diff --git a/specs/mvp/image/Snipaste_2025-12-30_16-06-37.png b/specs/mvp/image/Snipaste_2025-12-30_16-06-37.png
new file mode 100644
index 0000000..fb62750
Binary files /dev/null and b/specs/mvp/image/Snipaste_2025-12-30_16-06-37.png differ
diff --git a/specs/mvp/image/Snipaste_2025-12-30_17-02-20.png b/specs/mvp/image/Snipaste_2025-12-30_17-02-20.png
new file mode 100644
index 0000000..4005c93
Binary files /dev/null and b/specs/mvp/image/Snipaste_2025-12-30_17-02-20.png differ
diff --git a/specs/mvp/image/roadmap_v3.0.png b/specs/mvp/image/roadmap_v3.0.png
new file mode 100644
index 0000000..8424345
Binary files /dev/null and b/specs/mvp/image/roadmap_v3.0.png differ
diff --git a/specs/mvp/v3.0/README.md b/specs/mvp/v3.0/README.md
new file mode 100644
index 0000000..b338179
--- /dev/null
+++ b/specs/mvp/v3.0/README.md
@@ -0,0 +1,15 @@
+# MVP v3.0（Design）— WebUI + 用户数据上传/下载（SFTPGo）→ 首个可发布版本
+
+本目录基于：
+- `specs/mvp/mvp_roadmap_v2.md`（总体路线图）
+- `specs/mvp/image/roadmap_v3.0.png`（v3.0 迭代图）
+- 当前已落地的 v2.5（User Mgmt + Stateless Ray Node Pool）
+
+目标是在 v2.5 的基础上补齐 **用户数据闭环**（上传→训练可见→产物下载）以及最小可用的 **WebUI**，形成“可发布”的 v3.0 版本。
+
+文档：
+- `specs/mvp/v3.0/v3.0_design.md`：总体架构与关键机制（WebUI、SFTPGo、数据/权限模型、任务流）。
+- `specs/mvp/v3.0/v3.0_api.md`：v3.0 API 扩展设计（UI、数据、SFTPGo 管理、权限约束）。
+- `specs/mvp/v3.0/v3.0_acceptance.md`：部署/升级/验收流程与可验证标准（含故障注入与回归清单）。
+- `specs/mvp/v3.0/v3.0_dev_plan.md`：TDD 驱动的工程化开发计划（里程碑拆分、测试分层、E2E 验收）。
+- `specs/mvp/v3.0/v3.0_progress.md`：实施进展记录（每个里程碑完成后追加记录）。
diff --git a/specs/mvp/v3.0/v3.0_acceptance.md b/specs/mvp/v3.0/v3.0_acceptance.md
new file mode 100644
index 0000000..fe710e5
--- /dev/null
+++ b/specs/mvp/v3.0/v3.0_acceptance.md
@@ -0,0 +1,55 @@
+# MVP v3.0 — 部署与验收流程（草案）
+
+## 0) 环境前提
+- Ray 集群：延续 v2.5 的 head + stateless worker（自动 join）
+- 共享存储：容器内挂载 `/private`（dev/prod 对齐）
+- API server：宿主机代码挂载到 head 容器，在 head 容器内启动
+- 新增：SFTPGo 服务（建议容器化部署）
+
+## 1) 部署步骤（高层）
+
+1) 部署/升级 Ray 节点镜像（沿用 v2.5 的 `argus/argus-ray-node:v2.5` 或更高版本）
+2) 启动 Ray 集群（compose 或平台创建容器）
+3) 启动/配置 SFTPGo（挂载 `/private`）
+4) 启动 API server（head 容器内）
+5) 启动 WebUI（由 API server 托管）
+
+## 2) 验收用例（必须通过）
+
+### A. 用户与凭据
+1) admin 创建用户 `alice`，签发 API token
+2) 系统联动在 SFTPGo 创建 `alice`（home=/private/users/alice）
+3) `alice` 使用 token 登录 WebUI（或调用 `/api/v2/me` 成功）
+
+### B. 上传数据闭环（核心）
+1) `alice` 通过 SFTP 上传数据集到 `/private/users/alice/datasets/...`
+2) `alice` 通过 WebUI/API 提交任务，TaskSpec 引用该路径
+3) Ray worker 读取该数据，任务 RUNNING 并最终 SUCCEEDED
+
+### C. 下载产物闭环
+1) 训练完成后，产物落到 `/private/users/alice/jobs/<submission_id>/...`
+2) `alice` 通过 SFTP 下载 checkpoints/logs 成功
+3) （新增）`alice` 将需要长期保留的权重从 `jobs/<submission_id>/...` 移动到 `models/`，确认移动后可长期存在
+
+### C2. Jobs 回收站与自动清理（3 天移入回收站，7 天后永久删除）
+1) 将 `jobs_trash_after_days`/`jobs_purge_after_days` 配置为较小值（例如分钟级，用于验证）
+2) 训练完成进入 terminal 状态
+3) 等待 API server 内置 janitor 扫描周期后，确认对应 `jobs/<submission_id>` 被移动到 `trash/jobs/<submission_id>`
+4) 在回收站窗口内，把某个文件从 `trash/jobs/<submission_id>` 移动到 `models/`，确认移动成功
+5) 等待超过 `jobs_purge_after_days` 后，确认 `trash/jobs/<submission_id>` 被永久删除
+6) 确认已移动到 `models/` 的文件不被删除
+
+### D. 安全隔离（必须）
+1) `bob` 不能通过 API 查询 `alice` 的 task（404）
+2) `bob` 不能提交引用 `/private/users/alice/...` 的 TaskSpec（400/403）
+3) `bob` 通过 SFTP 无法访问 `/private/users/alice/...`（chroot 生效）
+
+## 3) 故障注入（推荐通过）
+1) kill worker watchdog 或 raylet → worker 自动恢复并重新加入集群
+2) 重启 head 容器 → head 重新写 `head.json`，worker 自动重连
+3) SFTPGo 重启 → 不影响 Ray 集群；用户可重新连接上传/下载
+
+## 4) 回归清单（与 v2.5 一致）
+- 任务队列、重试（INSUFFICIENT_RESOURCES → PENDING_RESOURCES → retry）
+- PPO/GRPO/SFT 三种 workload 均可跑通
+- head 不跑训练（driver 强制落 worker）
diff --git a/specs/mvp/v3.0/v3.0_api.md b/specs/mvp/v3.0/v3.0_api.md
new file mode 100644
index 0000000..67f113b
--- /dev/null
+++ b/specs/mvp/v3.0/v3.0_api.md
@@ -0,0 +1,109 @@
+# MVP v3.0 — API 扩展设计（基于 v2.5）
+
+v3.0 的原则是：**尽量复用 v2.5 API**，只增量增加 “数据闭环” 与 “WebUI 支持” 所需的最小接口。
+
+## 1) 认证与权限
+
+沿用 v2.5：
+- Header：`Authorization: Bearer <token>`
+- admin token：来自 `MVP_INTERNAL_TOKEN`
+- 普通用户 token：由 admin 颁发并持久化在 SQLite
+
+权限规则：
+- 非 admin：只能访问自己的 task、自己的数据空间（`/private/users/<user_id>/...`）。
+- 跨用户访问返回 404（不泄露存在性）。
+
+## 2) 用户与 SFTPGo 联动（管理员接口）
+
+### 2.1 创建用户（复用 v2.5）
+`POST /api/v2/users`
+- v3.0 行为：成功后，**可选**联动创建 SFTPGo 用户
+  - v3.0 默认启用联动：创建 SFTPGo 用户 + 生成一次性密码（password 认证）
+  - v3.0 仅保留该方案（方案 A）：不做外部认证/SSO 集成（留到更后续版本）
+  - `data.sftpgo.admin_api_base` 推荐形如：`http://argus-sftpgo:8080/api/v2`（包含 `/api/v2` 前缀）
+
+### 2.2 下发 token（复用 v2.5）
+`POST /api/v2/users/{user_id}/tokens`
+
+### 2.3 禁用用户（复用 v2.5）
+`POST /api/v2/users/{user_id}:disable`
+- v3.0 行为：联动禁用 SFTPGo 用户（可选）
+
+### 2.4 SFTP 凭据管理（新增，管理员或用户自助）
+（具体由你确认 v3.0 需要“用户自助”还是“管理员操作”）
+
+#### 重置 SFTP 密码（管理员）
+`POST /api/v2/users/{user_id}/sftp:reset_password`
+- 返回：一次性密码（只返回一次，服务端不保存明文）
+> v3.0 先只做 password 方案；SSH public key 作为后续版本可选增强（不在 v3.0 范围）。
+
+## 3) 用户自助信息（新增）
+
+### 3.1 获取当前用户信息
+`GET /api/v2/me`
+- 返回示例：
+```json
+{
+  "user_id": "alice",
+  "is_admin": false,
+  "paths": {
+    "home": "/private/users/alice",
+    "datasets": "/private/users/alice/datasets",
+    "models": "/private/users/alice/models",
+    "code": "/private/users/alice/code",
+    "jobs": "/private/users/alice/jobs",
+    "trash_jobs": "/private/users/alice/trash/jobs"
+  },
+  "retention": {
+    "jobs_trash_after_days": 3,
+    "jobs_purge_after_days": 7
+  },
+  "sftp": {
+    "host": "h1.example.internal",
+    "port": 2022,
+    "username": "alice"
+  }
+}
+```
+
+## 3.2 Jobs Retention 提示（新增）
+为了支撑 WebUI 展示与用户预期管理，可在 `/api/v2/me` 或单独接口返回：
+- `jobs_trash_after_days`：默认 3
+- `jobs_purge_after_days`：默认 7
+- `jobs_root`：`/private/users/<me>/jobs`
+- `trash_jobs_root`：`/private/users/<me>/trash/jobs`
+- `recommendations`：提示用户把需要长期保存的产物移动到 `models/` 或 `datasets/`
+
+## 4) 数据浏览/下载（可选，v3.0 最小化）
+
+说明：上传/下载主通道仍是 SFTP。  
+WebUI 如果要提供“快速浏览/查看”，可实现只读接口（避免实现大文件上传/断点等复杂逻辑）。
+
+### 4.1 列目录
+`GET /api/v2/files?path=/private/users/alice`
+- 权限：path 必须在 `/private/common/` 或 `/private/users/<me>/` 下
+- 返回：文件列表（name/type/size/mtime）
+
+### 4.2 下载文件（小文件为主）
+`GET /api/v2/files:download?path=/private/users/alice/jobs/.../logs/...`
+- 返回：流式下载
+- 大文件仍建议走 SFTP
+
+## 5) TaskSpec 路径校验升级（v3.0 关键）
+
+v2.5：仅允许 `/private/common/...`  
+v3.0：允许 `/private/common/...` 与 `/private/users/<me>/...`
+
+应用字段（至少）：
+- `train_file` / `val_file`
+- `code_path`：仍仅允许 `/private/common/...`（v3.0 不支持执行用户 code）
+- 本地模型路径字段（如果引入）：允许 `/private/users/<me>/models/...`
+
+## 6) WebUI 路由（新增）
+
+由 API server 托管：
+- `GET /ui`：主页面
+- `GET /ui/login`：token 登录页
+- 静态资源：`/ui/static/...`
+
+WebUI 的所有操作均调用同源 API（不额外开 CORS）。
diff --git a/specs/mvp/v3.0/v3.0_design.md b/specs/mvp/v3.0/v3.0_design.md
new file mode 100644
index 0000000..4edf84b
--- /dev/null
+++ b/specs/mvp/v3.0/v3.0_design.md
@@ -0,0 +1,358 @@
+# MVP v3.0 详细设计方案（基于 v2.5）
+
+## 0. 结论摘要（v3.0 要交付什么）
+
+v3.0 = v2.5 + **WebUI** + **用户数据上传/下载（SFTPGo）**，形成第一个可对外发布的版本：
+- 用户可以通过 **SFTP** 上传数据/模型/代码（至少数据），落到 GPFS（容器内 `/private`）并对 Ray worker 可见。
+- 用户可以通过 API/WebUI 提交训练任务，任务读取自己上传的数据。
+- 用户可以下载训练产物（checkpoints/logs 等），最小闭环跑通。
+
+## 1. 范围与原则
+
+### 1.1 继承 v2.5 的前提（不回退）
+- **Stateless Ray Node Pool**：head 写 `head.json`，worker watchdog 自动 join/自愈。
+- **User Management**：token 鉴权、任务可见性隔离（跨用户 404 不泄漏）。
+- **作业产物隔离**：Ray job 目录落到 `/private/users/<user_id>/jobs/<ray_submission_id>/...`。
+- **API server 短期运行方式**：代码在宿主机，挂载到 head 容器，在 head 容器内启动（保持现状）。
+
+### 1.2 v3.0 新增目标
+1) **Data Management（SFTPGo）**
+   - 提供用户上传/下载入口（SFTP 为主）。
+   - 数据落到 GPFS（dev 环境 NFS/GPFS，生产环境 GPFS），训练 job 在 worker 容器内可直接读取。
+2) **WebUI**
+   - 用户可视化创建任务、查看队列/状态/日志、查看“数据路径约定”和自己的 SFTP 信息。
+   - 目标是 “可用而非豪华”，支持核心工作流。
+3) **权限闭环**
+   - 用户只能使用自己目录下的数据（`/private/users/<user_id>/...`）或公共目录（`/private/common/...`）。
+   - 防止用户提交任务读取其他用户的文件路径。
+
+### 1.3 v3.0 明确不做（留给 v3.5）
+- 不做 “自定义 reward function / 自定义 verl 代码 / 多版本 verl 共存”（路线图 v3.5）。
+- 不做复杂 Serving/训推一体（路线图 v3.5）。
+- 不做 IB 网络/拓扑优化（路线图 v3.5）。
+- 不做系统级可观测性平台（路线图 v4.0）。
+
+## 2. 架构概览
+
+参考 `roadmap_v3.0.png`，v3.0 的控制面与数据面：
+
+### 2.1 控制面（Control Plane）
+- **API Server（FastAPI）**
+  - v2.5 的任务队列/调度/重试 + 用户管理能力继续复用
+  - 新增：数据管理能力（与 SFTPGo 对接） + WebUI
+- **WebUI**
+  - 通过 API 使用 token 登录
+  - 提供任务/日志/数据入口（不直接运行训练）
+- **Ray Head（状态节点）**
+  - 仍在 head 容器内（或单独节点）
+  - job server/dashbaord 提供 job submit/status/logs 能力
+
+### 2.2 数据面（Data Plane）
+- **GPFS（容器内挂载 `/private`）**
+  - 存放 common 与 users 两大根目录
+- **Ray Worker Node（无状态）**
+  - 自动连接 head，执行训练
+  - 读取 `/private/users/<user>/...` 的数据
+
+### 2.3 新增组件：SFTPGo（Data Management）
+- 作为独立服务运行（容器化优先），后端存储使用 **filesystem**（GPFS 挂载路径）。
+- 用户的 home directory 指向 `/private/users/<user_id>`（或其子目录）。
+
+## 3. 存储与目录规范（v3.0 统一约定）
+
+### 3.1 目录层级
+统一以容器内 `/private` 作为根路径（dev/prod 对齐）：
+- `/private/common/`：公共资源
+  - `hf/`：HF cache
+  - `datasets/`：公共数据集（可选）
+  - `code/`：公共代码（例如公共 verl repo snapshot）
+  - `db/`：SQLite（队列、用户、token）
+  - `logs/`：API/supervisor/watchdog 日志
+- `/private/users/<user_id>/`：用户空间（v3.0 重点）
+  - `datasets/`：用户上传的数据集（推荐）
+  - `models/`：用户保存/上传的本地模型（允许；也用于“把 job 产物移动到长期保存目录”）
+  - `code/`：用户上传的代码（v3.0 **不支持执行**；仅存放/下载）
+  - `jobs/`：训练任务产物（已在 v2.5 落地）
+  - `tmp/`：临时文件（可选）
+
+### 3.2 Jobs Retention（两段式：3 天移入回收站，7 天后永久删除）
+v3.0 引入 **jobs 目录两段式保留策略**：
+- 第 1 阶段（soft-delete）：job 结束后 **3 天**，将该 job 目录从 `jobs/` **移动到用户回收目录**；
+- 第 2 阶段（hard-delete）：进入回收目录后再过 **7 天**，从回收目录 **永久删除**。
+
+目录约定（建议）：
+- jobs 根目录：`/private/users/<user_id>/jobs/<ray_submission_id>/...`
+- 回收目录：`/private/users/<user_id>/trash/jobs/<ray_submission_id>/...`
+
+计时规则：
+- 以 job 进入 terminal 状态（SUCCEEDED/FAILED/CANCELED）的结束时间为起点；
+- “3 天”用于从 `jobs/` 移入 `trash/jobs/`；
+- “7 天”用于从 `trash/jobs/` 永久删除（即总共最多 10 天窗口）。
+
+用户保留关键产物的方式（无需 keep 标记）：
+- 在 “3 天窗口”内把需要长期保存的文件从 `jobs/<submission_id>/...` **移动/复制**到 `models/`（例如权重）或 `datasets/`（例如评估输出数据）；
+- 即便已被移动到回收目录，用户仍可在 “7 天窗口”内从 `trash/jobs/<submission_id>/...` 把需要的文件移到 `models/` / `datasets/`；
+- janitor 只管理 `jobs/` 与 `trash/jobs/`，不会触碰 `models/` 与 `datasets/`。
+
+这里的“清理程序”我们称为 **janitor**：
+- 定义：一个后台清理执行器，按固定周期扫描“已结束且已过期”的 job 目录并删除
+- v3.0 目标：实现“3 天移入回收站 + 7 天后删除”这一条产品规则（不提供 keep/延长保留标记）
+
+实现建议（按你的偏好）：
+- **janitor 作为 API server 内置后台线程**运行：
+  - 优点：天然可访问 SQLite（任务状态、结束时间、user_id、ray_submission_id），并能把清理结果写回 events 表用于审计
+  - 部署更简单：不额外引入 cronjob/独立服务
+- 删除/移动动作建议 **直接在 GPFS/NFS 文件系统上操作**（API server 运行在 head 容器，已挂载 `/private`）：
+  - 第 1 阶段：`os.rename`（同文件系统原子移动）把 `jobs/<sid>` 移到 `trash/jobs/<sid>`；
+    - 若跨文件系统（理论上不应发生），则降级为 copy+delete；
+    - 移动前做严格路径前缀校验（必须在 `.../users/<u>/jobs/` 下）。
+  - 第 2 阶段：对 `trash/jobs/<sid>` 执行递归删除（例如 `shutil.rmtree`），同样做路径前缀校验（必须在 `.../users/<u>/trash/jobs/` 下）。
+  - 为什么不依赖 SFTPGo API：SFTPGo 只是用户访问协议层（SFTP/Web），目录物理就在同一份文件系统；文件系统直连更简单、也不依赖 SFTPGo 在线。
+- 如果你强烈希望“通过 SFTPGo API 删除”：
+  - 可以作为可选实现/补充（例如用于统一审计或未来接入配额/策略），但不建议作为唯一手段（SFTPGo 停机不应阻塞清理）。
+
+### 3.3 用户在 SFTPGo 内移动/整理文件（确认点）
+支持用户在 SFTPGo 中进行“移动/重命名/整理”（例如把权重从 `jobs/` 移动到 `models/`）：
+- 前提：SFTPGo 用户权限允许对其 home 目录进行 `rename/mkdir/remove` 等操作（v3.0 默认可写）。
+- 行为：用户可以把 `jobs/` 下某些文件移动到 `models/` 或 `datasets/`，用于长期保存权重/评估产物等。
+- 与 retention 的关系：只要文件被移动出 `jobs/`，就不会被 jobs 清理逻辑删除。
+
+### 3.4 路径权限规则（API 侧校验）
+v2.5 约束是 “只允许 `/private/common/...`”。  
+v3.0 需要升级为：
+- 允许：
+  - `/private/common/...`
+  - `/private/users/<current_user_id>/...`
+- 禁止：
+  - 任何其他绝对路径（例如 `/private/users/other/...`、`/etc/...`）
+
+并把该规则应用到 TaskSpec 的相关字段（至少）：
+- `train_file` / `val_file`
+- `code_path`：仍仅允许 `/private/common/...`（v3.0 不支持执行用户 code）
+- 本地模型路径字段：允许 `/private/users/<me>/models/...`（确认：v3.0 允许）
+
+## 4. SFTPGo 方案设计（Data Management）
+
+### 4.1 运行形态
+推荐用容器运行 SFTPGo（与 Ray/API 解耦），挂载同一份 `/private`：
+- `sftpgo` 容器挂载 `../../shared:/private`
+- 对外暴露：
+  - SFTP 端口（建议 2022）
+  - WebAdmin/API 端口（建议 8081，仅内网或管理员访问）
+
+#### 4.1.1 镜像来源（现成 Docker 镜像）
+SFTPGo 有现成可用的 Docker 镜像（无需自建）：
+- v3.0 推荐优先使用官方/上游发布的 `sftpgo` 镜像作为运行基座
+- 我们在 v3.0 里不需要定制 SFTPGo 代码，只需要：
+  - 正确挂载 GPFS/NFS（容器内 `/private`）
+  - 配置管理员账号（用于 API server 联动创建/禁用用户、重置密码）
+  - 配置每用户 home/chroot
+
+> 注意：具体镜像名/tag 在不同环境可能有差异（官方/镜像仓库策略会变动）。落地时建议在 `argus@h1` 上 `docker search sftpgo` 或由你们内部镜像仓库提供固定版本；v3.0 设计只要求“使用现成镜像”，不强依赖某个 tag。
+
+#### 4.1.2 docker-compose 服务草案（示意）
+下面给出一个**示意**（最终以实际镜像名/tag 与你们端口规划为准）：
+
+```yaml
+services:
+  sftpgo:
+    image: sftpgo/sftpgo:latest   # 示例：使用现成镜像
+    container_name: argus-sftpgo
+    ports:
+      - "2022:2022"   # SFTP
+      - "8081:8080"   # WebAdmin/API（建议仅内网/管理员）
+    volumes:
+      - ../../shared:/private
+      - ../../shared/common/sftpgo:/var/lib/sftpgo   # 持久化 SFTPGo 元数据（可选/建议）
+    environment:
+      # 管理员账号/密码（示意，具体变量名以镜像文档为准）
+      SFTPGO_ADMIN_USERNAME: "admin"
+      SFTPGO_ADMIN_PASSWORD: "${SFTPGO_ADMIN_PASSWORD}"
+```
+
+与 v3.0 的配合点：
+- API server 使用 `data.sftpgo.admin_api_base` + admin 凭据联动创建用户
+- 用户 home/chroot 统一指向 `/private/users/<user_id>`
+
+### 4.2 用户隔离
+每个用户在 SFTPGo 中的 home dir 绑定到：
+- `/private/users/<user_id>`（chroot），用户只能读写自己的目录。
+
+### 4.3 用户创建与凭据管理（两种实现，建议先做 A）
+
+**方案 A（v3.0 推荐）：API Server 负责“联动创建 SFTPGo 用户”**
+- 在 v2.5 的 `POST /api/v2/users` 成功后：
+  - API server 调用 SFTPGo 管理 API 创建同名用户
+  - 设置 home dir = `/private/users/<user_id>`
+  - 设置权限（默认可写；是否只读可配置）
+- 认证方式：
+  - v3.0 最小可用：用户名+密码（确认：v3.0 先 password；API 生成一次性密码，用户首次登录后要求改密）
+  - 或：SSH public key（WebUI 允许上传 public key，API 写入 SFTPGo）
+
+**方案 B（更强但复杂）：SFTPGo 外部认证**
+- SFTPGo 把认证委托给 API server（token/SSO），SFTP 也走内部 token。
+- 复杂度高，建议 v3.0 不做，放到 v3.5 或更后。
+
+### 4.4 用户上传/下载体验
+用户通过 SFTP 上传：
+- `datasets/...`（训练数据）
+- `models/...`（本地模型，可选）
+下载：
+- `jobs/<ray_submission_id>/...`（checkpoints/logs）
+
+WebUI/文档提供 “路径如何写进 TaskSpec” 的指引。
+
+## 5. WebUI 方案设计（最小可用）
+
+### 5.1 目标页面
+v3.0 WebUI 采用“**多子页面 + 侧边导航栏**”而不是把所有功能挤到单页：
+- 原因：信息密度更可控，后续可扩展（v3.5+）且不会把一个页面做成“巨型表单/巨型列表”。
+- 实现仍保持轻量：服务端渲染（或静态 HTML + 少量 JS），不引入复杂前端工程。
+
+信息架构（IA）建议如下：
+1) **登录页**（`/ui/login`）
+   - 用户粘贴 token（管理员发放），浏览器保存（localStorage/sessionStorage）
+   - 提供“退出登录/清空 token”
+2) **任务列表页**（`/ui/tasks`）
+   - 默认列表：最近 N 条任务（按 created_at 倒序）
+   - 支持过滤：workload、state（QUEUED/RUNNING/SUCCEEDED/FAILED/CANCELED）、时间范围
+   - 支持快捷操作：进入详情、取消任务
+3) **新建任务页**（`/ui/tasks/new`）
+   - 两种模式（二选一，均可实现）：
+     - **YAML 直接提交**：上传/粘贴 TaskSpec YAML（最省开发）
+     - **表单生成 YAML**：选择 workload，填写核心字段（train/val/model/nnodes/gpus），生成 YAML 预览后提交
+   - 提交后跳转到任务详情页
+4) **任务详情页**（`/ui/tasks/{task_id}`）
+   - 顶部：task_id、workload、state、created_at、updated_at、error_summary
+   - Attempt 卡片：latest attempt_no、ray_submission_id、ray_status、start/end
+   - 操作区：取消任务（若非 terminal）、刷新状态、复制路径/ID
+   - 链接到日志页与产物提示（SFTP 路径）
+5) **任务日志页**（`/ui/tasks/{task_id}/logs`）
+   - 默认 tail=2000，可选 200/1000/5000
+   - 提供“自动刷新（每 3~5 秒）”开关（简单轮询即可）
+6) **数据页**（`/ui/data`）
+   - 显示 SFTP 连接信息（host/port/username）
+   - 显示用户目录约定：
+     - home：`/private/users/<user_id>`
+     - datasets：`/private/users/<user_id>/datasets`
+     - models：`/private/users/<user_id>/models`
+     - jobs：`/private/users/<user_id>/jobs`
+     - trash/jobs：`/private/users/<user_id>/trash/jobs`
+   - 明确 retention：jobs 结束后 3 天移入回收站，回收站 7 天后删除；重要文件请移到 `models/` 或 `datasets/`
+7) **（仅管理员可见）用户管理页**（`/ui/admin/users`，可选但很有价值）
+   - 创建用户、禁用用户、签发 token、重置 SFTP 密码（方案 A）
+
+### 5.2 页面组织与导航（建议）
+侧边栏导航（普通用户）：
+- Tasks（列表）
+- New Task（新建）
+- Data（SFTP/目录说明）
+
+管理员侧边栏额外增加：
+- Admin / Users
+
+### 5.3 大致示意图（wireframe）
+
+下面是一个粗略示意（非最终 UI，仅表达信息结构与布局）：
+
+```
+┌──────────────────────────────────────────────────────────────────────┐
+│ Argus MVP v3.0                                      [user: alice]    │
+├───────────────┬──────────────────────────────────────────────────────┤
+│  Side Nav     │  /ui/tasks                                            │
+│               │                                                      │
+│  • Tasks      │  [Filter] workload=all  state=all  [Search task_id]  │
+│  • New Task   │                                                      │
+│  • Data       │  Task List                                           │
+│  • Admin(*)   │  ┌────────────────────────────────────────────────┐  │
+│               │  │ task_id                 workload  state   ...   │  │
+│               │  │ mvp2-alice-ppo-...      ppo       RUNNING ...   │  │
+│               │  │ mvp2-alice-sft-...      sft       SUCCEEDED...  │  │
+│               │  └────────────────────────────────────────────────┘  │
+│               │  [View] [Cancel]                                     │
+└───────────────┴──────────────────────────────────────────────────────┘
+```
+
+任务详情页（示意）：
+```
+┌──────────────────────────────────────────────────────────────────────┐
+│ /ui/tasks/{task_id}                                                   │
+├──────────────────────────────────────────────────────────────────────┤
+│ task_id: mvp2-alice-ppo-...   state: RUNNING   workload: ppo          │
+│ created_at: ...               updated_at: ...                         │
+│ error_summary: (empty)                                                │
+│                                                                        │
+│ latest_attempt: a01  ray_submission_id: ...--a01  ray_status: RUNNING  │
+│ [Open Logs] [Cancel Task] [Refresh]                                   │
+│                                                                        │
+│ Artifacts (SFTP paths):                                                │
+│   jobs/:   /private/users/alice/jobs/<ray_submission_id>/              │
+│   trash/:  /private/users/alice/trash/jobs/<ray_submission_id>/        │
+│   tip: move important files to /private/users/alice/models/            │
+└──────────────────────────────────────────────────────────────────────┘
+```
+
+### 5.2 技术取舍（建议：不引入 Node 构建）
+为了降低部署复杂度，建议 v3.0 WebUI 以 “服务端渲染 + 少量 JS/HTMX” 或 “纯静态 HTML+fetch” 实现：
+- 由 API server 提供静态资源（FastAPI StaticFiles）
+- 页面调用同源 API，避免跨域与复杂前端构建链
+
+## 6. API 扩展设计（概览）
+
+v3.0 可以保持 `/api/v2/...` 不变，增量加：
+- SFTPGo 集成管理端点（管理员）：
+  - 创建/禁用用户时联动 SFTPGo
+  - 重置 SFTP 密码 / 更新 SSH key
+- 用户数据端点（可选，最小化）：
+  - `/api/v2/me`：返回 user_id、SFTP 信息（host/port/home）
+  - `/api/v2/files`：仅用于浏览/下载（上传仍走 SFTP）
+
+详细见 `specs/mvp/v3.0/v3.0_api.md`。
+
+## 7. 配置与部署（v3.0 新增项）
+
+在 `configs/dev.yaml` 基础上扩展一组 `data` 配置（示意）：
+```yaml
+data:
+  shared_root: "/private"              # 通常与 ray.shared_root 一致
+  user_root: "/private/users"          # 用户空间根目录
+  allow_common_prefix: "/private/common/"
+  allow_user_prefix_template: "/private/users/{user_id}/"
+
+  sftpgo:
+    enabled: true
+    host: "127.0.0.1"
+    sftp_port: 2022
+    admin_api_base: "http://127.0.0.1:8081/api/v2"
+    admin_user: "admin"
+    admin_password_env: "SFTPGO_ADMIN_PASSWORD"   # 仅 head 容器内可读
+
+  retention:
+    jobs_trash_after_days: 3
+    jobs_purge_after_days: 7
+    trash_root_template: "/private/users/{user_id}/trash/jobs"
+    janitor_interval_s: 3600     # 每小时扫一次（可配置）
+```
+
+## 8. 风险点与对策
+
+1) **路径逃逸/越权读取**
+   - 必须在 API 提交任务时校验路径前缀
+   - SFTPGo 必须 chroot 到用户 home
+2) **大文件上传稳定性**
+   - 优先用 SFTP（断点续传/可靠性更好）
+3) **用户 token 与 SFTP 凭据的生命周期**
+   - token 走 v2.5 SQLite
+   - SFTP 凭据建议独立（密码/SSH key），并提供 reset 流程
+4) **GPFS/NFS 权限**
+   - 确保 `/private/users/<user>` 目录权限可被 SFTPGo 写入且 worker 可读
+
+## 9. 已确认结论（来自你的反馈）
+1) 允许用户上传并在训练时使用自定义数据集：允许（`/private/users/<u>/datasets/...`）。
+2) 允许用户上传并在训练时使用本地模型路径：允许（`/private/users/<u>/models/...`）。
+3) v3.0 不允许执行用户自定义代码（不注入 `PYTHONPATH` 作为可执行 code path）。
+4) SFTPGo 认证方式：v3.0 先 password。
+5) WebUI：按“简单最小必要功能”做（token 粘贴登录优先）。
+
+## 10. 待确认问题（需要你给结论）
