canceled after 0s Details Release / Build (macos-latest) (push) Canceled after 0s Details Release / Build (ubuntu-latest) (push) Canceled after 0s Details Release / Build (windows-latest) (push) Canceled after 0s Details Release / Attach macOS install script (push) Canceled after 0s Details	2026-09-20 09:49:49 +08:00
KimiSwitch Dev	35581103ed	feat(usage): plan:kimi_coding 兼容 kimi-code 0.43.1 quota 用量模型服务端 api.kimi.com/coding/v1/usages 已由绝对 used/limit 行切换为 quota 模型（usages.limit_5h/limit_7d/limit_month_total/limit_month_code，条目为 used_ratio 0-1 比率 + reset_time），见上游 #3787。 - parse_kimi_coding 优先解析新结构，旧 limits/usage 结构保留为回退 - used_ratio >1 时按已是百分数处理（防御） - 新 tier id month_code + 前端 planLabel/i18n 中英文案 - 新增 3 个测试：四窗口解析 / 比率防御 / 空窗口回退旧结构	2026-09-16 14:14:29 +08:00
KimiSwitch Dev	1b80d66239	chore(flags): 适配 kimi-code 0.43.0 — 移除转正 auto_session_title 实验 flag — bump to v0.7.19 Release / Version consistency (push) Canceled after 0s Details Release / Build (macos-latest) (push) Canceled after 0s Details Release / Build (ubuntu-latest) (push) Canceled after 0s Details Release / Build (windows-latest) (push) Canceled after 0s Details Release / Attach macOS install script (push) Canceled after 0s Details	2026-09-14 22:40:44 +08:00
KimiSwitch Dev	0f421b4e44	feat(sessions): 一键归档 + 归档用量快照落库 — bump to v0.7.18 Release / Version consistency (push) Canceled after 0s Details Release / Build (macos-latest) (push) Canceled after 0s Details Release / Build (ubuntu-latest) (push) Canceled after 0s Details Release / Build (windows-latest) (push) Canceled after 0s Details Release / Attach macOS install script (push) Canceled after 0s Details - archive_sessions_before：按 updatedAt 截止点批量元数据归档（1月/半月/周/指定日期） - 归档瞬间聚合 wire.jsonl 用量按天×模型快照进 SQLite archived_sessions 表 - 统计合并：文件已删除的归档会话由快照合成 UsageRecord 注入 get_summary/日明细，趋势/热力图/模型统计全兼容且绝不双计 - models-dev 快照同步 2026-09-14（7784 模型）；pricing 测试跟随 deepseek-v4-flash 调价（0.15/0.60） - cargo test 129 通过（含端到端归档→删目录→统计保留用例）、tsc 零错误	2026-09-14 01:52:21 +08:00