+（已确认）jobs 清理执行主体：v3.0 采用 **API server 内置 janitor 后台线程**。
diff --git a/specs/mvp/v3.0/v3.0_dev_plan.md b/specs/mvp/v3.0/v3.0_dev_plan.md
new file mode 100644
index 0000000..9a14b4b
--- /dev/null
+++ b/specs/mvp/v3.0/v3.0_dev_plan.md
@@ -0,0 +1,232 @@
+# MVP v3.0 开发计划（TDD 驱动）
+
+本文是 v3.0 的**工程化开发计划**，强调“先写测试，再写实现”（TDD），并将每个里程碑拆成**可独立验收**的小闭环。
+
+输入依据：
+- 路线图：`specs/mvp/mvp_roadmap_v2.md`
+- v3.0 设计：`specs/mvp/v3.0/v3.0_design.md`
+- v3.0 API：`specs/mvp/v3.0/v3.0_api.md`
+- v3.0 验收：`specs/mvp/v3.0/v3.0_acceptance.md`
+- 现状基线：v2.5（Task queue + User mgmt + Stateless ray pool + 单镜像节点守护）
+
+v3.0 已确认约束：
+- 允许用户数据集路径：`/private/users/<me>/datasets/...`
+- 允许用户本地模型路径：`/private/users/<me>/models/...`
+- **不允许执行用户自定义代码**（不注入 user code 到 PYTHONPATH；`code_path` 仍只允许 `/private/common/...`）
+- SFTPGo 先用 **password** 方案（方案 A：API 联动创建/管理 SFTPGo 用户）
+- jobs retention：**3 天移入回收站（trash/jobs），再 7 天永久删除**；不提供 keep/延长保留标记
+- janitor：**API server 内置后台线程**；删除/移动采用**文件系统直接操作**（不依赖 SFTPGo API）
+
+---
+
+## 0. TDD 规范（所有功能都遵循）
+
+### 0.1 测试分层
+
+1) **单元测试（fast）**
+- 纯 Python 逻辑：路径策略、SFTPGo client、retention 计算、文件移动/删除策略（用临时目录）。
+- 不依赖真实 Ray、不依赖 docker、不依赖网络。
+
+2) **组件测试（中等）**
+- FastAPI 路由（含 WebUI 路由）：`fastapi.testclient.TestClient`
+- mock/stub SFTPGo client 与 ray client
+
+3) **端到端（慢）**
+- 在 `argus@h1` 通过 docker compose + scripts：
+  - Ray 集群自动起来（head+2 worker）
+  - SFTPGo 服务可用
+  - 上传数据 → 提交训练 → 下载产物 → jobs 回收站/清理
+
+### 0.2 代码与测试约定
+- 测试目录：`src/mvp/py/tests/`
+- 新功能必须先补齐测试用例，并让其在未实现时失败（红）
+- 最小实现让测试变绿（绿）
+- 再做重构（重构）
+- 覆盖率：继续沿用当前阈值（>= 90%）
+
+---
+
+## 1. 里程碑拆分（v3.0 = 5 个可验证闭环）
+
+### M1：TaskSpec 路径策略升级（允许 user datasets/models；code_path 仍仅 common）
+
+**目标**
+- API submit 时的路径校验从 v2.5 的 “仅 `/private/common/`” 升级为：
+  - `train_file` / `val_file`：允许 `/private/common/...` 与 `/private/users/<me>/...`
+  - 本地模型路径：允许 `/private/users/<me>/models/...`（不改变 YAML 结构，见实现建议）
+  - `code_path`：仍仅允许 `/private/common/...`
+- 阻止越权路径（`/private/users/other/...`）与非 `/private/...` 路径。
+
+**实现建议（不扩展 TaskSpec）**
+- `model_id` 字段保持不变：
+  - 若 `model_id` 以 `/private/` 开头 → 视作本地模型路径
+  - 否则视作 HuggingFace repo id（如 `Qwen/...`）
+
+**TDD 用例（先写测试）**
+- 单测：
+  - `test_paths_allow_common_and_own_user_prefix()`
+  - `test_paths_reject_other_user_prefix()`
+  - `test_model_id_local_path_allowed_only_under_users_models()`
+  - `test_code_path_still_common_only()`
+- API 测试：
+  - `test_submit_accepts_user_datasets_paths()`
+  - `test_submit_rejects_cross_user_paths_404_or_400()`（按约定返回 400/403）
+
+**验收点**
+- `v3.0_acceptance.md` 的 D 类安全隔离用例可由 API 测试覆盖。
+
+---
+
+### M2：SFTPGo 集成（方案 A：用户联动创建 + password）
+
+**目标**
+- 引入 `data management (SFTPGo)`：
+  - admin 创建用户时联动创建 SFTPGo 用户（home=/private/users/<user_id>，chroot）
+  - password 模式：生成一次性密码（reset/create）并返回给 admin（明文只返回一次）
+- 提供用户自助信息：
+  - `GET /api/v2/me` 返回 SFTP 连接信息、目录约定、retention 提示。
+
+**实现建议**
+- 新增 `SFTPGoAdminClient`（同步调用）：
+  - 通过 `urllib` 或 `httpx`（建议 `urllib`，减少依赖；禁止 hard-code requests 使用）
+  - 支持：create user / disable user / reset password（最小集合）
+- API server 启动时校验配置（enabled 时必须具备 admin 密码 env）。
+- 同步创建用户目录结构（文件系统）：
+  - `/private/users/<u>/{datasets,models,code,jobs,trash/jobs}`（最小必需）
+
+**TDD 用例（先写测试）**
+- 单测：
+  - `test_sftpgo_client_builds_correct_requests()`（不发真实网络；mock urlopen）
+  - `test_user_dirs_created_on_user_create()`（tmp dir 断言目录存在）
+- API 测试：
+  - `test_create_user_calls_sftpgo_client()`（stub client，断言调用参数）
+  - `test_me_returns_sftp_info_and_paths()`（含 trash/jobs 与 TTL 字段）
+
+**验收点**
+- `v3.0_acceptance.md` 的 A 类（用户/凭据）与 B 类（上传闭环前置）覆盖。
+
+---
+
+### M3：WebUI（最小可用，多页面 + 侧边栏）
+
+**目标**
+- WebUI 由 API server 托管（同源，无额外 CORS）：
+  - `/ui/login`：token 粘贴登录（localStorage）
+  - `/ui/tasks`：任务列表 + 过滤（最小）
+  - `/ui/tasks/new`：YAML 提交（优先）+（可选）表单生成 YAML
+  - `/ui/tasks/{task_id}`：详情页
+  - `/ui/tasks/{task_id}/logs`：日志 tail + 可选自动刷新
+  - `/ui/data`：SFTP 信息 + 目录/retention 提示
+  - （可选）`/ui/admin/users`：管理员用户管理（若时间允许，强烈建议）
+
+**实现建议**
+- 先不引入 Node 构建：
+  - HTML 模板可用最简单的字符串拼接或 Jinja2（若引入 jinja2，则补齐依赖与测试）
+  - 页面通过 fetch 调用 `/api/v2/...`，并复用 token header
+
+**TDD 用例（先写测试）**
+- 组件测试（TestClient）：
+  - `test_ui_routes_render_200()`
+  - `test_ui_contains_sidebar_links()`（简单断言文本包含导航链接）
+  - `test_ui_tasks_detail_shows_ids()`（包含 task_id、state、ray_submission_id）
+
+**验收点**
+- WebUI 能完成：登录→创建任务→查看任务→查看日志→看到 data 页提示。
+
+---
+
+### M4：Jobs Retention janitor（3 天移入 trash，7 天后 purge）
+
+**目标**
+- API server 内置 janitor 后台线程：
+  - 周期性扫描 DB 中 terminal tasks
+  - 到期后执行：
+    - move：`/private/users/<u>/jobs/<sid>` → `/private/users/<u>/trash/jobs/<sid>`
+    - purge：递归删除 `/private/users/<u>/trash/jobs/<sid>`
+  - 全程严格 path 校验，禁止越界删除
+  - 清理操作记录到 DB events（审计）
+
+**实现建议（数据与状态）**
+- 需要稳定的时间锚点与幂等：
+  - 使用 attempts.end_time 作为 job 结束时间（latest attempt）
+  - 在 tasks 表新增字段（或新表）记录：
+    - `trashed_at`（首次成功 move 时间）
+    - `purged_at`（成功删除时间）
+    - `trash_path`（可选）
+  - 幂等：重复运行不会报错（目录不存在视为已处理）
+
+**TDD 用例（先写测试）**
+- 单测（用 tmpdir 构造 jobs/trash 目录）：
+  - `test_janitor_moves_job_to_trash_after_threshold()`
+  - `test_janitor_purges_trash_after_threshold()`
+  - `test_janitor_never_touches_models_or_datasets()`
+  - `test_janitor_path_escape_rejected()`（恶意 path 不可删）
+- API/组件测试：
+  - `test_me_includes_retention_fields()`（jobs_trash_after_days/jobs_purge_after_days）
+
+**验收点**
+- `v3.0_acceptance.md` 的 C2 用例可按“把阈值调小到分钟级”完成验证。
+
+---
+
+### M5：端到端（h1）— SFTP 上传→训练→产物下载→回收站/清理
+
+**目标**
+- 在 `argus@h1` 落一个一键脚本（或手册）跑通：
+  1) `docker compose up -d` 拉起 Ray（head+2 worker）+ SFTPGo
+  2) admin 创建用户 alice（联动创建 SFTPGo 用户 + password）
+  3) alice 通过 SFTP 上传：
+     - 数据集到 `/private/users/alice/datasets/...`
+     - （可选）本地模型到 `/private/users/alice/models/...`
+  4) alice 通过 API/WebUI 提交任务引用上述路径
+  5) 任务成功后：
+     - 从 `jobs/<sid>` 下载 logs/checkpoints
+     - 把权重移动到 `models/`，验证不会被清理
+  6) 把 retention 配置调小，验证 jobs→trash→purge
+
+**交付建议**
+- 新增脚本（命名示例）：
+  - `scripts/run_all_v30_api.sh`
+  - `scripts/run_e2e_v30_cases.sh`
+- 新增 `docker-compose.yaml` 中的 `sftpgo` service（或 `docker-compose.v30.yaml` 叠加文件）
+
+**验收点**
+- `v3.0_acceptance.md` 全部 MUST 用例通过。
+
+---
+
+## 2. 风险与测试关注点
+
+1) **权限与路径逃逸**
+- path policy 必须覆盖：train/val/model_id(local)/output dirs（jobs/trash）
+- 所有删除/移动必须做 prefix 校验
+
+2) **并发与竞态**
+- janitor 只处理 terminal tasks，避免清理正在写入的目录
+- move 使用同文件系统 `os.replace`（原子）
+
+3) **SFTPGo 可用性**
+- SFTPGo 不在线不应影响训练与 API 核心功能（除了用户创建联动）
+- janitor 不依赖 SFTPGo（文件系统直连）
+
+---
+
+## 3. 交付清单（代码/配置/脚本/文档）
+
+### 3.1 代码
+- Path policy（v3.0）
+- SFTPGoAdminClient + user create/disable/reset password 联动
+- `/api/v2/me` 扩展（SFTP/目录/retention）
+- WebUI 路由与静态资源
+- janitor（trash+purge）后台线程 + DB 记录
+
+### 3.2 配置
+- `configs/dev.yaml` 增加 `data.sftpgo`、`data.retention` 段（详见设计文档）
+
+### 3.3 scripts / compose
+- compose 增加 `sftpgo`（或新增 overlay compose 文件）
+- v3.0 e2e 脚本（上传/下载/清理验证）
+
+### 3.4 文档
+- 更新 `specs/mvp/v3.0/*` 与 `src/mvp/README.md`（运行方式、路径约定、SFTP 操作、retention 解释）
+
diff --git a/specs/mvp/v3.0/v3.0_progress.md b/specs/mvp/v3.0/v3.0_progress.md
new file mode 100644
index 0000000..49ebfe6
--- /dev/null
+++ b/specs/mvp/v3.0/v3.0_progress.md
@@ -0,0 +1,154 @@
+# MVP v3.0 进展记录（milestone log）
+
+本文档用于记录 v3.0 按 `specs/mvp/v3.0/v3.0_dev_plan.md` 实施过程中的里程碑完成情况。  
+约定：每完成一个里程碑，追加一条记录，包含**日期**、**完成内容**、**涉及文件**、**验证方式/结果**、**待办/风险**。
+
+---
+
+## M1：Path policy + tests（已完成）
+
+- 日期：2025-12-30
+- 范围：按 v3.0 路径策略升级 API submit 的路径校验（不扩展 TaskSpec YAML 结构）。
+- 完成内容：
+  - `code_path`：仍只允许 `/private/common/...`（v3.0 不执行 user code）。
+  - `train_file`/`val_file`：允许 `/private/common/datasets/...` 或 `/private/users/<me>/datasets/...`。
+  - `model_id`：若以 `/private/` 开头则视为本地路径，仅允许：
+    - `/private/common/models/...` 或
+    - `/private/users/<me>/models/...`
+    否则仍按 HuggingFace repo id（如 `Qwen/...`）处理。
+  - 拒绝跨用户路径（例如 `bob` 提交 `/private/users/alice/datasets/...`）。
+  - 拒绝本地模型路径不在 `models/`（例如指向 `jobs/`）。
+- 涉及文件：
+  - `src/mvp/py/argus/service/app.py`
+  - `src/mvp/py/tests/test_users.py`
+- 验证方式与结果：
+  - 本地单测：`.venv/bin/python -m pytest -q`
+  - 结果：全部通过（`54 passed`），覆盖率阈值保持 `>= 90%`。
+- 待办/风险：
+  - `model_id=/private/...` 的“本地模型路径语义”需要在用户文档/WebUI 中明确提示（避免误用）。
+  - 后续 M2/M3 需要把该路径策略同步到 UI 表单/提示文本（避免用户填错路径）。
+
+---
+
+## M2：SFTPGo 集成（方案 A：用户联动创建 + password）（已完成）
+
+- 日期：2025-12-30
+- 范围：SFTPGo（Data Management）最小集成 + 用户自助信息 `/api/v2/me` + 用户目录结构落盘。
+- 完成内容：
+  - 新增 `data` 配置段：
+    - `data.user_root`：用户数据根目录（默认 `/private/users`）
+    - `data.sftpgo`：SFTPGo 可选联动（enabled/host/sftp_port/admin_api_base/admin_user/admin_password_env）
+    - `data.retention`：jobs 过期策略配置（3 天移入 trash，7 天 purge；janitor 在 M4 实现）
+  - 新增 `SFTPGoAdminClient`（`urllib` 实现，不使用 `requests`）：
+    - `create_user` / `disable_user` / `reset_password`（最小集合）
+  - API server 增强：
+    - `POST /api/v2/users`：创建 DB user + 同步创建目录结构（`datasets/models/code/jobs/trash/jobs`）
+    - 当 `data.sftpgo.enabled=true` 时，创建用户会联动调用 SFTPGo admin API，并返回一次性密码（明文仅返回一次，服务端不保存）
+    - `POST /api/v2/users/{user_id}:disable`：禁用用户（SFTPGo 禁用 best-effort）
+    - `POST /api/v2/users/{user_id}/sftp:reset_password`：管理员重置一次性密码（SFTPGo enabled 才允许）
+    - `GET /api/v2/me`：返回当前用户的目录约定、retention 提示，以及（可选）SFTP 连接信息
+  - 同步更新 `src/mvp/configs/dev.yaml`：补齐 v3.0 相关 `data.*` 配置（默认关闭 sftpgo）。
+- 涉及文件：
+  - `src/mvp/py/argus/service/config.py`
+  - `src/mvp/py/argus/service/sftpgo.py`
+  - `src/mvp/py/argus/service/app.py`
+  - `src/mvp/py/tests/test_sftpgo.py`
+  - `src/mvp/py/tests/test_users.py`
+  - `src/mvp/py/tests/test_app.py`
+  - `src/mvp/py/tests/test_service_config.py`
+  - `src/mvp/configs/dev.yaml`
+  - `specs/mvp/v3.0/v3.0_api.md`
+- 验证方式与结果：
+  - 本地单测：`.venv/bin/python -m pytest -q`
+  - 结果：全部通过（`62 passed`），覆盖率 `90.11%`（阈值 `>= 90%`）。
+- 待办/风险：
+  - M2 仅做了“API 侧联动 + 单测”，未在真实 SFTPGo 容器上端到端验证（按计划在 M5 完成）。
+  - 目录创建依赖文件系统权限：生产部署时需确保 API/head 容器对 `/private/users` 可写。
+
+---
+
+## M3：WebUI（最小可用，多页面 + 侧边栏）（已完成）
+
+- 日期：2025-12-30
+- 范围：API server 托管最小 WebUI（同源，不引入 Node 构建），用于登录/提交/查看任务与日志、查看 data 信息。
+- 完成内容：
+  - 新增 UI 路由（HTML+少量 JS）：
+    - `/ui`（重定向到 tasks）
+    - `/ui/login`：token 粘贴并写入浏览器 localStorage（key=`mvp_token`）
+    - `/ui/tasks`：任务队列列表（调用 `/api/v2/queue`）
+    - `/ui/tasks/new`：提交 TaskSpec YAML（POST `/api/v2/tasks`）
+    - `/ui/tasks/{task_id}`：任务详情（GET `/api/v2/tasks/{task_id}`，支持 cancel）
+    - `/ui/tasks/{task_id}/logs`：日志查看（GET `/api/v2/tasks/{task_id}/logs`，可选自动刷新）
+    - `/ui/data`：展示 `/api/v2/me` 返回的路径/SFTP/retention 信息
+  - 统一侧边栏导航：Tasks / New Task / Data / Login。
+  - UI 不做服务端 session：所有 API 调用均由浏览器带 `Authorization: Bearer <token>`（localStorage 注入）。
+- 涉及文件：
+  - `src/mvp/py/argus/service/ui.py`
+  - `src/mvp/py/argus/service/app.py`
+  - `src/mvp/py/tests/test_ui.py`
+- 验证方式与结果：
+  - 本地单测：`.venv/bin/python -m pytest -q`
+  - 结果：全部通过（`65 passed`），覆盖率 `90.53%`（阈值 `>= 90%`）。
+- 待办/风险：
+  - WebUI 当前为“骨架+API 驱动”，不做复杂交互与大文件下载；上传/下载仍以 SFTP 为主（按设计）。
+  - Starlette TestClient 的 `allow_redirects` 有弃用告警（不影响功能，可在后续清理）。
+
+---
+
+## M4：Jobs Retention janitor（3 天移入 trash，7 天后 purge）（已完成）
+
+- 日期：2025-12-30
+- 范围：API server 内置后台线程，对“已结束 attempt”的 job 目录执行保留策略（文件系统直连，不依赖 SFTPGo）。
+- 完成内容：
+  - 新增 `JobsJanitor`：
+    - 以 `attempts.end_time` 为基准计算 TTL（从 job 结束开始算）
+    - `>= 3 天 && < 7 天`：把目录从 `.../jobs/<ray_submission_id>` 移动到 `.../trash/jobs/<ray_submission_id>`
+    - `>= 7 天`：确保目录进入 trash 后删除（`shutil.rmtree`）
+    - 对缺失目录、异常移动/删除为 best-effort（不影响服务主流程）
+  - DB 增强：新增查询 `list_ended_attempts_before()`，用于 janitor 扫描候选 attempt。
+  - API server 启动时启动 janitor 线程（可通过 `data.retention.janitor_interval_s` 控制；<=0 视为关闭）。
+- 涉及文件：
+  - `src/mvp/py/argus/service/janitor.py`
+  - `src/mvp/py/argus/service/db.py`
+  - `src/mvp/py/argus/service/app.py`
+  - `src/mvp/py/tests/test_janitor.py`
+- 验证方式与结果：
+  - 本地单测：`.venv/bin/python -m pytest -q`
+  - 结果：全部通过（`75 passed`），覆盖率 `90.72%`（阈值 `>= 90%`）。
+- 待办/风险：
+  - M4 只做“逻辑 + 单测”，实际 `/private/users/...` 的权限与在 `argus@h1` 的行为验证放到 M5（端到端）。
+
+---
+
+## M5：端到端（h1）— SFTPGo compose + v3.0 E2E 脚本（已完成：交付脚本/配置）
+
+- 日期：2025-12-30
+- 范围：补齐 h1 端到端所需的 compose/service、配置与一键脚本（实际运行/验收由你在 `argus@h1` 执行）。
+- 完成内容：
+  - SFTPGo 集成到 `docker compose`：
+    - 新增 `argus-sftpgo` service（SFTP 2022；Admin API/UI 8080→host 8081，避免与 MVP API 8080 冲突）
+    - 同挂载 `../../shared:/private`，并持久化元数据到 `../../shared/common/sftpgo`
+  - SFTPGoAdminClient 实装（对齐 upstream OpenAPI）：
+    - `GET /api/v2/token`（BasicAuth）获取 admin token
+    - `POST /api/v2/users` 创建用户（含 `permissions: {"/":["*"]}`）
+    - `PUT /api/v2/users/{username}` 禁用/重置密码
+  - 新增 v3.0 dev 配置：`configs/dev_v30.yaml`（启用 `data.sftpgo` 并配置 `admin_api_base=http://argus-sftpgo:8080/api/v2`）
+  - 新增 v3.0 一键脚本：
+    - `scripts/run_all_v30_api.sh`：起 Ray+SFTPGo、启动 API、创建用户并提交 PPO/GRPO/SFT（引用 user dataset 路径）
+    - `scripts/run_e2e_v30_cases.sh`：最小 E2E runner（HP-1）
+  - API 启动脚本增强：`scripts/60_start_api.sh` 支持透传 `SFTPGO_ADMIN_PASSWORD` 到 head 容器内的 API 进程。
+- 涉及文件：
+  - `src/mvp/docker-compose.yaml`
+  - `src/mvp/configs/dev_v30.yaml`
+  - `src/mvp/scripts/run_all_v30_api.sh`
+  - `src/mvp/scripts/run_e2e_v30_cases.sh`
+  - `src/mvp/scripts/60_start_api.sh`
+  - `src/mvp/py/argus/service/sftpgo.py`
+  - `src/mvp/py/tests/test_sftpgo.py`
+  - `src/mvp/README.md`
+  - `specs/mvp/v3.0/v3.0_api.md`
+- 验证方式与结果：
+  - 本地单测：`.venv/bin/python -m pytest -q`
+  - 结果：全部通过（`75 passed`），覆盖率 `90.35%`（阈值 `>= 90%`）。
+- 待办/风险：
+  - 需要你在 `argus@h1` 实跑 `scripts/run_all_v30_api.sh` 完成真正的 SFTP 上传/下载与 retention 验收（按 `v3.0_acceptance.md`）。
diff --git a/specs/mvp/v3.0/v3.0_summary.md b/specs/mvp/v3.0/v3.0_summary.md
new file mode 100644
index 0000000..c952025
--- /dev/null
+++ b/specs/mvp/v3.0/v3.0_summary.md
@@ -0,0 +1,166 @@
+# MVP v3.0 迭代总结（Ray + SFTPGo + API + WebUI）
+
+本文总结 v3.0 迭代最终落地的功能、架构、运行方式、验收点与已知限制，便于后续评审、交接与继续迭代。
+
+相关更详细文档：
+- `specs/mvp/v3.0/v3.0_design.md`
+- `specs/mvp/v3.0/v3.0_api.md`
+- `specs/mvp/v3.0/v3.0_dev_plan.md`
+- `specs/mvp/v3.0/v3.0_acceptance.md`
+- `specs/mvp/v3.0/v3.0_progress.md`
+
+---
+
+## 1. 目标与范围
+
+v3.0 作为“第一版可发布”的最小闭环，主要新增：
+- **WebUI**：最小可用的人机界面（登录、任务提交与查看、数据入口、管理员入口）。
+- **用户管理**：基于内部 token 的用户体系（admin 与普通用户），支持创建用户与签发 token。
+- **数据管理入口（SFTPGo）**：用户通过 SFTP/WebClient 上传下载自己的数据；同时暴露只读的共享数据/缓存目录（common）用于复用。
+- **保持训练闭环**：仍通过 Ray Job 提交到集群执行（PPO/GRPO/SFT 三类 workload 都验证）。
+
+明确不做（本迭代保持最小）：
+- 不支持用户自定义训练代码（TaskSpec 的 `code_path` 固定走 common 下的 verl snapshot 策略）。
+- 不做复杂资源排队优化/多集群/多租隔离策略（目前隔离粒度主要在用户 jobs 目录层）。
+
+---
+
+## 2. 系统架构（最终形态）
+
+核心组件：
+- **Ray 集群（容器）**
+  - `argus-ray-head`：head 节点（无 GPU/不跑训练），提供 Ray Dashboard 与 Job Server。
+  - `argus-ray-worker-0/1`：worker 节点（有 GPU），承载训练任务。
+  - worker 以 “stateless + watchdog 自动连接 head” 的方式加入集群。
+- **API Server（运行在 head 容器内）**
+  - 读取 YAML 配置（dev/prod），维护任务队列（sqlite），并周期性调度将任务提交到 Ray。
+  - 同时承载 WebUI（`/ui`）。
+- **SFTPGo（容器）**
+  - 提供 SFTP（端口 `2022`）与 Web Client/Admin（端口 `8081` 映射到容器 8080）。
+  - 用户 home 为 `/private/users/<user>`，默认可读写。
+  - 额外提供 `/common/*` 共享只读入口（见第 4 节）。
+- **共享存储（NFS/GPFS 等挂载到容器内 `/private`）**
+  - `/private/common`：共享缓存（hf、datasets、models、db、logs 等）。
+  - `/private/users/<user>`：用户隔离目录（jobs/datasets/models/code/trash 等）。
+
+---
+
+## 3. 任务与调度（Task / Ray Job）
+
+### 3.1 Task（平台概念）
+- 用户向 API 提交 TaskSpec（YAML），平台分配 `task_id`（可读、包含用户名）。
+- `task_id` 对应内部状态机与重试逻辑；底层每次提交 Ray Job 会产生 attempt 与 `ray_submission_id`。
+
+### 3.2 Ray Job（Ray 概念）
+- 真正执行训练的 driver 通过 Ray Job 运行在集群 worker 上（避免 head 承载训练）。
+- head 节点通过 `--num-cpus=0` / 自定义资源等策略避免调度到 head。
+
+### 3.3 VERL 资源预检查的处理
+- VERL 在创建资源池时会做 fail-fast 资源预检查（如“可用 GPU 不足”直接报错退出）。
+- v3.0 延续 v2.x 的策略：服务端识别失败原因并按策略重试/回退（具体见 scheduler 实现与 v2.5/3.0 文档）。
+
+---
+
+## 4. 数据管理（SFTPGo）与 common 只读目录
+
+### 4.1 用户目录（读写）
+- 用户通过 SFTP/WebClient 访问自己的 home：`/private/users/<user>`
+- 目录结构（至少）：`datasets/ models/ code/ jobs/ trash/ common/`
+
+### 4.2 common 只读（方案 A：Virtual Folder）
+本迭代采用 SFTPGo 的 Virtual Folder + 路径权限覆盖，实现用户可读共享目录但不可写。
+
+最终对外暴露为：
+- `/common/datasets`（只读）
+  - **mapped_path 指向真实目录 `/private/datasets`**（避免 `/private/common/datasets` 中大量 symlink 导致的 WebClient “权限不足/越界”问题）
+- `/common/hf`（只读）
+  - mapped_path 指向 `/private/hf`
+
+备注：
+- `/private/common/datasets` 内部存在 symlink（如 `gsm8k -> /private/datasets/gsm8k`），如果虚拟目录映射到 symlink 根目录，SFTPGo 会把 symlink 跳转视为“逃逸 root”，导致点击进入时报权限不足；因此选择直接映射到真实目录根。
+
+---
+
+## 5. WebUI（最小可用）
+
+入口：
+- `/ui/login`：粘贴 token（存 browser `localStorage`）
+- `/ui/tasks`：任务列表（Running/Pending/Completed），Completed 支持分页
+- `/ui/tasks/new`：提交任务（PPO/GRPO/SFT 三套样例可一键填充）
+- `/ui/data`：展示当前用户名、支持重置 SFTPGo 密码并复制；提供跳转到 SFTPGo WebClient；提示 FileZilla 等客户端用法
+- `/ui/admin`：管理员入口（创建用户、签发 token、用户列表）
+- 导航栏提供 Ray Dashboard 快捷跳转（当前 IP 的 `:8265`）
+
+关于 admin 页面权限：
+- admin 页面本身可访问，但其数据请求必须携带 admin token；否则会在页面内显示 401/403/错误信息（满足“需要先提供 admin token 才能看到内容”）。
+
+---
+
+## 6. API（v3.0 新增/强化点）
+
+核心接口（节选）：
+- 认证：
+  - Bearer token：`MVP_INTERNAL_TOKEN`（admin）或用户 token（由 admin 签发）
+- 用户管理（admin）：
+  - `POST /api/v2/users` 创建用户（并初始化用户目录）
+  - `GET /api/v2/users` 获取用户列表（包含最新 token、创建/更新时间等）
+  - `POST /api/v2/users/{user_id}/tokens` 签发用户 token
+- 任务：
+  - `POST /api/v2/tasks` 提交 TaskSpec（YAML）
+  - `GET /api/v2/tasks` 任务列表（支持 states/limit/offset，用于 Completed 分页）
+  - `GET /api/v2/tasks/{task_id}`、`POST /api/v2/tasks/{task_id}:cancel`、`GET /api/v2/tasks/{task_id}/logs`
+  - `GET /api/v2/queue`（运行中/待调度概览）
+- 数据/SFTP：
+  - `GET /api/v2/me` 返回用户路径信息、SFTP 连接信息，并 best-effort 对齐 SFTPGo 用户配置
+  - `POST /api/v2/me/sftp:reset_password` 用户自助重置 SFTPGo 密码（一次性返回明文）
+
+安全取舍说明（当前为内网/开发优先）：
+- 为了 Admin WebUI “可查看并复制 token”，数据库持久化存储了 `token_plain`（明文 token）。
+  - 这在生产场景通常不建议；未来可改为只展示“重置/重新签发”而不回显明文，或只回显一次。
+
+---
+
+## 7. 持久化与清理
+
+- 任务队列：sqlite（WAL 模式）
+- SFTPGo：自带 sqlite db（容器挂载持久化目录）
+- Jobs 目录清理策略（服务端 janitor）：
+  - job 结束后 3 天移动到回收目录（trash）
+  - 回收目录再保留 7 天后删除
+
+---
+
+## 8. 运行方式与脚本
+
+开发/验收脚本：
+- `src/mvp/scripts/run_all_v30_api.sh`：端到端拉起（Ray + SFTPGo + API），并通过 API 提交 PPO/GRPO/SFT，等待完成并验收
+- 其他脚本用于启动/停止 API、准备数据与模型、探测服务就绪等（详见 scripts 目录与 README）
+
+典型端到端（示例参数）：
+- `MVP_INTERNAL_TOKEN=my-dev-token`
+- `SFTPGO_ADMIN_PASSWORD=my-dev-sftpgo-admin`
+- 支持 `RESET_DB/RESET_SFTPGO` 用于测试环境重置
+
+---
+
+## 9. 验证结果（已跑通）
+
+在 `argus@h1` 环境中已完成端到端验证：
+- Ray 集群可用（head + 2 worker）
+- API server + WebUI 可用
+- SFTPGo（admin + 普通用户）可用
+- 通过 API 连续提交 PPO/GRPO/SFT 三种任务均能完成（SUCCEEDED）
+- 用户可以登录 SFTPGo WebClient/SFTP，访问自己的目录，并访问 `/common/datasets`、`/common/hf` 的只读内容
+
+同时本地单测通过：
+- pytest 全绿
+- 覆盖率阈值 >= 90%
+
+---
+
+## 10. 已知限制 & 后续可改进
+
+- WebUI 当前为最小版，交互与权限提示仍偏“工程化”而非产品化（后续可增强错误提示、搜索筛选、任务详情聚合等）。
+- token 明文持久化仅适合内网/开发场景；生产建议改为一次性展示或支持撤销/轮换策略。
+- SFTPGo 虚拟目录目前保留了历史遗留映射（例如 `/common/models` 可能残留），后续可在升级脚本中做一次性清理与迁移。
+
diff --git a/src/mvp/README.md b/src/mvp/README.md
index 9758140..4fd720e 100644
--- a/src/mvp/README.md
+++ b/src/mvp/README.md
@@ -16,3 +16,11 @@
 - CLI 提交流程：`scripts/run_all_cli.sh`
 - API 提交流程：`scripts/run_all_api.sh`
 - v2.5（Stateless worker + user 隔离 jobs）E2E：`scripts/run_all_v25_api.sh`
+- v3.0（WebUI + SFTPGo + user datasets/models + jobs retention）E2E：`scripts/run_all_v30_api.sh`
+
+v3.0 访问入口（dev/h1）：
+- WebUI：`http://127.0.0.1:8080/ui`
+- Ray Dashboard：`http://127.0.0.1:8265`
+- SFTPGo：
+  - SFTP：`127.0.0.1:2022`
+  - Admin API/UI：`http://127.0.0.1:8081`（容器内 8080，host 映射到 8081 避免与 API server 冲突）
diff --git a/src/mvp/configs/dev.yaml b/src/mvp/configs/dev.yaml
index d117f65..7d0f98e 100644
--- a/src/mvp/configs/dev.yaml
+++ b/src/mvp/configs/dev.yaml
@@ -32,3 +32,24 @@ service:
     retry_interval_s: 60
     max_running_tasks: 1
 
+# v3.0: user data management (filesystem + SFTPGo)
+data:
+  # All user writable data is placed under this root:
+  #   /private/users/<user_id>/{datasets,models,code,jobs,trash/jobs}
+  user_root: "/private/users"
+
+  # SFTPGo is optional in dev; when enabled, admin endpoints will call SFTPGo admin API.
+  # Admin password is provided by env var `data.sftpgo.admin_password_env`.
+  sftpgo:
+    enabled: false
+    host: ""              # shown to users via GET /api/v2/me
+    sftp_port: 2022
+    admin_api_base: ""    # e.g. http://argus-sftpgo:8080
+    admin_user: "admin"
+    admin_password_env: "SFTPGO_ADMIN_PASSWORD"
+
+  # Jobs retention policy (v3.0 janitor): move to trash after 3d, purge after 7d.
+  retention:
+    jobs_trash_after_days: 3
+    jobs_purge_after_days: 7
+    janitor_interval_s: 3600
diff --git a/src/mvp/configs/dev_v30.yaml b/src/mvp/configs/dev_v30.yaml
new file mode 100644
index 0000000..3c2649d
--- /dev/null
+++ b/src/mvp/configs/dev_v30.yaml
@@ -0,0 +1,51 @@
+ray:
+  # Ray Job server address (head 容器内视角)
+  address: "http://127.0.0.1:8265"
+
+  # 共享根路径（容器内统一 /private，对齐生产）
+  shared_root: "/private"
+
+  # 强制 driver 落 worker（head 不跑训练）
+  entrypoint_num_cpus: 1
+  entrypoint_resources:
+    worker_node: 1
+
+  # 所有 job 通用 runtime env
+  runtime_env:
+    env_vars:
+      HF_ENDPOINT: "https://hf-mirror.com"
+      PYTHONUNBUFFERED: "1"
+
+  # v3.0 先不支持 user code 执行
+  user_code_path: "/private/user/code"
+
+service:
+  api:
+    host: "0.0.0.0"
+    port: 8080
+  auth:
+    token_env: "MVP_INTERNAL_TOKEN"
+  sqlite:
+    db_path: "/private/common/db/mvp.sqlite3"
+  scheduler:
+    tick_s: 5
+    retry_interval_s: 60
+    max_running_tasks: 1
+
+data:
+  user_root: "/private/users"
+  sftpgo:
+    enabled: true
+    # Returned to users by GET /api/v2/me. For h1 E2E, usually connect to the host IP.
+    host: "127.0.0.1"
+    sftp_port: 2022
+    # Admin API base should include /api/v2 (SFTPGo OpenAPI server base).
+    # From head container, access SFTPGo by service name on the compose network.
+    admin_api_base: "http://argus-sftpgo:8080/api/v2"
+    admin_user: "admin"
+    admin_password_env: "SFTPGO_ADMIN_PASSWORD"
+  retention:
+    jobs_trash_after_days: 3
+    jobs_purge_after_days: 7
+    janitor_interval_s: 3600
+
diff --git a/src/mvp/docker-compose.yaml b/src/mvp/docker-compose.yaml
index a74c1a3..6ed9194 100644
--- a/src/mvp/docker-compose.yaml
+++ b/src/mvp/docker-compose.yaml
@@ -38,6 +38,36 @@ services:
       HF_ENDPOINT: "https://hf-mirror.com"
       PYTHONUNBUFFERED: "1"
 
+  # v3.0: Data management service (SFTPGo).
+  # - SFTP: 2022
+  # - Admin API/UI: 8080 (mapped to host 8081 to avoid collision with MVP API server on 8080)
+  #
+  # NOTE: This is for dev / h1 E2E. In prod you may use an internal mirror image/tag and different ports.
+  sftpgo:
+    image: drakkan/sftpgo:latest
+    container_name: argus-sftpgo
+    user: "0:0"
+    ports:
+      - "2022:2022"
+      - "8081:8080"
+    volumes:
+      - ../../shared:/private
+      - ../../shared/common/sftpgo:/var/lib/sftpgo
+    networks:
+      - argus-ray-net
+    environment:
+      # Create a default admin on first start (used by API server to manage users).
+      # Override on host as needed:
+      #   export SFTPGO_ADMIN_PASSWORD=...
+      SFTPGO_DATA_PROVIDER__CREATE_DEFAULT_ADMIN: "true"
+      # Persist the sqlite DB under the mounted metadata dir; otherwise it defaults to a relative path.
+      SFTPGO_DATA_PROVIDER__NAME: "/var/lib/sftpgo/sftpgo.db"
+      SFTPGO_DEFAULT_ADMIN_USERNAME: "admin"
+      SFTPGO_DEFAULT_ADMIN_PASSWORD: "${SFTPGO_ADMIN_PASSWORD:-my-dev-sftpgo-admin}"
+      # Explicitly pin default ports via env-var schema (double-underscore for nesting).
+      SFTPGO_HTTPD__BINDINGS__0__PORT: "8080"
+      SFTPGO_SFTPD__BINDINGS__0__PORT: "2022"
+
   ray_worker_0:
     image: argus/argus-ray-node:v2.5
     container_name: argus-ray-worker-0
diff --git a/src/mvp/py/argus/service/app.py b/src/mvp/py/argus/service/app.py
index 0d7c2db..b09ee5c 100644
--- a/src/mvp/py/argus/service/app.py
+++ b/src/mvp/py/argus/service/app.py
@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import os
+import secrets
 import threading
 from typing import Any
 
@@ -12,7 +13,10 @@ from argus.ray.models import JobSpec, RayConfig
 
 from .config import V2Config
 from .db import Db
+from .janitor import JobsJanitor
 from .scheduler import Scheduler
+from .sftpgo import SFTPGoAdminClient, SFTPGoError
+from .ui import register_ui_routes
 
 
 def _utc_now_iso() -> str:
@@ -38,11 +42,48 @@ def create_app(config_path: str) -> FastAPI:
     db.init()
 
     scheduler = Scheduler(db=db, ray_cfg=ray_cfg, v2_cfg=v2_cfg)
+    janitor = JobsJanitor(
+        db=db,
+        user_root=v2_cfg.data.user_root,
+        trash_after_days=v2_cfg.data.retention.jobs_trash_after_days,
+        purge_after_days=v2_cfg.data.retention.jobs_purge_after_days,
+        interval_s=v2_cfg.data.retention.janitor_interval_s,
+    )
     stop_flag = threading.Event()
     tool = scheduler.tool
 
     app = FastAPI(title="mvp-v2", version="2.0")
 
+    def _user_home(user_id: str) -> str:
+        base = v2_cfg.data.user_root.rstrip("/")
+        return f"{base}/{user_id}"
+
+    def _ensure_user_dirs(user_id: str) -> None:
+        home = _user_home(user_id)
+        # Note: create a physical /common directory under each user's home so that
+        # SFTPGo WebClient shows a "common" entry at the root. The actual shared
+        # content is exposed via SFTPGo virtual folders under /common/{datasets,models}.
+        for rel in ("datasets", "models", "code", "jobs", "trash/jobs", "common"):
+            os.makedirs(f"{home}/{rel}", exist_ok=True)
+
+    def _sftpgo_enabled() -> bool:
+        return bool(v2_cfg.data.sftpgo.enabled)
+
+    def _sftpgo_client() -> SFTPGoAdminClient:
+        cfg = v2_cfg.data.sftpgo
+        if not cfg.admin_api_base:
+            raise HTTPException(status_code=500, detail="sftpgo enabled but data.sftpgo.admin_api_base is empty")
+        pw = os.environ.get(cfg.admin_password_env, "")
+        if not pw:
+            raise HTTPException(status_code=500, detail=f"missing env: {cfg.admin_password_env}")
+        shared_root = ray_cfg.shared_root.rstrip("/")
+        return SFTPGoAdminClient(
+            admin_api_base=str(cfg.admin_api_base),
+            admin_user=str(cfg.admin_user),
+            admin_password=pw,
+            common_root=f"{shared_root}/common",
+        )
+
     def _auth(req: Request) -> dict[str, Any]:
         token_env = v2_cfg.auth.token_env
         admin_token = os.environ.get(token_env, "")
@@ -74,6 +115,9 @@ def create_app(config_path: str) -> FastAPI:
     def _startup() -> None:
         t = threading.Thread(target=scheduler.run_forever, args=(stop_flag,), daemon=True)
         t.start()
+        if int(janitor.interval_s) > 0:
+            tj = threading.Thread(target=janitor.run_forever, args=(stop_flag,), daemon=True)
+            tj.start()
 
     @app.on_event("shutdown")
     def _shutdown() -> None:
@@ -93,7 +137,42 @@ def create_app(config_path: str) -> FastAPI:
             row = db.create_user(user_id=user_id, display_name=str(display_name) if display_name is not None else None)
         except Exception as e:
             raise HTTPException(status_code=409, detail=f"user create failed: {e!r}")
-        return {"user_id": row.get("user_id", user_id), "state": row.get("state", "ACTIVE")}
+
+        # v3.0: create user dir structure (datasets/models/code/jobs/trash).
+        try:
+            _ensure_user_dirs(user_id)
+        except Exception as e:
+            raise HTTPException(status_code=500, detail=f"failed to create user dirs: {e!r}")
+
+        out: dict[str, Any] = {"user_id": row.get("user_id", user_id), "state": row.get("state", "ACTIVE")}
+
+        # v3.0: scheme A (password) SFTPGo integration.
+        if _sftpgo_enabled():
+            pw = secrets.token_urlsafe(12)
+            try:
+                _sftpgo_client().create_user(username=user_id, password=pw, home_dir=_user_home(user_id))
+            except SFTPGoError as e:
+                # Make create user idempotent for retries:
+                # If the user already exists in SFTPGo, reset password and enable it.
+                if "http error: 409" in str(e):
+                    try:
+                        _sftpgo_client().reset_password(username=user_id, new_password=pw, home_dir=_user_home(user_id))
+                        _sftpgo_client().enable_user(username=user_id, home_dir=_user_home(user_id))
+                    except Exception as e2:
+                        raise HTTPException(status_code=502, detail=f"sftpgo upsert user failed: {e2!r}")
+                else:
+                    raise HTTPException(status_code=502, detail=f"sftpgo create user failed: {e!r}")
+            except HTTPException:
+                raise
+            except Exception as e:
+                raise HTTPException(status_code=502, detail=f"sftpgo create user failed: {e!r}")
+            out["sftp"] = {
+                "username": user_id,
+                "password": pw,  # one-time return to admin; do not persist plaintext
+                "host": v2_cfg.data.sftpgo.host,
+                "port": int(v2_cfg.data.sftpgo.sftp_port),
+            }
+        return out
 
     @app.post("/api/v2/users/{user_id}/tokens")
     async def issue_token(user_id: str, req: Request) -> dict[str, Any]:
@@ -104,6 +183,12 @@ def create_app(config_path: str) -> FastAPI:
         token = db.issue_token(user_id=user_id)
         return {"user_id": user_id, "token": token}
 
+    @app.get("/api/v2/users")
+    async def list_users(req: Request, limit: int = 200) -> dict[str, Any]:
+        _require_admin(req)
+        lim = max(1, min(int(limit), 1000))
+        return {"users": db.list_users(limit=lim), "limit": lim}
+
     @app.post("/api/v2/users/{user_id}:disable")
     async def disable_user(user_id: str, req: Request) -> dict[str, Any]:
         _require_admin(req)
@@ -111,8 +196,122 @@ def create_app(config_path: str) -> FastAPI:
         if not u:
             raise HTTPException(status_code=404, detail="user not found")
         db.disable_user(user_id=user_id)
+        if _sftpgo_enabled():
+            try:
+                _sftpgo_client().disable_user(username=user_id, home_dir=_user_home(user_id))
+            except Exception:
+                # Best-effort: DB state is source of truth for API auth; SFTPGo sync can lag.
+                pass
         return {"user_id": user_id, "state": "DISABLED"}
 
+    @app.post("/api/v2/users/{user_id}/sftp:reset_password")
+    async def reset_sftp_password(user_id: str, req: Request) -> dict[str, Any]:
+        _require_admin(req)
+        u = db.get_user(user_id)
+        if not u:
+            raise HTTPException(status_code=404, detail="user not found")
+        if not _sftpgo_enabled():
+            raise HTTPException(status_code=400, detail="sftpgo is not enabled")
+        pw = secrets.token_urlsafe(12)
+        try:
+            _sftpgo_client().reset_password(username=user_id, new_password=pw, home_dir=_user_home(user_id))
+        except HTTPException:
+            raise
+        except Exception as e:
+            raise HTTPException(status_code=502, detail=f"sftpgo reset password failed: {e!r}")
+        return {"user_id": user_id, "password": pw}
+
+    @app.post("/api/v2/me/sftp:reset_password")
+    async def reset_my_sftp_password(req: Request) -> dict[str, Any]:
+        """
+        v3.0 WebUI: allow a user to rotate their own SFTPGo password and copy it.
+        Note: SFTPGo does not allow reading the current password, so this endpoint returns a new one-time password.
+        """
+        subject = _auth(req)
+        user_id = str(subject["user_id"])
+        u = db.get_user(user_id)
+        if not u:
+            raise HTTPException(status_code=404, detail="user not found")
+        if not _sftpgo_enabled():
+            raise HTTPException(status_code=400, detail="sftpgo is not enabled")
+        pw = secrets.token_urlsafe(12)
+        try:
+            _sftpgo_client().reset_password(username=user_id, new_password=pw, home_dir=_user_home(user_id))
+            _sftpgo_client().enable_user(username=user_id, home_dir=_user_home(user_id))
+        except Exception as e:
+            raise HTTPException(status_code=502, detail=f"sftpgo reset password failed: {e!r}")
+        return {"user_id": user_id, "password": pw}
+
+    @app.get("/api/v2/me")
+    async def me(req: Request) -> dict[str, Any]:
+        subject = _auth(req)
+        user_id = str(subject["user_id"])
+        try:
+            _ensure_user_dirs(user_id)
+        except Exception:
+            # Best-effort: user may still be able to use API even if FS init fails.
+            pass
+        # Best-effort: reconcile SFTPGo user to include /common read-only mounts.
+        # This makes the virtual folders visible without requiring a password reset.
+        if _sftpgo_enabled() and not subject.get("is_admin"):
+            try:
+                _sftpgo_client().enable_user(username=user_id, home_dir=_user_home(user_id))
+            except Exception:
+                pass
+        home = _user_home(user_id)
+        out: dict[str, Any] = {
+            "user_id": user_id,
+            "is_admin": bool(subject.get("is_admin")),
+            "paths": {
+                "home": home,
+                "datasets": f"{home}/datasets",
+                "models": f"{home}/models",
+                "code": f"{home}/code",
+                "jobs": f"{home}/jobs",
+                "trash_jobs": f"{home}/trash/jobs",
+            },
+            "retention": {
+                "jobs_trash_after_days": int(v2_cfg.data.retention.jobs_trash_after_days),
+                "jobs_purge_after_days": int(v2_cfg.data.retention.jobs_purge_after_days),
+            },
+        }
+        if _sftpgo_enabled():
+            out["sftp"] = {
+                "host": v2_cfg.data.sftpgo.host,
+                "port": int(v2_cfg.data.sftpgo.sftp_port),
+                "username": user_id,
+            }
+        return out
+
+    @app.get("/api/v2/tasks")
+    async def list_tasks(req: Request, limit: int = 200, offset: int = 0, states: str | None = None) -> dict[str, Any]:
+        subject = _auth(req)
+        lim = max(1, min(int(limit), 1000))
+        off = max(0, int(offset))
+        state_list: list[str] | None = None
+        if states:
+            raw = [s.strip() for s in str(states).split(",") if s.strip()]
+            # Keep it strict to avoid typos silently returning empty results.
+            allowed = {
+                "QUEUED",
+                "PENDING_RESOURCES",
+                "SUBMITTING",
+                "SUBMITTED",
+                "RUNNING",
+                "SUCCEEDED",
+                "FAILED",
+                "CANCELED",
+            }
+            for s in raw:
+                if s not in allowed:
+                    raise HTTPException(status_code=400, detail=f"invalid state: {s}")
+            state_list = raw or None
+        if subject.get("is_admin"):
+            tasks = db.list_tasks(user_id=None, states=state_list, limit=lim, offset=off)
+        else:
+            tasks = db.list_tasks(user_id=str(subject["user_id"]), states=state_list, limit=lim, offset=off)
+        return {"tasks": tasks, "limit": lim, "offset": off, "has_more": bool(len(tasks) == lim)}
+
     @app.post("/api/v2/tasks")
     async def submit_task(req: Request) -> dict[str, Any]:
         subject = _auth(req)
@@ -126,13 +325,40 @@ def create_app(config_path: str) -> FastAPI:
         except Exception as e:
             raise HTTPException(status_code=400, detail=f"invalid jobspec: {e!r}")
 
-        # v2.5 constraint: training inputs must come from /private/common (dev/prod统一)。
-        common_prefix = ray_cfg.shared_root.rstrip("/") + "/common/"
-        for k, v in (("code_path", spec.code_path), ("train_file", spec.train_file), ("val_file", spec.val_file)):
+        # v3.0 path policy:
+        # - code_path: only allow /private/common/...
+        # - train/val: allow /private/common/datasets/... OR /private/users/<me>/datasets/...
+        # - model_id: if it looks like a local path (/private/...), allow only models dirs:
+        #   /private/common/models/... OR /private/users/<me>/models/...
+        root = ray_cfg.shared_root.rstrip("/")
+        common_prefix = f"{root}/common/"
+        user_prefix = f"{root}/users/{str(subject['user_id']).strip()}/"
+
+        common_datasets_prefix = f"{common_prefix}datasets/"
+        user_datasets_prefix = f"{user_prefix}datasets/"
+        common_models_prefix = f"{common_prefix}models/"
+        user_models_prefix = f"{user_prefix}models/"
+
+        if not str(spec.code_path).startswith(common_prefix):
+            raise HTTPException(status_code=400, detail=f"code_path must start with {common_prefix}")
+
+        for k, v in (("train_file", spec.train_file), ("val_file", spec.val_file)):
             if v is None:
                 continue
-            if not str(v).startswith(common_prefix):
-                raise HTTPException(status_code=400, detail=f"{k} must start with {common_prefix}")
+            sv = str(v)
+            if not (sv.startswith(common_datasets_prefix) or sv.startswith(user_datasets_prefix)):
+                raise HTTPException(
+                    status_code=400,
+                    detail=f"{k} must start with {common_datasets_prefix} or {user_datasets_prefix}",
+                )
+
+        model_id = str(spec.model_id)
+        if model_id.startswith(f"{root}/"):
+            if not (model_id.startswith(common_models_prefix) or model_id.startswith(user_models_prefix)):
+                raise HTTPException(
+                    status_code=400,
+                    detail=f"model_id local path must start with {common_models_prefix} or {user_models_prefix}",
+                )
 
         task_id = new_task_id(spec.workload, user_id=str(subject["user_id"]))
         db.create_task_v25(
@@ -257,4 +483,7 @@ def create_app(config_path: str) -> FastAPI:
             return db.list_queue()
         return db.list_queue(user_id=str(subject["user_id"]))
 
+    # v3.0: minimal WebUI (no server-side session; token stored in browser localStorage).
+    register_ui_routes(app)
+
     return app
diff --git a/src/mvp/py/argus/service/config.py b/src/mvp/py/argus/service/config.py
index 7d3f5dd..3ece213 100644
--- a/src/mvp/py/argus/service/config.py
+++ b/src/mvp/py/argus/service/config.py
@@ -27,12 +27,37 @@ class V2SchedulerConfig:
     max_running_tasks: int = 1
 
 
+@dataclass(frozen=True)
+class V2RetentionConfig:
+    jobs_trash_after_days: int = 3
+    jobs_purge_after_days: int = 7
+    janitor_interval_s: int = 3600
+
+
+@dataclass(frozen=True)
+class V2SFTPGoConfig:
+    enabled: bool = False
+    host: str = ""
+    sftp_port: int = 2022
+    admin_api_base: str = ""
+    admin_user: str = "admin"
+    admin_password_env: str = "SFTPGO_ADMIN_PASSWORD"
+
+
+@dataclass(frozen=True)
+class V2DataConfig:
+    user_root: str
+    sftpgo: V2SFTPGoConfig
+    retention: V2RetentionConfig
+
+
 @dataclass(frozen=True)
 class V2Config:
     api: V2ApiConfig
     auth: V2AuthConfig
     sqlite: V2SqliteConfig
     scheduler: V2SchedulerConfig
+    data: V2DataConfig
 
     @staticmethod
     def from_root_dict(root: dict[str, Any]) -> "V2Config":
@@ -58,9 +83,19 @@ class V2Config:
         else:
             shared_root = str(root.get("shared_root") or "/private")
 
+        data = root.get("data") or {}
+        if not isinstance(data, dict):
+            raise ValueError("config.data must be a mapping")
+        sftpgo = data.get("sftpgo") or {}
+        retention = data.get("retention") or {}
+        if not isinstance(sftpgo, dict) or not isinstance(retention, dict):
+            raise ValueError("config.data.{sftpgo,retention} must be mappings")
+
         default_db_path = f"{shared_root}/common/db/mvp.sqlite3"
         db_path = str(sqlite.get("db_path") or default_db_path)
 
+        user_root = str(data.get("user_root") or f"{shared_root}/users")
+
         return V2Config(
             api=V2ApiConfig(
                 host=str(api.get("host") or "0.0.0.0"),
@@ -73,4 +108,20 @@ class V2Config:
                 retry_interval_s=int(scheduler.get("retry_interval_s") or 60),
                 max_running_tasks=int(scheduler.get("max_running_tasks") or 1),
             ),
+            data=V2DataConfig(
+                user_root=user_root,
+                sftpgo=V2SFTPGoConfig(
+                    enabled=bool(sftpgo.get("enabled") or False),
+                    host=str(sftpgo.get("host") or ""),
+                    sftp_port=int(sftpgo.get("sftp_port") or 2022),
+                    admin_api_base=str(sftpgo.get("admin_api_base") or ""),
+                    admin_user=str(sftpgo.get("admin_user") or "admin"),
+                    admin_password_env=str(sftpgo.get("admin_password_env") or "SFTPGO_ADMIN_PASSWORD"),
+                ),
+                retention=V2RetentionConfig(
+                    jobs_trash_after_days=int(retention.get("jobs_trash_after_days") or 3),
+                    jobs_purge_after_days=int(retention.get("jobs_purge_after_days") or 7),
+                    janitor_interval_s=int(retention.get("janitor_interval_s") or 3600),
+                ),
+            ),
         )
diff --git a/src/mvp/py/argus/service/db.py b/src/mvp/py/argus/service/db.py
index 0aa36d9..8ca314b 100644
--- a/src/mvp/py/argus/service/db.py
+++ b/src/mvp/py/argus/service/db.py
@@ -40,7 +40,8 @@ class Db:
                   user_id TEXT PRIMARY KEY,
                   display_name TEXT,
                   state TEXT NOT NULL,
-                  created_at TEXT NOT NULL
+                  created_at TEXT NOT NULL,
+                  updated_at TEXT NOT NULL
                 )
                 """
             )
@@ -49,6 +50,7 @@ class Db:
                 CREATE TABLE IF NOT EXISTS api_tokens (
                   token_hash TEXT PRIMARY KEY,
                   user_id TEXT NOT NULL,
+                  token_plain TEXT NOT NULL,
                   created_at TEXT NOT NULL,
                   last_used_at TEXT,
                   FOREIGN KEY (user_id) REFERENCES users(user_id) ON DELETE CASCADE
@@ -78,6 +80,15 @@ class Db:
                 conn.execute("ALTER TABLE tasks ADD COLUMN user_id TEXT")
             except sqlite3.OperationalError:
                 pass
+            # Best-effort: add missing columns (forward compatibility).
+            try:
+                conn.execute("ALTER TABLE users ADD COLUMN updated_at TEXT")
+            except sqlite3.OperationalError:
+                pass
+            try:
+                conn.execute("ALTER TABLE api_tokens ADD COLUMN token_plain TEXT")
+            except sqlite3.OperationalError:
+                pass
             conn.execute(
                 """
                 CREATE TABLE IF NOT EXISTS attempts (
@@ -168,10 +179,10 @@ class Db:
         with self.tx() as conn:
             conn.execute(
                 """
-                INSERT INTO users (user_id, display_name, state, created_at)
-                VALUES (?, ?, 'ACTIVE', ?)
+                INSERT INTO users (user_id, display_name, state, created_at, updated_at)
+                VALUES (?, ?, 'ACTIVE', ?, ?)
                 """,
-                (user_id, display_name, now),
+                (user_id, display_name, now, now),
             )
             conn.execute(
                 "INSERT INTO events (task_id, ts, event_type, payload_json) VALUES (NULL, ?, 'USER_CREATED', ?)",
@@ -183,7 +194,7 @@ class Db:
     def disable_user(self, *, user_id: str) -> None:
         now = _utc_now_iso()
         with self.tx() as conn:
-            conn.execute("UPDATE users SET state = 'DISABLED' WHERE user_id = ?", (user_id,))
+            conn.execute("UPDATE users SET state = 'DISABLED', updated_at = ? WHERE user_id = ?", (now, user_id))
             conn.execute(
                 "INSERT INTO events (task_id, ts, event_type, payload_json) VALUES (NULL, ?, 'USER_DISABLED', ?)",
                 (now, user_id),
@@ -195,15 +206,18 @@ class Db:
             return dict(row) if row else None
 
     def issue_token(self, *, user_id: str) -> str:
-        # Returns plaintext token once; stores hash only.
+        # Returns plaintext token once.
+        # Note: For v3.0 WebUI admin convenience we persist the plaintext token so admins
+        # can re-copy it later. This is acceptable only for internal/dev deployments.
         now = _utc_now_iso()
         token = f"mvp_u_{user_id}_{secrets.token_urlsafe(18)}"
         token_hash = self._hash_token(token)
         with self.tx() as conn:
             conn.execute(
-                "INSERT INTO api_tokens (token_hash, user_id, created_at) VALUES (?, ?, ?)",
-                (token_hash, user_id, now),
+                "INSERT INTO api_tokens (token_hash, user_id, token_plain, created_at) VALUES (?, ?, ?, ?)",
+                (token_hash, user_id, token, now),
             )
+            conn.execute("UPDATE users SET updated_at = ? WHERE user_id = ?", (now, user_id))
             conn.execute(
                 "INSERT INTO events (task_id, ts, event_type, payload_json) VALUES (NULL, ?, 'TOKEN_ISSUED', ?)",
                 (now, user_id),
@@ -226,8 +240,48 @@ class Db:
                 return None
             now = _utc_now_iso()
             conn.execute("UPDATE api_tokens SET last_used_at = ? WHERE token_hash = ?", (now, token_hash))
+            conn.execute("UPDATE users SET updated_at = ? WHERE user_id = ?", (now, str(row["user_id"])))
             return {"user_id": row["user_id"], "state": row["state"]}
 
+    def list_users(self, limit: int = 200) -> list[dict[str, Any]]:
+        with self._connect() as conn:
+            rows = conn.execute(
+                """
+                SELECT
+                  u.user_id,
+                  u.display_name,
+                  u.state,
+                  u.created_at,
+                  u.updated_at,
+                  (
+                    SELECT t.token_plain
+                    FROM api_tokens t
+                    WHERE t.user_id = u.user_id
+                    ORDER BY t.created_at DESC
+                    LIMIT 1
+                  ) AS token,
+                  (
+                    SELECT t.created_at
+                    FROM api_tokens t
+                    WHERE t.user_id = u.user_id
+                    ORDER BY t.created_at DESC
+                    LIMIT 1
+                  ) AS token_created_at,
+                  (
+                    SELECT t.last_used_at
+                    FROM api_tokens t
+                    WHERE t.user_id = u.user_id
+                    ORDER BY t.created_at DESC
+                    LIMIT 1
+                  ) AS token_last_used_at
+                FROM users u
+                ORDER BY u.created_at DESC
+                LIMIT ?
+                """,
+                (int(limit),),
+            ).fetchall()
+            return [dict(r) for r in rows]
+
     def get_task(self, task_id: str) -> dict[str, Any] | None:
         with self._connect() as conn:
             row = conn.execute("SELECT * FROM tasks WHERE task_id = ?", (task_id,)).fetchone()
@@ -269,6 +323,40 @@ class Db:
             running = conn.execute(running_sql, tuple(params)).fetchall()
             return {"pending": [dict(r) for r in pending], "running": [dict(r) for r in running]}
 
+    def list_tasks(
+        self,
+        *,
+        user_id: str | None = None,
+        states: list[str] | None = None,
+        limit: int = 200,
+        offset: int = 0,
+    ) -> list[dict[str, Any]]:
+        """
+        Returns recent tasks (including terminal tasks).
+        """
+        with self._connect() as conn:
+            params: list[Any] = []
+            where_clauses: list[str] = []
+            if user_id is not None:
+                where_clauses.append("user_id = ?")
+                params.append(user_id)
+            if states:
+                placeholders = ",".join(["?"] * len(states))
+                where_clauses.append(f"state IN ({placeholders})")
+                params.extend(states)
+            where_sql = f" WHERE {' AND '.join(where_clauses)}" if where_clauses else ""
+            sql = (
+                "SELECT task_id, user_id, workload, state, nnodes, n_gpus_per_node, latest_attempt_no, created_at, updated_at, error_summary "
+                "FROM tasks"
+                f"{where_sql} "
+                "ORDER BY created_at DESC "
+                "LIMIT ? OFFSET ?"
+            )
+            params.append(int(limit))
+            params.append(max(0, int(offset)))
+            rows = conn.execute(sql, tuple(params)).fetchall()
+            return [dict(r) for r in rows]
+
     def count_running(self) -> int:
         with self._connect() as conn:
             row = conn.execute(
@@ -383,3 +471,25 @@ class Db:
                 "INSERT INTO events (task_id, ts, event_type, payload_json) VALUES (?, ?, 'ATTEMPT_UPDATE', ?)",
                 (task_id, now, None),
             )
+
+    def list_ended_attempts_before(self, *, end_time_le: str, limit: int = 2000) -> list[dict[str, Any]]:
+        """
+        Used by the jobs janitor:
+        - Only considers tasks with a non-null user_id (v2.5+).
+        - Returns attempts that have end_time <= end_time_le.
+        """
+        with self._connect() as conn:
+            rows = conn.execute(
+                """
+                SELECT t.task_id, t.user_id, a.ray_submission_id, a.end_time
+                FROM attempts a
+                JOIN tasks t ON t.task_id = a.task_id
+                WHERE t.user_id IS NOT NULL
+                  AND a.end_time IS NOT NULL
+                  AND a.end_time <= ?
+                ORDER BY a.end_time ASC
+                LIMIT ?
+                """,
+                (str(end_time_le), int(limit)),
+            ).fetchall()
+            return [dict(r) for r in rows]
diff --git a/src/mvp/py/argus/service/janitor.py b/src/mvp/py/argus/service/janitor.py
new file mode 100644
index 0000000..61a5e59
--- /dev/null
+++ b/src/mvp/py/argus/service/janitor.py
@@ -0,0 +1,105 @@
+from __future__ import annotations
+
+import os
+import shutil
+import time
+from dataclasses import dataclass
+from datetime import datetime, timedelta, timezone
+
+from .db import Db
+
+
+def _parse_iso_z(ts: str) -> datetime:
+    # Stored as "YYYY-MM-DDTHH:MM:SSZ" (naive UTC). Parse into aware UTC.
+    return datetime.fromisoformat(ts.replace("Z", "+00:00")).astimezone(timezone.utc)
+
+
+def _iso_z(dt: datetime) -> str:
+    return dt.astimezone(timezone.utc).replace(microsecond=0).isoformat().replace("+00:00", "Z")
+
+
+@dataclass
+class JobsJanitor:
+    db: Db
+    user_root: str
+    trash_after_days: int = 3
+    purge_after_days: int = 7
+    interval_s: int = 3600
+
+    def __post_init__(self) -> None:
+        if int(self.trash_after_days) < 0 or int(self.purge_after_days) < 0:
+            raise ValueError("retention days must be non-negative")
+        if int(self.purge_after_days) and int(self.purge_after_days) < int(self.trash_after_days):
+            raise ValueError("purge_after_days must be >= trash_after_days")
+
+    def _job_dir(self, *, user_id: str, ray_submission_id: str) -> str:
+        base = self.user_root.rstrip("/")
+        return f"{base}/{user_id}/jobs/{ray_submission_id}"
+
+    def _trash_dir(self, *, user_id: str, ray_submission_id: str) -> str:
+        base = self.user_root.rstrip("/")
+        return f"{base}/{user_id}/trash/jobs/{ray_submission_id}"
+
+    def tick_once(self, *, now: datetime | None = None, limit: int = 2000) -> None:
+        if int(self.trash_after_days) <= 0 and int(self.purge_after_days) <= 0:
+            return
+
+        now_dt = now or datetime.now(timezone.utc)
+        move_cutoff = now_dt - timedelta(days=int(self.trash_after_days))
+        move_cutoff_iso = _iso_z(move_cutoff)
+
+        rows = self.db.list_ended_attempts_before(end_time_le=move_cutoff_iso, limit=int(limit))
+        for r in rows:
+            user_id = str(r.get("user_id") or "").strip()
+            sid = str(r.get("ray_submission_id") or "").strip()
+            end_time = str(r.get("end_time") or "").strip()
+            if not user_id or not sid or not end_time:
+                continue
+
+            try:
+                ended_at = _parse_iso_z(end_time)
+            except Exception:
+                continue
+
+            age_days = (now_dt - ended_at).total_seconds() / 86400.0
+            src = self._job_dir(user_id=user_id, ray_submission_id=sid)
+            dst = self._trash_dir(user_id=user_id, ray_submission_id=sid)
+            dst_parent = os.path.dirname(dst)
+
+            # Between trash and purge: ensure in trash.
+            if age_days >= float(self.trash_after_days) and age_days < float(self.purge_after_days or 10**9):
+                if os.path.exists(src) and not os.path.exists(dst):
+                    os.makedirs(dst_parent, exist_ok=True)
+                    try:
+                        shutil.move(src, dst)
+                    except Exception:
+                        pass
+                continue
+
+            # Purge: move to trash (if still in jobs) then delete from trash.
+            if int(self.purge_after_days) > 0 and age_days >= float(self.purge_after_days):
+                if os.path.exists(src) and not os.path.exists(dst):
+                    os.makedirs(dst_parent, exist_ok=True)
+                    try:
+                        shutil.move(src, dst)
+                    except Exception:
+                        pass
+                if os.path.exists(dst):
+                    try:
+                        shutil.rmtree(dst)
+                    except Exception:
+                        pass
+
+    def run_forever(self, stop_flag: object) -> None:
+        try:
+            is_set = stop_flag.is_set  # type: ignore[attr-defined]
+        except Exception:
+            raise ValueError("stop_flag must be a threading.Event-like object with is_set()")
+
+        while not is_set():
+            try:
+                self.tick_once()
+            except Exception:
+                pass
+            time.sleep(max(1, int(self.interval_s)))
+
diff --git a/src/mvp/py/argus/service/sftpgo.py b/src/mvp/py/argus/service/sftpgo.py
new file mode 100644
index 0000000..09b2681
--- /dev/null
+++ b/src/mvp/py/argus/service/sftpgo.py
@@ -0,0 +1,221 @@
+from __future__ import annotations
+
+import base64
+import json
+from dataclasses import dataclass
+from urllib.error import HTTPError, URLError
+from urllib.request import Request, urlopen
+
+
+class SFTPGoError(RuntimeError):
+    pass
+
+
+@dataclass(frozen=True)
+class SFTPGoAdminClient:
+    """
+    Minimal SFTPGo admin API client (v3.0).
+
+    SFTPGo OpenAPI documents the admin token flow:
+      - GET /api/v2/token with HTTP BasicAuth -> returns {"access_token": "..."}.
+      - Use Authorization: Bearer <token> for admin endpoints like POST /api/v2/users.
+
+    See upstream OpenAPI for details:
+      https://raw.githubusercontent.com/drakkan/sftpgo/main/openapi/openapi.yaml
+    """
+
+    admin_api_base: str
+    admin_user: str
+    admin_password: str
+    common_root: str = "/private/common"
+
+    def _url(self, path: str) -> str:
+        base = self.admin_api_base.rstrip("/")
+        p = path if path.startswith("/") else f"/{path}"
+        return f"{base}{p}"
+
+    def _basic_auth_header(self) -> str:
+        raw = f"{self.admin_user}:{self.admin_password}".encode("utf-8")
+        return "Basic " + base64.b64encode(raw).decode("ascii")
+
+    def _get_json(self, url: str, headers: dict[str, str]) -> dict:
+        req = Request(url, headers=headers, method="GET")
+        try:
+            with urlopen(req, timeout=10) as resp:
+                return json.loads(resp.read().decode("utf-8"))
+        except HTTPError as e:
+            raise SFTPGoError(f"sftpgo http error: {e.code} {e.reason}") from e
+        except URLError as e:
+            raise SFTPGoError(f"sftpgo connection error: {e!r}") from e
+
+    def _post_json(self, url: str, payload: dict, headers: dict[str, str]) -> None:
+        data = json.dumps(payload).encode("utf-8")
+        req = Request(url, data=data, headers=headers, method="POST")
+        try:
+            with urlopen(req, timeout=10) as resp:
+                _ = resp.read()
+        except HTTPError as e:
+            raise SFTPGoError(f"sftpgo http error: {e.code} {e.reason}") from e
+        except URLError as e:
+            raise SFTPGoError(f"sftpgo connection error: {e!r}") from e
+
+    def _put_json(self, url: str, payload: dict, headers: dict[str, str]) -> None:
+        data = json.dumps(payload).encode("utf-8")
+        req = Request(url, data=data, headers=headers, method="PUT")
+        try:
+            with urlopen(req, timeout=10) as resp:
+                _ = resp.read()
+        except HTTPError as e:
+            raise SFTPGoError(f"sftpgo http error: {e.code} {e.reason}") from e
+        except URLError as e:
+            raise SFTPGoError(f"sftpgo connection error: {e!r}") from e
+
+    def _admin_token(self) -> str:
+        url = self._url("/token")
+        obj = self._get_json(url, headers={"Authorization": self._basic_auth_header()})
+        tok = str(obj.get("access_token") or "").strip()
+        if not tok:
+            raise SFTPGoError("sftpgo token response missing access_token")
+        return tok
+
+    def _auth_headers(self, tok: str) -> dict[str, str]:
+        return {"Authorization": f"Bearer {tok}", "Content-Type": "application/json"}
+
+    def _ensure_folder(self, *, tok: str, name: str, mapped_path: str) -> None:
+        url = self._url("/folders")
+        try:
+            self._post_json(url, {"name": name, "mapped_path": mapped_path}, headers=self._auth_headers(tok))
+        except SFTPGoError as e:
+            # Idempotent + self-healing:
+            # If it already exists, update mapped_path to the desired value.
+            if "409" in str(e):
+                self._put_json(
+                    self._url(f"/folders/{name}"),
+                    {"name": name, "mapped_path": mapped_path},
+                    headers=self._auth_headers(tok),
+                )
+                return
+            raise
+
+    def _ensure_common_folders(self, *, tok: str) -> None:
+        # Important: do NOT map datasets to /private/common/datasets because that path is
+        # a symlink farm (e.g. gsm8k -> /private/datasets/gsm8k) and SFTPGo WebClient
+        # will treat symlink traversal as escaping the virtual folder root, resulting
+        # in "permission denied". Map directly to the real datasets root.
+        common = self.common_root.rstrip("/")
+        if common.endswith("/common"):
+            shared_root = common[: -len("/common")] or "/private"
+        else:
+            shared_root = "/private"
+
+        self._ensure_folder(tok=tok, name="common_datasets", mapped_path=f"{shared_root}/datasets")
+        # Expose HF cache read-only so users can inspect downloaded models/datasets.
+        self._ensure_folder(tok=tok, name="common_hf", mapped_path=f"{shared_root}/hf")
+
+    def _get_user(self, *, tok: str, username: str) -> dict:
+        url = self._url(f"/users/{username}")
+        return self._get_json(url, headers={"Authorization": f"Bearer {tok}"})
+
+    def _put_user(self, *, tok: str, username: str, payload: dict) -> None:
+        url = self._url(f"/users/{username}")
+        self._put_json(url, payload, headers=self._auth_headers(tok))
+
+    def _apply_common_readonly(self, user_payload: dict) -> dict:
+        # Path-based permissions: make /common/* read-only while keeping home writeable.
+        perms = dict(user_payload.get("permissions") or {"/": ["*"]})
+        # Ensure /common is visible as a directory and can be traversed.
+        perms["/common"] = ["list"]
+        perms["/common/datasets"] = ["list", "download"]
+        perms["/common/hf"] = ["list", "download"]
+        user_payload["permissions"] = perms
+
+        desired_vf = [
+            {"name": "common_datasets", "virtual_path": "/common/datasets"},
+            {"name": "common_hf", "virtual_path": "/common/hf"},
+        ]
+        existing = user_payload.get("virtual_folders") or []
+        if not isinstance(existing, list):
+            existing = []
+        seen = {(vf.get("name"), vf.get("virtual_path")) for vf in existing if isinstance(vf, dict)}
+        merged = list(existing)
+        for vf in desired_vf:
+            key = (vf["name"], vf["virtual_path"])
+            if key not in seen:
+                merged.append(vf)
+        user_payload["virtual_folders"] = merged
+        return user_payload
+
+    def create_user(self, *, username: str, password: str, home_dir: str) -> None:
+        tok = self._admin_token()
+        self._ensure_common_folders(tok=tok)
+        url = self._url("/users")
+        payload: dict = {
+            "status": 1,
+            "username": username,
+            "password": password,
+            "home_dir": home_dir,
+            "permissions": {
+                "/": ["*"],
+                "/common": ["list"],
+                "/common/datasets": ["list", "download"],
+                "/common/hf": ["list", "download"],
+            },
+            "virtual_folders": [
+                {"name": "common_datasets", "virtual_path": "/common/datasets"},
+                {"name": "common_hf", "virtual_path": "/common/hf"},
+            ],
+        }
+        self._post_json(
+            url,
+            payload,
+            headers=self._auth_headers(tok),
+        )
+
+    def disable_user(self, *, username: str, home_dir: str) -> None:
+        tok = self._admin_token()
+        self._ensure_common_folders(tok=tok)
+        cur = self._get_user(tok=tok, username=username)
+        payload: dict = {
+            "username": username,
+            "status": 0,
+            "home_dir": home_dir,
+            "uid": cur.get("uid", 0),
+            "gid": cur.get("gid", 0),
+            "permissions": cur.get("permissions") or {"/": ["*"]},
+            "virtual_folders": cur.get("virtual_folders") or [],
+        }
+        self._apply_common_readonly(payload)
+        self._put_user(tok=tok, username=username, payload=payload)
+
+    def enable_user(self, *, username: str, home_dir: str) -> None:
+        tok = self._admin_token()
+        self._ensure_common_folders(tok=tok)
+        cur = self._get_user(tok=tok, username=username)
+        payload: dict = {
+            "username": username,
+            "status": 1,
+            "home_dir": home_dir,
+            "uid": cur.get("uid", 0),
+            "gid": cur.get("gid", 0),
+            "permissions": cur.get("permissions") or {"/": ["*"]},
+            "virtual_folders": cur.get("virtual_folders") or [],
+        }
+        self._apply_common_readonly(payload)
+        self._put_user(tok=tok, username=username, payload=payload)
+
+    def reset_password(self, *, username: str, new_password: str, home_dir: str) -> None:
+        tok = self._admin_token()
+        self._ensure_common_folders(tok=tok)
+        cur = self._get_user(tok=tok, username=username)
+        payload: dict = {
+            "username": username,
+            "status": 1,
+            "home_dir": home_dir,
+            "uid": cur.get("uid", 0),
+            "gid": cur.get("gid", 0),
+            "permissions": cur.get("permissions") or {"/": ["*"]},
+            "virtual_folders": cur.get("virtual_folders") or [],
+            "password": new_password,
+        }
+        self._apply_common_readonly(payload)
+        self._put_user(tok=tok, username=username, payload=payload)
diff --git a/src/mvp/py/argus/service/ui.py b/src/mvp/py/argus/service/ui.py
new file mode 100644
index 0000000..27de826
--- /dev/null
+++ b/src/mvp/py/argus/service/ui.py
@@ -0,0 +1,629 @@
+from __future__ import annotations
+
+import html
+import json
+
+from fastapi import FastAPI
+from fastapi.responses import HTMLResponse, RedirectResponse
+
+
+_BASE_CSS = """
+:root { --bg:#0b1020; --panel:#111a33; --muted:#95a3c6; --fg:#e8eeff; --accent:#7aa2ff; --danger:#ff6b6b; --ok:#3ddc97; }
+* { box-sizing: border-box; }
+body { margin:0; font-family: ui-sans-serif, system-ui, -apple-system, Segoe UI, Roboto, Helvetica, Arial; background:var(--bg); color:var(--fg); }
+a { color:var(--accent); text-decoration:none; }
+.layout { display:flex; min-height:100vh; }
+.nav { width: 240px; padding:16px; background: linear-gradient(180deg, #0e1630, #0b1020); border-right: 1px solid rgba(255,255,255,0.06); }
+.brand { font-weight: 700; letter-spacing: .2px; margin-bottom: 12px; }
+.nav a { display:block; padding:10px 10px; border-radius:10px; color: var(--fg); opacity: .9; }
+.nav a.active { background: rgba(122,162,255,0.14); border: 1px solid rgba(122,162,255,0.22); }
+.nav a:hover { background: rgba(255,255,255,0.06); }
+.main { flex:1; padding: 20px 24px; }
+.card { background: rgba(255,255,255,0.04); border: 1px solid rgba(255,255,255,0.08); border-radius: 14px; padding: 16px; }
+.row { display:flex; gap: 12px; align-items:center; flex-wrap: wrap; }
+.muted { color: var(--muted); }
+.btn { border: 1px solid rgba(255,255,255,0.16); background: rgba(255,255,255,0.06); color: var(--fg); padding: 10px 12px; border-radius: 10px; cursor: pointer; }
+.btn:hover { background: rgba(255,255,255,0.10); }
+.btn.danger { border-color: rgba(255,107,107,0.35); background: rgba(255,107,107,0.10); }
+.pill { display:inline-block; padding: 2px 10px; border-radius: 999px; border: 1px solid rgba(255,255,255,0.16); font-size: 12px; }
+.pill.ok { border-color: rgba(61,220,151,0.35); background: rgba(61,220,151,0.12); }
+.pill.bad { border-color: rgba(255,107,107,0.35); background: rgba(255,107,107,0.12); }
+textarea, input { width: 100%; color: var(--fg); background: rgba(255,255,255,0.06); border: 1px solid rgba(255,255,255,0.12); border-radius: 12px; padding: 10px 12px; outline: none; }
+button:disabled { opacity: .45; cursor: not-allowed; }
+pre { white-space: pre-wrap; word-break: break-word; }
+table { width:100%; border-collapse: collapse; }
+th, td { padding: 10px 8px; border-bottom: 1px solid rgba(255,255,255,0.08); text-align:left; }
+""".strip()
+
+
+_BASE_JS = """
+function mvpTokenGet() {
+  return (localStorage.getItem("mvp_token") || "").trim();
+}
+function mvpTokenSet(v) {
+  localStorage.setItem("mvp_token", (v || "").trim());
+}
+function mvpSftpPasswordGet() {
+  return (localStorage.getItem("mvp_sftp_password") || "").trim();
+}
+function mvpSftpPasswordSet(v) {
+  localStorage.setItem("mvp_sftp_password", (v || "").trim());
+}
+async function apiFetch(path, opts) {
+  opts = opts || {};
+  opts.headers = opts.headers || {};
+  const tok = mvpTokenGet();
+  if (tok) opts.headers["Authorization"] = "Bearer " + tok;
+  return fetch(path, opts);
+}
+async function apiJson(path, opts) {
+  const resp = await apiFetch(path, opts);
+  const text = await resp.text();
+  if (!resp.ok) {
+    const err = new Error("HTTP " + resp.status);
+    err.status = resp.status;
+    err.body = text;
+    throw err;
+  }
+  return JSON.parse(text);
+}
+function fmtJson(obj) {
+  try { return JSON.stringify(obj, null, 2); } catch (e) { return String(obj); }
+}
+function curOriginWithPort(port) {
+  const proto = window.location.protocol;
+  const host = window.location.hostname;
+  return proto + "//" + host + ":" + port;
+}
+async function copyText(v) {
+  if (!v) return false;
+  try {
+    await navigator.clipboard.writeText(v);
+    return true;
+  } catch (e) {
+    // Fallback for non-secure contexts (http) or older browsers.
+    try {
+      const ta = document.createElement("textarea");
+      ta.value = v;
+      ta.style.position = "fixed";
+      ta.style.opacity = "0";
+      document.body.appendChild(ta);
+      ta.focus();
+      ta.select();
+      const ok = document.execCommand("copy");
+      document.body.removeChild(ta);
+      return ok;
+    } catch (e2) {
+      return false;
+    }
+  }
+}
+document.addEventListener("DOMContentLoaded", () => {
+  const el = document.getElementById("nav-ray-dashboard");
+  if (el) el.href = curOriginWithPort(8265);
+});
+""".strip()
+
+
+def _nav(active: str) -> str:
+    links = [
+        ("login", "/ui/login", "Login"),
+        ("tasks", "/ui/tasks", "Tasks"),
+        ("new", "/ui/tasks/new", "New Task"),
+        ("data", "/ui/data", "Data"),
+        ("admin", "/ui/admin", "Admin"),
+        ("ray", "#", "Ray Dashboard"),
+    ]
+    items = []
+    for key, href, label in links:
+        cls = "active" if key == active else ""
+        extra = ""
+        if key == "ray":
+            extra = ' id="nav-ray-dashboard" target="_blank" rel="noopener"'
+        items.append(f'<a class="{cls}" href="{href}"{extra}>{html.escape(label)}</a>')
+    return "\n".join(items)
+
+
+def _page(title: str, active: str, body: str, script: str = "") -> str:
+    return f"""<!doctype html>
+<html lang="en">
+<head>
+  <meta charset="utf-8" />
+  <meta name="viewport" content="width=device-width, initial-scale=1" />
+  <title>{html.escape(title)}</title>
+  <style>{_BASE_CSS}</style>
+</head>
+<body>
+  <div class="layout">
+    <nav class="nav">
+      <div class="brand">Argus MVP</div>
+      {_nav(active)}
+      <div style="margin-top:12px" class="muted">
+        Token stored in browser localStorage as <code>mvp_token</code>.
+      </div>
+    </nav>
+    <main class="main">
+      {body}
+    </main>
+  </div>
+  <script>{_BASE_JS}</script>
+  <script>{script}</script>
+</body>
+</html>"""
+
+
+def register_ui_routes(app: FastAPI) -> None:
+    @app.get("/ui")
+    async def ui_root() -> RedirectResponse:
+        return RedirectResponse(url="/ui/tasks")
+
+    @app.get("/ui/login")
+    async def ui_login() -> HTMLResponse:
+        body = """
+<h1>Login</h1>
+<div class="card">
+  <div class="muted">Paste your API token (without the <code>Bearer</code> prefix).</div>
+  <div style="height:10px"></div>
+  <input id="tok" placeholder="token..." />
+  <div style="height:10px"></div>
+  <div class="row">
+    <button class="btn" id="save">Save</button>
+    <button class="btn" id="clear">Clear</button>
+    <a href="/ui/tasks" class="btn" style="display:inline-block">Go to Tasks</a>
+  </div>
+  <div style="height:10px"></div>
+  <div id="msg" class="muted"></div>
+</div>
+""".strip()
+        script = """
+const tokEl = document.getElementById("tok");
+const msg = document.getElementById("msg");
+tokEl.value = mvpTokenGet();
+document.getElementById("save").onclick = () => { mvpTokenSet(tokEl.value); msg.textContent = "Saved."; };
+document.getElementById("clear").onclick = () => { mvpTokenSet(""); tokEl.value = ""; msg.textContent = "Cleared."; };
+""".strip()
+        return HTMLResponse(content=_page("Login", "login", body, script))
+
+    @app.get("/ui/tasks")
+    async def ui_tasks() -> HTMLResponse:
+        body = """
+<h1>Tasks</h1>
+<div class="card">
+  <div class="row">
+    <button class="btn" id="refresh">Refresh</button>
+    <a class="btn" href="/ui/tasks/new" style="display:inline-block">New Task</a>
+  </div>
+  <div style="height:10px"></div>
+  <div id="out" class="muted">Loading...</div>
+</div>
+""".strip()
+        script = """
+const out = document.getElementById("out");
+async function refresh() {
+  out.textContent = "Loading...";
+  try {
+    const q = await apiJson("/api/v2/queue");
+    const completedLimit = 25;
+    const completedOffset = Number(localStorage.getItem("mvp_completed_offset") || "0") || 0;
+    const done = await apiJson("/api/v2/tasks?limit=" + completedLimit + "&offset=" + completedOffset + "&states=SUCCEEDED,FAILED,CANCELED");
+
+    function pill(state) {
+    const s = String(state || "");
+    if (s === "SUCCEEDED") return `<span class="pill ok">${s}</span>`;
+    if (s === "FAILED") return `<span class="pill bad">${s}</span>`;
+    if (s === "CANCELED") return `<span class="pill">${s}</span>`;
+    if (s === "RUNNING") return `<span class="pill ok">${s}</span>`;
+    if (s === "QUEUED" || s === "PENDING_RESOURCES" || s === "SUBMITTING" || s === "SUBMITTED") return `<span class="pill">${s}</span>`;
+    return `<span class="pill">${s}</span>`;
+  }
+  function row(t) {
+    const id = t.task_id;
+    return `<tr>
+      <td><a href="/ui/tasks/${id}">${id}</a></td>
+      <td>${t.workload}</td>
+      <td>${pill(t.state)}</td>
+      <td>${t.nnodes} x ${t.n_gpus_per_node} GPU</td>
+      <td>${t.updated_at || ""}</td>
+    </tr>`;
+  }
+
+  const running = (q.running || []).map(row).join("");
+  const pending = (q.pending || []).map(row).join("");
+  const doneRows = (done.tasks || []).map(row).join("");
+  const pageNo = Math.floor(completedOffset / completedLimit) + 1;
+  const prevDisabled = completedOffset <= 0;
+  const nextDisabled = !done.has_more;
+  out.innerHTML = `
+    <div class="muted">Tip: configure token in <a href="/ui/login">Login</a>.</div>
+    <div style="height:10px"></div>
+    <h3>Running</h3>
+    <table><thead><tr><th>Task</th><th>Workload</th><th>State</th><th>Resources</th><th>Updated</th></tr></thead><tbody>${running || "<tr><td colspan=5 class=muted>(none)</td></tr>"}</tbody></table>
+    <div style="height:12px"></div>
+    <h3>Pending</h3>
+    <table><thead><tr><th>Task</th><th>Workload</th><th>State</th><th>Resources</th><th>Updated</th></tr></thead><tbody>${pending || "<tr><td colspan=5 class=muted>(none)</td></tr>"}</tbody></table>
+    <div style="height:12px"></div>
+    <h3>Completed</h3>
+    <div class="row" style="justify-content: space-between; margin-bottom: 8px;">
+      <div class="muted">Page ${pageNo}</div>
+      <div class="row">
+        <button class="btn" id="done-prev" ${prevDisabled ? "disabled" : ""}>Prev</button>
+        <button class="btn" id="done-next" ${nextDisabled ? "disabled" : ""}>Next</button>
+      </div>
+    </div>
+    <table><thead><tr><th>Task</th><th>Workload</th><th>State</th><th>Resources</th><th>Updated</th></tr></thead><tbody>${doneRows || "<tr><td colspan=5 class=muted>(none)</td></tr>"}</tbody></table>
+  `;
+
+  const prevBtn = document.getElementById("done-prev");
+  const nextBtn = document.getElementById("done-next");
+  if (prevBtn) prevBtn.onclick = () => {
+    const cur = Number(localStorage.getItem("mvp_completed_offset") || "0") || 0;
+    const next = Math.max(0, cur - completedLimit);
+    localStorage.setItem("mvp_completed_offset", String(next));
+    refresh();
+  };
+  if (nextBtn) nextBtn.onclick = () => {
+    const cur = Number(localStorage.getItem("mvp_completed_offset") || "0") || 0;
+    const next = cur + completedLimit;
+    localStorage.setItem("mvp_completed_offset", String(next));
+    refresh();
+  };
+  } catch (e) {
+    out.textContent = "Error: " + (e.status || "") + "\\n" + (e.body || String(e));
+  }
+}
+document.getElementById("refresh").onclick = refresh;
+refresh();
+""".strip()
+        return HTMLResponse(content=_page("Tasks", "tasks", body, script))
+
+    @app.get("/ui/tasks/new")
+    async def ui_new_task() -> HTMLResponse:
+        ppo = """# PPO TaskSpec (YAML)
+workload: ppo
+nnodes: 2
+n_gpus_per_node: 4
+code_path: /private/common/code/verl/verl_repo
+train_file: /private/common/datasets/gsm8k/train.parquet
+val_file: /private/common/datasets/gsm8k/test.parquet
+model_id: Qwen/Qwen2.5-0.5B-Instruct
+""".strip()
+        grpo = """# GRPO TaskSpec (YAML)
+workload: grpo
+nnodes: 2
+n_gpus_per_node: 4
+code_path: /private/common/code/verl/verl_repo
+train_file: /private/common/datasets/gsm8k/train.parquet
+val_file: /private/common/datasets/gsm8k/test.parquet
+model_id: Qwen/Qwen2.5-0.5B-Instruct
+""".strip()
+        sft = """# SFT TaskSpec (YAML)
+workload: sft
+nnodes: 1
+n_gpus_per_node: 1
+code_path: /private/common/code/verl/verl_repo
+train_file: /private/common/datasets/gsm8k_sft/train.parquet
+val_file: /private/common/datasets/gsm8k_sft/test.parquet
+model_id: Qwen/Qwen2.5-0.5B-Instruct
+""".strip()
+        body = f"""
+<h1>New Task</h1>
+<div class="card">
+  <div class="muted">Paste TaskSpec YAML and submit to API server. Note: <code>code_path</code> is required (v3.0 does not execute user code; use the common snapshot).</div>
+  <div style="height:10px"></div>
+  <div class="row">
+    <button class="btn" id="tpl-ppo">PPO example</button>
+    <button class="btn" id="tpl-grpo">GRPO example</button>
+    <button class="btn" id="tpl-sft">SFT example</button>
+  </div>
+  <div style="height:10px"></div>
+  <textarea id="yaml" rows="16">{html.escape(ppo)}</textarea>
+  <div style="height:10px"></div>
+  <div class="row">
+    <button class="btn" id="submit">Submit</button>
+    <a class="btn" href="/ui/tasks" style="display:inline-block">Back</a>
+  </div>
+  <div style="height:10px"></div>
+  <pre id="msg" class="muted"></pre>
+</div>
+""".strip()
+        tpl_ppo = json.dumps(ppo)
+        tpl_grpo = json.dumps(grpo)
+        tpl_sft = json.dumps(sft)
+        script = (
+            """
+const msg = document.getElementById("msg");
+const yamlEl = document.getElementById("yaml");
+const TPL_PPO = __TPL_PPO__;
+const TPL_GRPO = __TPL_GRPO__;
+const TPL_SFT = __TPL_SFT__;
+document.getElementById("tpl-ppo").onclick = () => { yamlEl.value = TPL_PPO; msg.textContent = ""; };
+document.getElementById("tpl-grpo").onclick = () => { yamlEl.value = TPL_GRPO; msg.textContent = ""; };
+document.getElementById("tpl-sft").onclick = () => { yamlEl.value = TPL_SFT; msg.textContent = ""; };
+document.getElementById("submit").onclick = async () => {
+  msg.textContent = "Submitting...";
+  const body = yamlEl.value;
+  const resp = await apiFetch("/api/v2/tasks", { method: "POST", headers: {"Content-Type":"text/plain"}, body });
+  const text = await resp.text();
+  if (!resp.ok) { msg.textContent = "Error: " + resp.status + "\\n" + text; return; }
+  const obj = JSON.parse(text);
+  msg.textContent = "OK: " + fmtJson(obj);
+  if (obj.task_id) window.location.href = "/ui/tasks/" + obj.task_id;
+};
+""".strip()
+            .replace("__TPL_PPO__", tpl_ppo)
+            .replace("__TPL_GRPO__", tpl_grpo)
+            .replace("__TPL_SFT__", tpl_sft)
+        )
+        return HTMLResponse(content=_page("New Task", "new", body, script))
+
+    @app.get("/ui/tasks/{task_id}")
+    async def ui_task_detail(task_id: str) -> HTMLResponse:
+        safe_id = html.escape(task_id)
+        body = f"""
+<h1>Task: <code>{safe_id}</code></h1>
+<div class="card">
+  <div class="row">
+    <a class="btn" href="/ui/tasks/{safe_id}/logs" style="display:inline-block">Logs</a>
+    <button class="btn" id="refresh">Refresh</button>
+    <button class="btn danger" id="cancel">Cancel</button>
+    <a class="btn" href="/ui/tasks" style="display:inline-block">Back</a>
+  </div>
+  <div style="height:10px"></div>
+  <pre id="out" class="muted">Loading...</pre>
+</div>
+""".strip()
+        script = f"""
+document.getElementById("nav-ray-dashboard").href = curOriginWithPort(8265);
+const out = document.getElementById("out");
+async function refresh() {{
+  out.textContent = "Loading...";
+  const resp = await apiFetch("/api/v2/tasks/{task_id}");
+  const text = await resp.text();
+  if (!resp.ok) {{ out.textContent = "Error: " + resp.status + "\\n" + text; return; }}
+  out.textContent = fmtJson(JSON.parse(text));
+}}
+document.getElementById("refresh").onclick = refresh;
+document.getElementById("cancel").onclick = async () => {{
+  if (!confirm("Cancel this task?")) return;
+  const resp = await apiFetch("/api/v2/tasks/{task_id}:cancel", {{ method: "POST" }});
+  const text = await resp.text();
+  out.textContent = (resp.ok ? "Canceled.\\n" : "Error: " + resp.status + "\\n") + text;
+  setTimeout(refresh, 800);
+}};
+refresh();
+""".strip()
+        return HTMLResponse(content=_page(f"Task {task_id}", "tasks", body, script))
+
+    @app.get("/ui/tasks/{task_id}/logs")
+    async def ui_task_logs(task_id: str) -> HTMLResponse:
+        safe_id = html.escape(task_id)
+        body = f"""
+<h1>Logs: <code>{safe_id}</code></h1>
+<div class="card">
+  <div class="row">
+    <button class="btn" id="refresh">Refresh</button>
+    <label class="muted">Auto refresh <input type="checkbox" id="auto" /></label>
+    <a class="btn" href="/ui/tasks/{safe_id}" style="display:inline-block">Back</a>
+  </div>
+  <div style="height:10px"></div>
+  <pre id="out" class="muted">Loading...</pre>
+</div>
+""".strip()
+        script = f"""
+document.getElementById("nav-ray-dashboard").href = curOriginWithPort(8265);
+const out = document.getElementById("out");
+let timer = null;
+async function refresh() {{
+  const resp = await apiFetch("/api/v2/tasks/{task_id}/logs?tail=4000");
+  const text = await resp.text();
+  out.textContent = resp.ok ? text : ("Error: " + resp.status + "\\n" + text);
+}}
+document.getElementById("refresh").onclick = refresh;
+document.getElementById("auto").onchange = (e) => {{
+  if (e.target.checked) {{
+    timer = setInterval(refresh, 2000);
+  }} else {{
+    if (timer) clearInterval(timer);
+    timer = null;
+  }}
+}};
+refresh();
+""".strip()
+        return HTMLResponse(content=_page(f"Logs {task_id}", "tasks", body, script))
+
+    @app.get("/ui/data")
+    async def ui_data() -> HTMLResponse:
+        body = """
+<h1>Data</h1>
+<div class="card">
+  <div class="muted">User files live under your home directory. Keep long-term artifacts in <code>models/</code> or <code>datasets/</code>.</div>
+  <div style="height:14px"></div>
+
+  <div class="row">
+    <div style="flex:1; min-width:260px">
+      <div class="muted">Username</div>
+      <div style="height:6px"></div>
+      <div class="row" style="gap:8px">
+        <input id="u" readonly />
+        <button class="btn" id="copy-u">Copy</button>
+      </div>
+    </div>
+    <div style="flex:1; min-width:260px">
+      <div class="muted">SFTPGo password</div>
+      <div style="height:6px"></div>
+      <div class="row" style="gap:8px">
+        <input id="p" placeholder="Click Reset to generate..." />
+        <button class="btn" id="copy-p">Copy</button>
+        <button class="btn" id="reset-p">Reset</button>
+      </div>
+    </div>
+  </div>
+
+  <div style="height:12px"></div>
+  <div class="row">
+    <a class="btn" id="sftp-web" target="_blank" rel="noopener" href="#">Open SFTPGo Web Client (:8081)</a>
+  </div>
+
+  <div style="height:12px"></div>
+  <div class="muted">
+    You can also use an SFTP client (e.g. FileZilla) with the same username/password.
+    Host: <code id="sftp-host"></code>, Port: <code id="sftp-port"></code>.
+  </div>
+
+  <div style="height:14px"></div>
+  <pre id="out" class="muted">Loading...</pre>
+</div>
+""".strip()
+        script = """
+const out = document.getElementById("out");
+document.getElementById("nav-ray-dashboard").href = curOriginWithPort(8265);
+const u = document.getElementById("u");
+const p = document.getElementById("p");
+const sftpWeb = document.getElementById("sftp-web");
+const sftpHost = document.getElementById("sftp-host");
+const sftpPort = document.getElementById("sftp-port");
+document.getElementById("copy-u").onclick = async () => { await copyText(u.value || ""); };
+document.getElementById("copy-p").onclick = async () => { await copyText(p.value || ""); };
+
+async function refresh() {
+  const resp = await apiFetch("/api/v2/me");
+  const text = await resp.text();
+  if (!resp.ok) { out.textContent = "Error: " + resp.status + "\\n" + text; return; }
+  const obj = JSON.parse(text);
+  u.value = (obj.user_id || "");
+  const cached = mvpSftpPasswordGet();
+  if (cached) p.value = cached;
+  const host = curOriginWithPort(8081);
+  sftpWeb.href = host + "/web/client";
+  sftpHost.textContent = (obj.sftp && obj.sftp.host) ? obj.sftp.host : window.location.hostname;
+  sftpPort.textContent = (obj.sftp && obj.sftp.port) ? String(obj.sftp.port) : "2022";
+  out.textContent = fmtJson(obj);
+}
+document.getElementById("reset-p").onclick = async () => {
+  p.value = "";
+  const resp = await apiFetch("/api/v2/me/sftp:reset_password", { method: "POST" });
+  const text = await resp.text();
+  if (!resp.ok) { out.textContent = "Error: " + resp.status + "\\n" + text; return; }
+  const obj = JSON.parse(text);
+  p.value = obj.password || "";
+  mvpSftpPasswordSet(p.value);
+  out.textContent = "SFTPGo password rotated.\\n\\n" + fmtJson(obj);
+};
+refresh();
+""".strip()
+        return HTMLResponse(content=_page("Data", "data", body, script))
+
+    @app.get("/ui/admin")
+    async def ui_admin() -> HTMLResponse:
+        body = """
+<h1>Admin</h1>
+<div class="card">
+  <div class="muted">
+    This page requires the <code>admin</code> token (set it in <a href="/ui/login">Login</a>).
+  </div>
+  <div style="height:14px"></div>
+
+  <h3>Create user</h3>
+  <div class="row">
+    <input id="new-user-id" placeholder="user_id (e.g. alice)" style="max-width:320px" />
+    <input id="new-display-name" placeholder="display_name (optional)" style="max-width:320px" />
+    <button class="btn" id="create-user">Create</button>
+  </div>
+  <div style="height:10px"></div>
+  <pre id="create-msg" class="muted"></pre>
+
+  <div style="height:14px"></div>
+  <div class="row">
+    <button class="btn" id="refresh">Refresh</button>
+  </div>
+  <div style="height:10px"></div>
+  <div id="out" class="muted">Loading...</div>
+</div>
+""".strip()
+        script = """
+const out = document.getElementById("out");
+const createMsg = document.getElementById("create-msg");
+const userIdEl = document.getElementById("new-user-id");
+const displayNameEl = document.getElementById("new-display-name");
+
+function esc(s) {
+  s = String(s || "");
+  return s.replaceAll("&","&amp;").replaceAll("<","&lt;").replaceAll(">","&gt;");
+}
+
+async function refresh() {
+  out.textContent = "Loading...";
+  try {
+    const obj = await apiJson("/api/v2/users?limit=200");
+    const users = (obj.users || []);
+    function row(u) {
+      const uid = u.user_id;
+      const tok = u.token || "";
+      const tokShort = tok ? (tok.length > 18 ? (tok.slice(0, 18) + "…") : tok) : "";
+      const created = u.created_at || "";
+      const updated = u.updated_at || "";
+      const tCreated = u.token_created_at || "";
+      const tUsed = u.token_last_used_at || "";
+      return `<tr>
+        <td><code>${esc(uid)}</code></td>
+        <td class="muted">${esc(created)}</td>
+        <td class="muted">${esc(updated)}</td>
+        <td>
+          <div class="row" style="gap:8px">
+            <code title="${esc(tok)}">${esc(tokShort)}</code>
+            <button class="btn" data-copy="${esc(tok)}">Copy</button>
+            <button class="btn" data-issue="${esc(uid)}">Issue token</button>
+          </div>
+          <div class="muted" style="margin-top:6px">token_created_at: ${esc(tCreated)}; last_used_at: ${esc(tUsed)}</div>
+        </td>
+      </tr>`;
+    }
+    out.innerHTML = `
+      <table>
+        <thead><tr><th>User</th><th>Created</th><th>Updated</th><th>Token</th></tr></thead>
+        <tbody>${users.map(row).join("") || "<tr><td colspan=4 class=muted>(none)</td></tr>"}</tbody>
+      </table>
+    `;
+    for (const btn of out.querySelectorAll("button[data-copy]")) {
+      btn.onclick = async () => { await copyText(btn.getAttribute("data-copy") || ""); };
+    }
+    for (const btn of out.querySelectorAll("button[data-issue]")) {
+      btn.onclick = async () => {
+        const uid = btn.getAttribute("data-issue");
+        if (!uid) return;
+        try {
+          const r = await apiJson("/api/v2/users/" + encodeURIComponent(uid) + "/tokens", { method: "POST" });
+          createMsg.textContent = "Issued token:\\n" + fmtJson(r);
+          await refresh();
+        } catch (e) {
+          createMsg.textContent = "Error issuing token: " + (e.status || "") + "\\n" + (e.body || String(e));
+        }
+      };
+    }
+  } catch (e) {
+    out.textContent = "Error: " + (e.status || "") + "\\n" + (e.body || String(e));
+  }
+}
+
+document.getElementById("refresh").onclick = refresh;
+document.getElementById("create-user").onclick = async () => {
+  createMsg.textContent = "Creating...";
+  const user_id = (userIdEl.value || "").trim();
+  const display_name = (displayNameEl.value || "").trim();
+  if (!user_id) { createMsg.textContent = "user_id is required"; return; }
+  const payload = { user_id: user_id };
+  if (display_name) payload.display_name = display_name;
+  try {
+    const r = await apiJson("/api/v2/users", { method: "POST", headers: {"Content-Type":"application/json"}, body: JSON.stringify(payload) });
+    createMsg.textContent = "Created:\\n" + fmtJson(r);
+    userIdEl.value = "";
+    displayNameEl.value = "";
+    await refresh();
+  } catch (e) {
+    createMsg.textContent = "Error: " + (e.status || "") + "\\n" + (e.body || String(e));
+  }
+};
+
+refresh();
+""".strip()
+        return HTMLResponse(content=_page("Admin", "admin", body, script))
diff --git a/src/mvp/py/tests/test_app.py b/src/mvp/py/tests/test_app.py
index 99e2243..b96bf77 100644
--- a/src/mvp/py/tests/test_app.py
+++ b/src/mvp/py/tests/test_app.py
@@ -16,6 +16,11 @@ def _write_config(tmp_path: Path) -> Path:
             "entrypoint_resources": {"worker_node": 1},
             "runtime_env": {"env_vars": {}},
         },
+        "data": {
+            # Avoid touching real /private in tests. Keep ray.shared_root as /private
+            # so existing path validation tests remain unchanged.
+            "user_root": str(tmp_path / "users"),
+        },
         "service": {
             "api": {"host": "127.0.0.1", "port": 0},
             "auth": {"token_env": "MVP_INTERNAL_TOKEN"},
@@ -95,6 +100,17 @@ def test_task_submit_get_cancel_logs_queue(tmp_path: Path, monkeypatch):
         assert r3.status_code == 200
         assert "pending" in r3.json()
 
+        r3b = c.get("/api/v2/tasks?limit=10", headers=headers)
+        assert r3b.status_code == 200
+        assert any(t.get("task_id") == "tid1" for t in r3b.json().get("tasks", []))
+
+        r3c = c.get("/api/v2/tasks?limit=10&offset=0&states=QUEUED", headers=headers)
+        assert r3c.status_code == 200
+        assert all(t.get("state") == "QUEUED" for t in r3c.json().get("tasks", []))
+
+        r3d = c.get("/api/v2/tasks?states=NOPE", headers=headers)
+        assert r3d.status_code == 400
+
         r4 = c.post("/api/v2/tasks/tid1:cancel", headers=headers)
         assert r4.status_code == 200
         assert r4.json()["state"] == "CANCELED"
@@ -118,6 +134,14 @@ def test_task_submit_get_cancel_logs_queue(tmp_path: Path, monkeypatch):
         db.create_attempt(task_id="tid2", attempt_no=1, ray_submission_id="sid2")
         db.set_task_state(task_id="tid2", state="RUNNING", latest_attempt_no=1)
 
+        r6 = c.get("/api/v2/tasks?limit=1&offset=0&states=RUNNING", headers=headers)
+        assert r6.status_code == 200
+        assert any(t.get("task_id") == "tid2" for t in r6.json().get("tasks", []))
+
+        r7 = c.get("/api/v2/tasks?limit=1&offset=1&states=RUNNING", headers=headers)
+        assert r7.status_code == 200
+        assert "has_more" in r7.json()
+
         r5 = c.get("/api/v2/tasks/tid2/logs?tail=1", headers=headers)
         assert r5.status_code == 200
         assert r5.text.strip() == "c"
@@ -163,3 +187,102 @@ def test_submit_rejects_invalid_jobspec(tmp_path: Path, monkeypatch):
     with TestClient(app) as c:
         r = c.post("/api/v2/tasks", headers={"authorization": "Bearer token1"}, data="workload: nope\n")
         assert r.status_code == 400
+
+
+def test_me_sftp_reset_password_disabled_returns_400(tmp_path: Path, monkeypatch):
+    from argus.service import app as app_mod
+
+    cfg_path = _write_config(tmp_path)
+    monkeypatch.setenv("MVP_INTERNAL_TOKEN", "token1")
+
+    class _Scheduler:
+        def __init__(self, **kwargs):
+            self.tool = object()
+
+        def run_forever(self, stop_flag):
+            return None
+
+    monkeypatch.setattr(app_mod, "Scheduler", _Scheduler)
+    app = app_mod.create_app(str(cfg_path))
+
+    # seed user + token
+    from argus.service.config import V2Config
+    from argus.service.db import Db
+
+    root = yaml.safe_load(cfg_path.read_text(encoding="utf-8"))
+    v2_cfg = V2Config.from_root_dict(root)
+    db = Db(v2_cfg.sqlite.db_path)
+    db.init()
+    db.create_user(user_id="u1", display_name=None)
+    token = db.issue_token(user_id="u1")
+
+    with TestClient(app) as c:
+        r = c.post("/api/v2/me/sftp:reset_password", headers={"authorization": f"Bearer {token}"})
+        assert r.status_code == 400
+
+
+def test_me_sftp_reset_password_enabled_returns_password(tmp_path: Path, monkeypatch):
+    from argus.service import app as app_mod
+
+    cfg = yaml.safe_load(_write_config(tmp_path).read_text(encoding="utf-8"))
+    cfg["data"]["sftpgo"] = {
+        "enabled": True,
+        "host": "127.0.0.1",
+        "sftp_port": 2022,
+        "admin_api_base": "http://127.0.0.1:8081",
+        "admin_user": "admin",
+        "admin_password_env": "SFTPGO_ADMIN_PASSWORD",
+    }
+    cfg_path = tmp_path / "cfg_sftp.yaml"
+    cfg_path.write_text(yaml.safe_dump(cfg), encoding="utf-8")
+
+    monkeypatch.setenv("MVP_INTERNAL_TOKEN", "token1")
+    monkeypatch.setenv("SFTPGO_ADMIN_PASSWORD", "pw1")
+
+    class _FakeSFTPGo:
+        def __init__(self, **kwargs):
+            self.reset = []
+            self.enabled = []
+
+        def reset_password(self, username: str, new_password: str, home_dir: str):
+            assert username
+            assert new_password
+            assert home_dir
+            self.reset.append((username, home_dir))
+
+        def enable_user(self, username: str, home_dir: str):
+            self.enabled.append((username, home_dir))
+
+    fake_client = _FakeSFTPGo()
+
+    class _FakeSFTPGoFactory:
+        def __call__(self, **kwargs):
+            return fake_client
+
+    class _Scheduler:
+        def __init__(self, **kwargs):
+            self.tool = object()
+
+        def run_forever(self, stop_flag):
+            return None
+
+    monkeypatch.setattr(app_mod, "Scheduler", _Scheduler)
+    monkeypatch.setattr(app_mod, "SFTPGoAdminClient", _FakeSFTPGoFactory())
+    app = app_mod.create_app(str(cfg_path))
+
+    # seed user in DB
+    from argus.service.db import Db
+    from argus.service.config import V2Config
+
+    v2_cfg = V2Config.from_root_dict(cfg)
+    db = Db(v2_cfg.sqlite.db_path)
+    db.init()
+    db.create_user(user_id="u1", display_name=None)
+    token = db.issue_token(user_id="u1")
+
+    with TestClient(app) as c:
+        r = c.post("/api/v2/me/sftp:reset_password", headers={"authorization": f"Bearer {token}"})
+        assert r.status_code == 200
+        j = r.json()
+        assert j["user_id"] == "u1"
+        assert isinstance(j["password"], str) and len(j["password"]) >= 8
diff --git a/src/mvp/py/tests/test_janitor.py b/src/mvp/py/tests/test_janitor.py
new file mode 100644
index 0000000..8528555
--- /dev/null
+++ b/src/mvp/py/tests/test_janitor.py
@@ -0,0 +1,225 @@
+from __future__ import annotations
+
+from datetime import datetime, timedelta, timezone
+from pathlib import Path
+
+from argus.service.db import Db
+from argus.service.janitor import JobsJanitor
+
+
+def _iso_z(dt: datetime) -> str:
+    return dt.astimezone(timezone.utc).replace(microsecond=0).isoformat().replace("+00:00", "Z")
+
+
+def _mk_job_dir(user_root: Path, user_id: str, sid: str) -> Path:
+    p = user_root / user_id / "jobs" / sid
+    p.mkdir(parents=True, exist_ok=True)
+    (p / "marker.txt").write_text("x", encoding="utf-8")
+    return p
+
+
+def _mk_trash_dir(user_root: Path, user_id: str, sid: str) -> Path:
+    p = user_root / user_id / "trash" / "jobs" / sid
+    p.mkdir(parents=True, exist_ok=True)
+    (p / "marker.txt").write_text("x", encoding="utf-8")
+    return p
+
+
+def test_janitor_moves_jobs_to_trash_after_3_days(tmp_path: Path) -> None:
+    db_path = tmp_path / "mvp.sqlite3"
+    user_root = tmp_path / "users"
+    db = Db(str(db_path))
+    db.init()
+
+    task_id = "t1"
+    user_id = "alice"
+    sid = "sid-a01"
+    db.create_task_v25(task_id=task_id, user_id=user_id, workload="sft", jobspec_yaml="{}", nnodes=1, n_gpus_per_node=1)
+    db.create_attempt(task_id=task_id, attempt_no=1, ray_submission_id=sid)
+
+    now = datetime(2025, 1, 10, tzinfo=timezone.utc)
+    ended = now - timedelta(days=4)
+    db.update_attempt(task_id=task_id, attempt_no=1, end_time=_iso_z(ended), ray_status="SUCCEEDED")
+
+    src = _mk_job_dir(user_root, user_id, sid)
+    jan = JobsJanitor(db=db, user_root=str(user_root), trash_after_days=3, purge_after_days=7, interval_s=1)
+    jan.tick_once(now=now)
+
+    assert not src.exists()
+    dst = user_root / user_id / "trash" / "jobs" / sid
+    assert dst.exists()
+    assert (dst / "marker.txt").read_text(encoding="utf-8") == "x"
+
+
+def test_janitor_purges_from_trash_after_7_days(tmp_path: Path) -> None:
+    db_path = tmp_path / "mvp.sqlite3"
+    user_root = tmp_path / "users"
+    db = Db(str(db_path))
+    db.init()
+
+    task_id = "t2"
+    user_id = "alice"
+    sid = "sid-a01"
+    db.create_task_v25(task_id=task_id, user_id=user_id, workload="ppo", jobspec_yaml="{}", nnodes=1, n_gpus_per_node=1)
+    db.create_attempt(task_id=task_id, attempt_no=1, ray_submission_id=sid)
+
+    now = datetime(2025, 1, 10, tzinfo=timezone.utc)
+    ended = now - timedelta(days=8)
+    db.update_attempt(task_id=task_id, attempt_no=1, end_time=_iso_z(ended), ray_status="FAILED")
+
+    _mk_trash_dir(user_root, user_id, sid)
+    jan = JobsJanitor(db=db, user_root=str(user_root), trash_after_days=3, purge_after_days=7, interval_s=1)
+    jan.tick_once(now=now)
+
+    dst = user_root / user_id / "trash" / "jobs" / sid
+    assert not dst.exists()
+
+
+def test_janitor_does_not_touch_recent_jobs(tmp_path: Path) -> None:
+    db_path = tmp_path / "mvp.sqlite3"
+    user_root = tmp_path / "users"
+    db = Db(str(db_path))
+    db.init()
+
+    task_id = "t3"
+    user_id = "alice"
+    sid = "sid-a01"
+    db.create_task_v25(task_id=task_id, user_id=user_id, workload="grpo", jobspec_yaml="{}", nnodes=1, n_gpus_per_node=1)
+    db.create_attempt(task_id=task_id, attempt_no=1, ray_submission_id=sid)
+
+    now = datetime(2025, 1, 10, tzinfo=timezone.utc)
+    ended = now - timedelta(days=1)
+    db.update_attempt(task_id=task_id, attempt_no=1, end_time=_iso_z(ended), ray_status="FAILED")
+
+    src = _mk_job_dir(user_root, user_id, sid)
+    jan = JobsJanitor(db=db, user_root=str(user_root), trash_after_days=3, purge_after_days=7, interval_s=1)
+    jan.tick_once(now=now)
+
+    assert src.exists()
+    assert not (user_root / user_id / "trash" / "jobs" / sid).exists()
+
+
+def test_janitor_skips_tasks_without_user_id(tmp_path: Path) -> None:
+    db_path = tmp_path / "mvp.sqlite3"
+    user_root = tmp_path / "users"
+    db = Db(str(db_path))
+    db.init()
+
+    task_id = "legacy"
+    sid = "sid-legacy"
+    db.create_task(task_id=task_id, workload="sft", jobspec_yaml="{}", nnodes=1, n_gpus_per_node=1)
+    db.create_attempt(task_id=task_id, attempt_no=1, ray_submission_id=sid)
+
+    now = datetime(2025, 1, 10, tzinfo=timezone.utc)
+    ended = now - timedelta(days=10)
+    db.update_attempt(task_id=task_id, attempt_no=1, end_time=_iso_z(ended), ray_status="SUCCEEDED")
+
+    # Even if someone created a matching directory under user_root, janitor shouldn't touch it because user_id is NULL.
+    src = _mk_job_dir(user_root, "alice", sid)
+    jan = JobsJanitor(db=db, user_root=str(user_root), trash_after_days=3, purge_after_days=7, interval_s=1)
+    jan.tick_once(now=now)
+    assert src.exists()
+
+
+def test_janitor_validates_retention_days(tmp_path: Path) -> None:
+    db = Db(str(tmp_path / "mvp.sqlite3"))
+    db.init()
+    try:
+        JobsJanitor(db=db, user_root="/tmp", trash_after_days=-1, purge_after_days=7, interval_s=1)
+        raise AssertionError("expected ValueError")
+    except ValueError:
+        pass
+    try:
+        JobsJanitor(db=db, user_root="/tmp", trash_after_days=3, purge_after_days=1, interval_s=1)
+        raise AssertionError("expected ValueError")
+    except ValueError:
+        pass
+
+
+def test_janitor_noop_when_disabled(tmp_path: Path) -> None:
+    db = Db(str(tmp_path / "mvp.sqlite3"))
+    db.init()
+    user_root = tmp_path / "users"
+    jan = JobsJanitor(db=db, user_root=str(user_root), trash_after_days=0, purge_after_days=0, interval_s=1)
+    jan.tick_once(now=datetime(2025, 1, 1, tzinfo=timezone.utc))
+
+
+def test_janitor_handles_invalid_end_time_and_missing_fields(tmp_path: Path) -> None:
+    db_path = tmp_path / "mvp.sqlite3"
+    user_root = tmp_path / "users"
+    db = Db(str(db_path))
+    db.init()
+
+    now = datetime(2025, 1, 10, tzinfo=timezone.utc)
+    cutoff_ended = now - timedelta(days=10)
+
+    # Missing end_time (empty string) => should be skipped.
+    db.create_task_v25(task_id="t4", user_id="alice", workload="sft", jobspec_yaml="{}", nnodes=1, n_gpus_per_node=1)
+    db.create_attempt(task_id="t4", attempt_no=1, ray_submission_id="sid-empty")
+    db.update_attempt(task_id="t4", attempt_no=1, end_time="", ray_status="SUCCEEDED")
+
+    # Invalid ISO but still lexicographically <= cutoff => should be skipped.
+    db.create_task_v25(task_id="t5", user_id="alice", workload="sft", jobspec_yaml="{}", nnodes=1, n_gpus_per_node=1)
+    db.create_attempt(task_id="t5", attempt_no=1, ray_submission_id="sid-bad")
+    db.update_attempt(task_id="t5", attempt_no=1, end_time="2025-01-01T00:00:00ZZ", ray_status="FAILED")
+
+    _mk_job_dir(user_root, "alice", "sid-empty")
+    _mk_job_dir(user_root, "alice", "sid-bad")
+    jan = JobsJanitor(db=db, user_root=str(user_root), trash_after_days=3, purge_after_days=7, interval_s=1)
+    jan.tick_once(now=cutoff_ended + timedelta(days=10))
+
+    assert (user_root / "alice" / "jobs" / "sid-empty").exists()
+    assert (user_root / "alice" / "jobs" / "sid-bad").exists()
+
+
+def test_janitor_purge_moves_from_jobs_then_deletes(tmp_path: Path, monkeypatch) -> None:
+    db_path = tmp_path / "mvp.sqlite3"
+    user_root = tmp_path / "users"
+    db = Db(str(db_path))
+    db.init()
+
+    task_id = "t6"
+    user_id = "alice"
+    sid = "sid-a01"
+    db.create_task_v25(task_id=task_id, user_id=user_id, workload="ppo", jobspec_yaml="{}", nnodes=1, n_gpus_per_node=1)
+    db.create_attempt(task_id=task_id, attempt_no=1, ray_submission_id=sid)
+
+    now = datetime(2025, 1, 10, tzinfo=timezone.utc)
+    ended = now - timedelta(days=9)
+    db.update_attempt(task_id=task_id, attempt_no=1, end_time=_iso_z(ended), ray_status="SUCCEEDED")
+
+    src = _mk_job_dir(user_root, user_id, sid)
+    jan = JobsJanitor(db=db, user_root=str(user_root), trash_after_days=3, purge_after_days=7, interval_s=1)
+
+    jan.tick_once(now=now)
+    assert not src.exists()
+    assert not (user_root / user_id / "trash" / "jobs" / sid).exists()
+
+
+def test_janitor_run_forever_requires_event_like(tmp_path: Path) -> None:
+    db = Db(str(tmp_path / "mvp.sqlite3"))
+    db.init()
+    jan = JobsJanitor(db=db, user_root=str(tmp_path / "users"), trash_after_days=3, purge_after_days=7, interval_s=1)
+    try:
+        jan.run_forever(object())
+        raise AssertionError("expected ValueError")
+    except ValueError:
+        pass
+
+
+def test_janitor_run_forever_survives_tick_exceptions(tmp_path: Path, monkeypatch) -> None:
+    db = Db(str(tmp_path / "mvp.sqlite3"))
+    db.init()
+    jan = JobsJanitor(db=db, user_root=str(tmp_path / "users"), trash_after_days=3, purge_after_days=7, interval_s=1)
+
+    class Flag:
+        def __init__(self) -> None:
+            self.n = 0
+
+        def is_set(self) -> bool:
+            self.n += 1
+            return self.n >= 2
+
+    monkeypatch.setattr(jan, "tick_once", lambda **_: (_ for _ in ()).throw(RuntimeError("boom")))
+    monkeypatch.setattr("argus.service.janitor.time.sleep", lambda *_: None)
+    jan.run_forever(Flag())
diff --git a/src/mvp/py/tests/test_service_config.py b/src/mvp/py/tests/test_service_config.py
index c6ceb0e..9fe7226 100644
--- a/src/mvp/py/tests/test_service_config.py
+++ b/src/mvp/py/tests/test_service_config.py
@@ -38,3 +38,18 @@ def test_v2_config_requires_mappings():
         V2Config.from_root_dict({"service": ["nope"]})
     with pytest.raises(ValueError, match="config\\.service\\.\\{api,auth,sqlite,scheduler\\} must be mappings"):
         V2Config.from_root_dict({"service": {"api": [1], "auth": {}, "sqlite": {}, "scheduler": {}}})
+
+
+def test_v2_config_requires_data_mappings():
+    from argus.service.config import V2Config
+
+    base = {
+        "ray": {"shared_root": "/private"},
+        "service": {"api": {}, "auth": {}, "sqlite": {}, "scheduler": {}},
+    }
+
+    with pytest.raises(ValueError, match="config\\.data must be a mapping"):
+        V2Config.from_root_dict({**base, "data": ["nope"]})
+
+    with pytest.raises(ValueError, match="config\\.data\\.\\{sftpgo,retention\\} must be mappings"):
+        V2Config.from_root_dict({**base, "data": {"sftpgo": ["x"], "retention": {}}})
diff --git a/src/mvp/py/tests/test_sftpgo.py b/src/mvp/py/tests/test_sftpgo.py
new file mode 100644
index 0000000..bc0440f
--- /dev/null
+++ b/src/mvp/py/tests/test_sftpgo.py
@@ -0,0 +1,322 @@
+from __future__ import annotations
+
+from pathlib import Path
+
+import yaml
+from fastapi.testclient import TestClient
+
+
+def _write_config(tmp_path: Path, *, enabled: bool) -> Path:
+    cfg = {
+        "ray": {
+            "address": "http://127.0.0.1:8265",
+            "shared_root": "/private",
+            "entrypoint_resources": {"worker_node": 1},
+            "runtime_env": {"env_vars": {}},
+        },
+        "data": {
+            "user_root": str(tmp_path / "users"),
+            "sftpgo": {
+                "enabled": bool(enabled),
+                "admin_api_base": "http://127.0.0.1:8081/api/v2",
+                "admin_user": "admin",
+                "admin_password_env": "SFTPGO_ADMIN_PASSWORD",
+                "host": "h1.internal",
+                "sftp_port": 2022,
+            },
+        },
+        "service": {
+            "api": {"host": "127.0.0.1", "port": 0},
+            "auth": {"token_env": "MVP_INTERNAL_TOKEN"},
+            "sqlite": {"db_path": str(tmp_path / "mvp.sqlite3")},
+            "scheduler": {"tick_s": 1, "retry_interval_s": 1, "max_running_tasks": 1},
+        },
+    }
+    p = tmp_path / "cfg.yaml"
+    p.write_text(yaml.safe_dump(cfg), encoding="utf-8")
+    return p
+
+
+def test_create_user_calls_sftpgo_when_enabled(tmp_path: Path, monkeypatch):
+    from argus.service import app as app_mod
+
+    cfg_path = _write_config(tmp_path, enabled=True)
+    monkeypatch.setenv("MVP_INTERNAL_TOKEN", "adm1")
+    monkeypatch.setenv("SFTPGO_ADMIN_PASSWORD", "pw1")
+
+    calls = {"create": [], "disable": [], "reset": []}
+
+    class _Client:
+        def __init__(self, **kwargs):
+            pass
+
+        def create_user(self, *, username: str, password: str, home_dir: str) -> None:
+            calls["create"].append((username, password, home_dir))
+
+        def enable_user(self, *, username: str, home_dir: str) -> None:
+            # Not used in this test, but required by app for idempotent upsert.
+            return None
+
+        def disable_user(self, *, username: str, home_dir: str) -> None:
+            calls["disable"].append(username)
+
+        def reset_password(self, *, username: str, new_password: str, home_dir: str) -> None:
+            calls["reset"].append((username, new_password))
+
+    monkeypatch.setattr(app_mod, "SFTPGoAdminClient", _Client)
+
+    class _Scheduler:
+        def __init__(self, **kwargs):
+            self.tool = object()
+
+        def run_forever(self, stop_flag):
+            return None
+
+    monkeypatch.setattr(app_mod, "Scheduler", _Scheduler)
+    app = app_mod.create_app(str(cfg_path))
+
+    admin_headers = {"authorization": "Bearer adm1"}
+    with TestClient(app) as c:
+        r = c.post("/api/v2/users", headers=admin_headers, json={"user_id": "alice"})
+        assert r.status_code == 200
+        assert calls["create"]
+        username, password, home_dir = calls["create"][-1]
+        assert username == "alice"
+        assert password
+        assert home_dir.endswith("/users/alice")
+
+        r2 = c.post("/api/v2/users/alice:disable", headers=admin_headers)
+        assert r2.status_code == 200
+        assert calls["disable"] == ["alice"]
+
+        r3 = c.post("/api/v2/users/alice/sftp:reset_password", headers=admin_headers)
+        assert r3.status_code == 200
+        body = r3.json()
+        assert body["user_id"] == "alice"
+        assert body["password"]
+        assert calls["reset"] and calls["reset"][-1][0] == "alice"
+
+
+def test_create_user_upserts_when_sftpgo_user_already_exists(tmp_path: Path, monkeypatch):
+    from argus.service import app as app_mod
+    from argus.service.sftpgo import SFTPGoError
+
+    cfg_path = _write_config(tmp_path, enabled=True)
+    monkeypatch.setenv("MVP_INTERNAL_TOKEN", "adm1")
+    monkeypatch.setenv("SFTPGO_ADMIN_PASSWORD", "pw1")
+
+    calls = {"create": 0, "reset": [], "enable": []}
+
+    class _Client:
+        def __init__(self, **kwargs):
+            pass
+
+        def create_user(self, *, username: str, password: str, home_dir: str) -> None:
+            calls["create"] += 1
+            raise SFTPGoError("sftpgo http error: 409 Conflict")
+
+        def reset_password(self, *, username: str, new_password: str, home_dir: str) -> None:
+            calls["reset"].append((username, new_password))
+
+        def enable_user(self, *, username: str, home_dir: str) -> None:
+            calls["enable"].append(username)
+
+        def disable_user(self, *, username: str, home_dir: str) -> None:
+            return None
+
+    monkeypatch.setattr(app_mod, "SFTPGoAdminClient", _Client)
+
+    class _Scheduler:
+        def __init__(self, **kwargs):
+            self.tool = object()
+
+        def run_forever(self, stop_flag):
+            return None
+
+    monkeypatch.setattr(app_mod, "Scheduler", _Scheduler)
+    app = app_mod.create_app(str(cfg_path))
+
+    admin_headers = {"authorization": "Bearer adm1"}
+    with TestClient(app) as c:
+        r = c.post("/api/v2/users", headers=admin_headers, json={"user_id": "bob"})
+        assert r.status_code == 200
+        body = r.json()
+        assert body["user_id"] == "bob"
+        assert body.get("sftp", {}).get("password")
+        assert calls["create"] == 1
+        assert calls["reset"] and calls["reset"][-1][0] == "bob"
+        assert calls["enable"] == ["bob"]
+
+
+def test_sftpgo_enabled_requires_admin_password_env(tmp_path: Path, monkeypatch):
+    from argus.service import app as app_mod
+
+    cfg_path = _write_config(tmp_path, enabled=True)
+    monkeypatch.setenv("MVP_INTERNAL_TOKEN", "adm1")
+    monkeypatch.delenv("SFTPGO_ADMIN_PASSWORD", raising=False)
+
+    class _Scheduler:
+        def __init__(self, **kwargs):
+            self.tool = object()
+
+        def run_forever(self, stop_flag):
+            return None
+
+    monkeypatch.setattr(app_mod, "Scheduler", _Scheduler)
+    app = app_mod.create_app(str(cfg_path))
+
+    admin_headers = {"authorization": "Bearer adm1"}
+    with TestClient(app) as c:
+        r = c.post("/api/v2/users", headers=admin_headers, json={"user_id": "alice"})
+        assert r.status_code == 500
+        assert "SFTPGO_ADMIN_PASSWORD" in r.text
+
+
+def test_sftpgo_admin_client_builds_requests(monkeypatch):
+    import json
+    from urllib.request import Request
+
+    from argus.service.sftpgo import SFTPGoAdminClient
+    import argus.service.sftpgo as mod
+
+    seen: list[Request] = []
+
+    class _Resp:
+        def __init__(self, body: bytes = b"ok"):
+            self._body = body
+
+        def __enter__(self):
+            return self
+
+        def __exit__(self, exc_type, exc, tb):
+            return False
+
+        def read(self):
+            return self._body
+
+    def fake_urlopen(req, timeout=0):
+        seen.append(req)
+        if req.full_url.endswith("/token"):
+            return _Resp(body=b'{"access_token":"t1","expires_at":0}')
+        # allow folder creates to be idempotent in tests
+        if req.full_url.endswith("/folders") and req.get_method() == "POST":
+            return _Resp(body=b'{"message":"created"}')
+        if req.full_url.endswith("/users/alice") and req.get_method() == "GET":
+            return _Resp(
+                body=b'{"username":"alice","status":1,"home_dir":"/private/users/alice","uid":0,"gid":0,"permissions":{"/":["*"]},"virtual_folders":[]}'
+            )
+        return _Resp()
+
+    monkeypatch.setattr(mod, "urlopen", fake_urlopen)
+
+    c = SFTPGoAdminClient(admin_api_base="http://sftpgo.local/api/v2", admin_user="admin", admin_password="pw", common_root="/private/common")
+    c.create_user(username="alice", password="p1", home_dir="/private/users/alice")
+    c.disable_user(username="alice", home_dir="/private/users/alice")
+    c.enable_user(username="alice", home_dir="/private/users/alice")
+    c.reset_password(username="alice", new_password="p2", home_dir="/private/users/alice")
+
+    # Each operation fetches a token, then issues a request.
+    # For create_user: token + ensure folders (2 POSTs) + create user
+    # For disable/enable/reset: token + ensure folders (2 POSTs) + GET user + PUT user
+    assert len(seen) == 1 + 2 + 1 + 3 * (1 + 2 + 1 + 1)
+    assert seen[0].full_url.endswith("/token")
+    assert seen[0].headers.get("Authorization", "").startswith("Basic ")
+    # create_user
+    assert seen[1].full_url.endswith("/folders")
+    assert seen[2].full_url.endswith("/folders")
+    assert seen[3].full_url.endswith("/users")
+    assert seen[3].headers.get("Authorization", "") == "Bearer t1"
+    created = json.loads(seen[3].data.decode("utf-8"))
+    assert created["username"] == "alice"
+    assert "/common" in created.get("permissions", {})
+    assert "/common/datasets" in created.get("permissions", {})
+    assert "/common/hf" in created.get("permissions", {})
+
+    # disable_user
+    assert seen[4].full_url.endswith("/token")
+    assert seen[5].full_url.endswith("/folders")
+    assert seen[6].full_url.endswith("/folders")
+    assert seen[7].full_url.endswith("/users/alice")
+    assert seen[7].get_method() == "GET"
+    assert seen[8].full_url.endswith("/users/alice")
+    assert seen[8].get_method() == "PUT"
+
+    # enable_user
+    assert seen[9].full_url.endswith("/token")
+    assert seen[10].full_url.endswith("/folders")
+    assert seen[11].full_url.endswith("/folders")
+    assert seen[12].full_url.endswith("/users/alice")
+    assert seen[12].get_method() == "GET"
+    assert seen[13].full_url.endswith("/users/alice")
+    assert seen[13].get_method() == "PUT"
+
+    # reset_password
+    assert seen[14].full_url.endswith("/token")
+    assert seen[15].full_url.endswith("/folders")
+    assert seen[16].full_url.endswith("/folders")
+    assert seen[17].full_url.endswith("/users/alice")
+    assert seen[17].get_method() == "GET"
+    assert seen[18].full_url.endswith("/users/alice")
+    assert seen[18].get_method() == "PUT"
+
+
+def test_sftpgo_admin_client_http_error_raises(monkeypatch):
+    from urllib.error import HTTPError
+
+    from argus.service.sftpgo import SFTPGoAdminClient, SFTPGoError
+    import argus.service.sftpgo as mod
+
+    def fake_urlopen(req, timeout=0):
+        if req.full_url.endswith("/token"):
+            class _Resp:
+                def __enter__(self):
+                    return self
+
+                def __exit__(self, exc_type, exc, tb):
+                    return False
+
+                def read(self):
+                    return b'{"access_token":"t1","expires_at":0}'
+
+            return _Resp()
+        raise HTTPError(req.full_url, 500, "boom", hdrs=None, fp=None)
+
+    monkeypatch.setattr(mod, "urlopen", fake_urlopen)
+
+    c = SFTPGoAdminClient(admin_api_base="http://sftpgo.local/api/v2", admin_user="admin", admin_password="pw", common_root="/private/common")
+    try:
+        c.create_user(username="alice", password="p1", home_dir="/private/users/alice")
+        assert False, "expected SFTPGoError"
+    except SFTPGoError as e:
+        assert "http error" in str(e)
+
+
+def test_sftpgo_admin_client_url_error_raises(monkeypatch):
+    from urllib.error import URLError
+
+    from argus.service.sftpgo import SFTPGoAdminClient, SFTPGoError
+    import argus.service.sftpgo as mod
+
+    def fake_urlopen(req, timeout=0):
+        if req.full_url.endswith("/token"):
+            class _Resp:
+                def __enter__(self):
+                    return self
+
+                def __exit__(self, exc_type, exc, tb):
+                    return False
+
+                def read(self):
+                    return b'{"access_token":"t1","expires_at":0}'
+
+            return _Resp()
+        raise URLError("no route")
+
+    monkeypatch.setattr(mod, "urlopen", fake_urlopen)
+
+    c = SFTPGoAdminClient(admin_api_base="http://sftpgo.local/api/v2", admin_user="admin", admin_password="pw")
+    try:
+        c.disable_user(username="alice", home_dir="/private/users/alice")
+        assert False, "expected SFTPGoError"
+    except SFTPGoError as e:
+        assert "connection error" in str(e)
diff --git a/src/mvp/py/tests/test_ui.py b/src/mvp/py/tests/test_ui.py
new file mode 100644
index 0000000..ab4ebd5
--- /dev/null
+++ b/src/mvp/py/tests/test_ui.py
@@ -0,0 +1,78 @@
+from __future__ import annotations
+
+from pathlib import Path
+
+from fastapi.testclient import TestClient
+
+from argus.service.app import create_app
+
+
+def _write_config(tmp_path: Path) -> Path:
+    p = tmp_path / "cfg.yaml"
+    p.write_text(
+        """
+ray:
+  address: "http://127.0.0.1:8265"
+  shared_root: "/private"
+  entrypoint_num_cpus: 1
+  entrypoint_resources: { worker_node: 1 }
+  runtime_env: { env_vars: { PYTHONUNBUFFERED: "1" } }
+service:
+  api: { host: "127.0.0.1", port: 8080 }
+  auth: { token_env: "MVP_INTERNAL_TOKEN" }
+  sqlite: { db_path: "%(db)s" }
+data:
+  user_root: "%(users)s"
+  sftpgo: { enabled: false }
+  retention: { jobs_trash_after_days: 3, jobs_purge_after_days: 7, janitor_interval_s: 3600 }
+"""
+        % {"db": str(tmp_path / "mvp.sqlite3"), "users": str(tmp_path / "users")}
+    )
+    return p
+
+
+def test_ui_routes_render_200(tmp_path, monkeypatch):
+    cfg = _write_config(tmp_path)
+    monkeypatch.setenv("MVP_INTERNAL_TOKEN", "admin-token")
+    app = create_app(str(cfg))
+    c = TestClient(app)
+
+    for path in (
+        "/ui",
+        "/ui/login",
+        "/ui/tasks",
+        "/ui/tasks/new",
+        "/ui/data",
+        "/ui/admin",
+        "/ui/tasks/any-task-id",
+        "/ui/tasks/any-task-id/logs",
+    ):
+        r = c.get(path, allow_redirects=True)
+        assert r.status_code == 200
+        assert "<html" in r.text.lower()
+
+
+def test_ui_contains_sidebar_links(tmp_path, monkeypatch):
+    cfg = _write_config(tmp_path)
+    monkeypatch.setenv("MVP_INTERNAL_TOKEN", "admin-token")
+    app = create_app(str(cfg))
+    c = TestClient(app)
+
+    r = c.get("/ui/tasks")
+    assert r.status_code == 200
+    for link in ("/ui/tasks", "/ui/tasks/new", "/ui/data", "/ui/login", "/ui/admin"):
+        assert link in r.text
+    assert "Ray Dashboard" in r.text
+
+
+def test_ui_task_detail_shows_ids(tmp_path, monkeypatch):
+    cfg = _write_config(tmp_path)
+    monkeypatch.setenv("MVP_INTERNAL_TOKEN", "admin-token")
+    app = create_app(str(cfg))
+    c = TestClient(app)
+
+    task_id = "mvp3-ppo-20250101-000000-aaaa"
+    r = c.get(f"/ui/tasks/{task_id}")
+    assert r.status_code == 200
+    assert task_id in r.text
+    assert f"/ui/tasks/{task_id}/logs" in r.text
diff --git a/src/mvp/py/tests/test_users.py b/src/mvp/py/tests/test_users.py
index 70ca876..92422d2 100644
--- a/src/mvp/py/tests/test_users.py
+++ b/src/mvp/py/tests/test_users.py
@@ -14,6 +14,11 @@ def _write_config(tmp_path: Path) -> Path:
             "entrypoint_resources": {"worker_node": 1},
             "runtime_env": {"env_vars": {}},
         },
+        "data": {
+            # Avoid touching real /private in tests. Keep ray.shared_root as /private
+            # so existing path validation tests remain unchanged.
+            "user_root": str(tmp_path / "users"),
+        },
         "service": {
             "api": {"host": "127.0.0.1", "port": 0},
             "auth": {"token_env": "MVP_INTERNAL_TOKEN"},
@@ -59,6 +64,9 @@ def test_admin_create_user_issue_token_and_disabled_rejected(tmp_path: Path, mon
 
     admin_headers = {"authorization": "Bearer adm1"}
     with TestClient(app) as c:
+        # list users requires admin
+        assert c.get("/api/v2/users", headers={"authorization": "Bearer nope"}).status_code in (401, 403)
+
         r1 = c.post("/api/v2/users", headers=admin_headers, json={"user_id": "alice", "display_name": "Alice"})
         assert r1.status_code == 200
         assert r1.json()["user_id"] == "alice"
@@ -68,6 +76,11 @@ def test_admin_create_user_issue_token_and_disabled_rejected(tmp_path: Path, mon
         token = r2.json()["token"]
         assert token
 
+        r2b = c.get("/api/v2/users", headers=admin_headers)
+        assert r2b.status_code == 200
+        users = r2b.json()["users"]
+        assert any(u.get("user_id") == "alice" for u in users)
+
         # non-admin token can access regular endpoints
         r3 = c.get("/api/v2/queue", headers={"authorization": f"Bearer {token}"})
         assert r3.status_code == 200
@@ -177,3 +190,165 @@ def test_submit_rejects_non_common_inputs(tmp_path: Path, monkeypatch):
         )
         assert r.status_code == 400
         assert "code_path must start with /private/common/" in r.text
+
+
+def test_submit_accepts_user_dataset_paths_and_local_model_paths(tmp_path: Path, monkeypatch):
+    from argus.service import app as app_mod
+
+    cfg_path = _write_config(tmp_path)
+    monkeypatch.setenv("MVP_INTERNAL_TOKEN", "adm1")
+
+    class _Scheduler:
+        def __init__(self, **kwargs):
+            self.tool = object()
+
+        def run_forever(self, stop_flag):
+            return None
+
+    monkeypatch.setattr(app_mod, "Scheduler", _Scheduler)
+    app = app_mod.create_app(str(cfg_path))
+
+    admin_headers = {"authorization": "Bearer adm1"}
+    with TestClient(app) as c:
+        assert c.post("/api/v2/users", headers=admin_headers, json={"user_id": "alice"}).status_code == 200
+        alice_tok = c.post("/api/v2/users/alice/tokens", headers=admin_headers).json()["token"]
+        alice_headers = {"authorization": f"Bearer {alice_tok}"}
+
+        # User dataset paths are allowed.
+        r1 = c.post(
+            "/api/v2/tasks",
+            headers=alice_headers,
+            data=(
+                "workload: ppo\n"
+                "code_path: /private/common/code/verl\n"
+                "model_id: Qwen/Qwen2.5-0.5B-Instruct\n"
+                "train_file: /private/users/alice/datasets/t\n"
+            ),
+        )
+        assert r1.status_code == 200
+
+        # Local model paths under user models/ are allowed (no TaskSpec schema change).
+        r2 = c.post(
+            "/api/v2/tasks",
+            headers=alice_headers,
+            data=(
+                "workload: ppo\n"
+                "code_path: /private/common/code/verl\n"
+                "model_id: /private/users/alice/models/m1\n"
+                "train_file: /private/common/datasets/t\n"
+            ),
+        )
+        assert r2.status_code == 200
+
+
+def test_submit_rejects_cross_user_paths_and_bad_local_model_dirs(tmp_path: Path, monkeypatch):
+    from argus.service import app as app_mod
+
+    cfg_path = _write_config(tmp_path)
+    monkeypatch.setenv("MVP_INTERNAL_TOKEN", "adm1")
+
+    class _Scheduler:
+        def __init__(self, **kwargs):
+            self.tool = object()
+
+        def run_forever(self, stop_flag):
+            return None
+
+    monkeypatch.setattr(app_mod, "Scheduler", _Scheduler)
+    app = app_mod.create_app(str(cfg_path))
+
+    admin_headers = {"authorization": "Bearer adm1"}
+    with TestClient(app) as c:
+        assert c.post("/api/v2/users", headers=admin_headers, json={"user_id": "alice"}).status_code == 200
+        assert c.post("/api/v2/users", headers=admin_headers, json={"user_id": "bob"}).status_code == 200
+        alice_tok = c.post("/api/v2/users/alice/tokens", headers=admin_headers).json()["token"]
+        bob_tok = c.post("/api/v2/users/bob/tokens", headers=admin_headers).json()["token"]
+        bob_headers = {"authorization": f"Bearer {bob_tok}"}
+
+        # Cross-user dataset path should be rejected.
+        r1 = c.post(
+            "/api/v2/tasks",
+            headers=bob_headers,
+            data=(
+                "workload: ppo\n"
+                "code_path: /private/common/code/verl\n"
+                "model_id: Qwen/Qwen2.5-0.5B-Instruct\n"
+                "train_file: /private/users/alice/datasets/t\n"
+            ),
+        )
+        assert r1.status_code == 400
+        assert "/private/users/bob/datasets/" in r1.text
+
+        # Local model path must be under models/.
+        r2 = c.post(
+            "/api/v2/tasks",
+            headers=bob_headers,
+            data=(
+                "workload: ppo\n"
+                "code_path: /private/common/code/verl\n"
+                "model_id: /private/users/bob/jobs/j1/checkpoints\n"
+                "train_file: /private/common/datasets/t\n"
+            ),
+        )
+        assert r2.status_code == 400
+        assert "model_id local path must start with" in r2.text
+
+
+def test_me_returns_paths_and_retention(tmp_path: Path, monkeypatch):
+    from argus.service import app as app_mod
+
+    cfg_path = _write_config(tmp_path)
+    monkeypatch.setenv("MVP_INTERNAL_TOKEN", "adm1")
+
+    class _Scheduler:
+        def __init__(self, **kwargs):
+            self.tool = object()
+
+        def run_forever(self, stop_flag):
+            return None
+
+    monkeypatch.setattr(app_mod, "Scheduler", _Scheduler)
+    app = app_mod.create_app(str(cfg_path))
+
+    admin_headers = {"authorization": "Bearer adm1"}
+    with TestClient(app) as c:
+        assert c.post("/api/v2/users", headers=admin_headers, json={"user_id": "alice"}).status_code == 200
+        alice_tok = c.post("/api/v2/users/alice/tokens", headers=admin_headers).json()["token"]
+
+        r = c.get("/api/v2/me", headers={"authorization": f"Bearer {alice_tok}"})
+        assert r.status_code == 200
+        obj = r.json()
+        assert obj["user_id"] == "alice"
+        assert obj["paths"]["home"].endswith("/users/alice")
+        assert obj["paths"]["jobs"].endswith("/users/alice/jobs")
+        assert obj["paths"]["trash_jobs"].endswith("/users/alice/trash/jobs")
+        assert obj["retention"]["jobs_trash_after_days"] == 3
+        assert obj["retention"]["jobs_purge_after_days"] == 7
+
+
+def test_create_user_creates_user_dirs(tmp_path: Path, monkeypatch):
+    from argus.service import app as app_mod
+
+    cfg_path = _write_config(tmp_path)
+    monkeypatch.setenv("MVP_INTERNAL_TOKEN", "adm1")
+
+    class _Scheduler:
+        def __init__(self, **kwargs):
+            self.tool = object()
+
+        def run_forever(self, stop_flag):
+            return None
+
+    monkeypatch.setattr(app_mod, "Scheduler", _Scheduler)
+    app = app_mod.create_app(str(cfg_path))
+
+    admin_headers = {"authorization": "Bearer adm1"}
+    with TestClient(app) as c:
+        assert c.post("/api/v2/users", headers=admin_headers, json={"user_id": "alice"}).status_code == 200
+
+    base = tmp_path / "users" / "alice"
+    assert (base / "datasets").is_dir()
+    assert (base / "models").is_dir()
+    assert (base / "code").is_dir()
+    assert (base / "jobs").is_dir()
+    assert (base / "trash" / "jobs").is_dir()
diff --git a/src/mvp/scripts/12_install_api_deps.sh b/src/mvp/scripts/12_install_api_deps.sh
index 565db17..90a2074 100755
--- a/src/mvp/scripts/12_install_api_deps.sh
+++ b/src/mvp/scripts/12_install_api_deps.sh
@@ -9,4 +9,4 @@ SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
 source "${SCRIPT_DIR}/lib.sh"
 
 dexec "${HEAD_CONTAINER}" bash -lc "python3 -m pip install -U pip >/dev/null 2>&1 || true"
-dexec "${HEAD_CONTAINER}" bash -lc "python3 -m pip install -r /workspace/mvp/py/requirements.txt"
+dexec "${HEAD_CONTAINER}" bash -lc "python3 -c 'import fastapi,uvicorn,yaml' >/dev/null 2>&1 && echo 'api_deps_ok: skip' || (python3 -m pip install -r /workspace/mvp/py/requirements.txt || echo 'WARN: api deps install failed; continuing with preinstalled deps')"
diff --git a/src/mvp/scripts/60_start_api.sh b/src/mvp/scripts/60_start_api.sh
index 29cef85..ed2cbd1 100755
--- a/src/mvp/scripts/60_start_api.sh
+++ b/src/mvp/scripts/60_start_api.sh
@@ -22,7 +22,12 @@ if [[ -z "${MVP_INTERNAL_TOKEN:-}" ]]; then
   exit 1
 fi
 
-docker exec -d -e MVP_INTERNAL_TOKEN="${MVP_INTERNAL_TOKEN}" "${HEAD_CONTAINER}" bash -lc "nohup python3 /workspace/mvp/py/server.py --config '${CONFIG_IN_CONTAINER}' >>'${LOG_PATH}' 2>&1 & echo \$! >'${PID_PATH}'"
+env_args=(-e "MVP_INTERNAL_TOKEN=${MVP_INTERNAL_TOKEN}")
+if [[ -n "${SFTPGO_ADMIN_PASSWORD:-}" ]]; then
+  env_args+=(-e "SFTPGO_ADMIN_PASSWORD=${SFTPGO_ADMIN_PASSWORD}")
+fi
+
+docker exec -d "${env_args[@]}" "${HEAD_CONTAINER}" bash -lc "nohup python3 /workspace/mvp/py/server.py --config '${CONFIG_IN_CONTAINER}' >>'${LOG_PATH}' 2>&1 & echo \$! >'${PID_PATH}'"
 
 echo "[host] started; pid stored in ${PID_PATH} (container path)"
 echo "[host] logs: ${LOG_PATH} (container path)"
diff --git a/src/mvp/scripts/run_all_v30_api.sh b/src/mvp/scripts/run_all_v30_api.sh
new file mode 100755
index 0000000..78b502b
--- /dev/null
+++ b/src/mvp/scripts/run_all_v30_api.sh
@@ -0,0 +1,242 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+# shellcheck source=lib.sh
+source "${SCRIPT_DIR}/lib.sh"
+
+# E2E v3.0:
+# - Start Ray (head + stateless workers) + SFTPGo (compose)
+# - Start API server with v3.0 config
+# - Create user (API triggers SFTPGo user create) and return one-time SFTP password
+# - (Optional) verify SFTP login
+# - Submit PPO/GRPO/SFT referencing user dataset paths and wait
+
+API_ADDR="${API_ADDR:-http://127.0.0.1:8080}"
+ADMIN_TOKEN="${MVP_INTERNAL_TOKEN:-}"
+USER_ID="${USER_ID:-alice}"
+RESET_DB="${RESET_DB:-1}"
+RESET_SFTPGO="${RESET_SFTPGO:-0}"
+EXPECTED_RAY_NODES="${EXPECTED_RAY_NODES:-3}" # head + 2 workers
+CLUSTER_NAME="${CLUSTER_NAME:-argus-ray}"
+
+CONFIG_IN_CONTAINER="${CONFIG_IN_CONTAINER:-/workspace/mvp/configs/dev_v30.yaml}"
+SFTPGO_ADMIN_PASSWORD="${SFTPGO_ADMIN_PASSWORD:-my-dev-sftpgo-admin}"
+export SFTPGO_ADMIN_PASSWORD
+
+if [[ -z "${ADMIN_TOKEN}" ]]; then
+  echo "ERROR: MVP_INTERNAL_TOKEN must be set in host env (admin token)" >&2
+  exit 1
+fi
+
+api_curl_admin() {
+  curl -sS -H "Authorization: Bearer ${ADMIN_TOKEN}" "$@"
+}
+
+api_wait_ready() {
+  local tries="${1:-60}"
+  for i in $(seq 1 "${tries}"); do
+    if curl -sS -m 2 "${API_ADDR}/docs" >/dev/null 2>&1; then
+      echo "[host] api_ready: ${API_ADDR}"
+      return 0
+    fi
+    echo "[host] waiting api... (${i}/${tries})"
+    sleep 2
+  done
+  echo "ERROR: api not ready: ${API_ADDR}" >&2
+  return 1
+}
+
+sftpgo_wait_ready() {
+  local tries="${1:-60}"
+  local url="${2:-http://127.0.0.1:8081/api/v2/token}"
+  for i in $(seq 1 "${tries}"); do
+    # SFTPGo admin endpoints require auth; readiness means "HTTP reachable and can issue token".
+    if curl -sS -m 2 -u "admin:${SFTPGO_ADMIN_PASSWORD}" "${url}" >/dev/null 2>&1; then
+      echo "[host] sftpgo_ready: ${url} (token ok)"
+      return 0
+    fi
+    echo "[host] waiting sftpgo... (${i}/${tries})"
+    sleep 2
+  done
+  echo "ERROR: sftpgo not ready: ${url}" >&2
+  return 1
+}
+
+ray_wait_ready() {
+  local tries="${1:-60}"
+  for i in $(seq 1 "${tries}"); do
+    if curl -sS -m 2 "${RAY_DASHBOARD_ADDR}/api/version" >/dev/null 2>&1; then
+      echo "[host] ray_dashboard_ready: ${RAY_DASHBOARD_ADDR}"
+      return 0
+    fi
+    echo "[host] waiting ray dashboard... (${i}/${tries})"
+    sleep 2
+  done
+  echo "ERROR: ray dashboard not ready: ${RAY_DASHBOARD_ADDR}" >&2
+  return 1
+}
+
+ray_wait_nodes() {
+  local want="${1:-3}"
+  local tries="${2:-60}"
+  for i in $(seq 1 "${tries}"); do
+    local out n
+    out="$(docker exec -i "${HEAD_CONTAINER}" python3 -c "import ray; ray.init(address='auto', ignore_reinit_error=True, log_to_driver=False, logging_level='ERROR'); print(sum(1 for n in ray.nodes() if n.get('Alive')))" 2>/dev/null || true)"
+    n="$(printf '%s\n' "${out}" | tail -n 1 | tr -cd '0-9' || true)"
+    if [[ "${n}" =~ ^[0-9]+$ ]]; then
+      echo "[host] ray_nodes_alive=${n} (want>=${want})"
+      if [[ "${n}" -ge "${want}" ]]; then
+        return 0
+      fi
+    else
+      echo "[host] waiting ray nodes... (${i}/${tries})"
+    fi
+    sleep 2
+  done
+  echo "ERROR: ray nodes not ready (want>=${want})" >&2
+  docker exec -i "${HEAD_CONTAINER}" bash -lc "ray status || true" >&2 || true
+  return 1
+}
+
+submit_taskspec_inline() {
+  local token="$1"
+  local yaml_body="$2"
+  local resp
+  resp="$(curl -sS -H "Authorization: Bearer ${token}" -H "Content-Type: application/yaml" --data-binary "${yaml_body}" "${API_ADDR}/api/v2/tasks")"
+  echo "[host] submit_resp: ${resp}" >&2
+  printf '%s' "${resp}" | python3 -c 'import sys,json; print(json.load(sys.stdin)["task_id"])'
+}
+
+wait_task() {
+  local token="$1"
+  local task_id="$2"
+  while true; do
+    local body state
+    body="$(curl -sS -H "Authorization: Bearer ${token}" "${API_ADDR}/api/v2/tasks/${task_id}")"
+    state="$(printf '%s' "${body}" | python3 -c 'import sys,json; print(json.load(sys.stdin)["state"])')"
+    echo "[host] task ${task_id}: ${state}"
+
+    if [[ "${state}" == "SUCCEEDED" ]]; then
+      return 0
+    fi
+    if [[ "${state}" == "FAILED" || "${state}" == "CANCELED" ]]; then
+      echo "[host] terminal=${state}; tail logs (best-effort):" >&2
+      curl -sS -H "Authorization: Bearer ${token}" "${API_ADDR}/api/v2/tasks/${task_id}/logs?tail=200" >&2 || true
+      return 1
+    fi
+    sleep 10
+  done
+}
+
+echo "[host] ===== run_all_v30_api.sh begin ====="
+
+"${SCRIPT_DIR}/00_prereq_check.sh"
+"${SCRIPT_DIR}/03_cleanup_v1_legacy.sh"
+"${SCRIPT_DIR}/04_cleanup_v2_legacy.sh"
+
+echo "[host] bring down existing containers (best-effort)"
+"${SCRIPT_DIR}/02_down.sh" || true
+
+if [[ "${RESET_SFTPGO}" == "1" ]]; then
+  echo "[host] reset sftpgo metadata dir (best-effort, via helper container)"
+  SFTPGO_META_DIR="${ROOT_DIR}/../../shared/common/sftpgo"
+  mkdir -p "${SFTPGO_META_DIR}"
+  docker run --rm --entrypoint sh -u 0:0 -v "${SFTPGO_META_DIR}:/mnt" drakkan/sftpgo:latest -lc "rm -rf /mnt/* || true"
+fi
+
+echo "[host] (re)create containers (Ray + SFTPGo)"
+"${SCRIPT_DIR}/01_up.sh"
+
+echo "[host] wait ray head ready"
+ray_wait_ready 60
+
+echo "[host] wait sftpgo ready"
+sftpgo_wait_ready 60 "http://127.0.0.1:8081/api/v2/token"
+
+echo "[host] render v3.0 config with SFTPGo container IP (work around docker DNS issues)"
+SFTPGO_IP="$(docker inspect -f '{{range .NetworkSettings.Networks}}{{.IPAddress}}{{end}}' argus-sftpgo)"
+RENDERED_CFG_HOST_PATH="/tmp/dev_v30_rendered.yaml"
+sed -E "s#^(\\s*admin_api_base:) .*#\\1 \"http://${SFTPGO_IP}:8080/api/v2\"#g" "${ROOT_DIR}/configs/dev_v30.yaml" >"${RENDERED_CFG_HOST_PATH}"
+docker cp "${RENDERED_CFG_HOST_PATH}" "${HEAD_CONTAINER}:/tmp/dev_v30_rendered.yaml"
+CONFIG_IN_CONTAINER="/tmp/dev_v30_rendered.yaml"
+
+echo "[host] verify head discovery record (supervised in-container)"
+HEAD_IP_FILE="${SHARED_ROOT}/ray/discovery/${CLUSTER_NAME}/head.json"
+dexec "${HEAD_CONTAINER}" bash -lc "test -f '${HEAD_IP_FILE}' && python3 -c 'import json,sys; print(json.load(open(sys.argv[1]))[\"job_server_url\"])' '${HEAD_IP_FILE}' || true"
+
+echo "[host] wait workers join"
+ray_wait_nodes "${EXPECTED_RAY_NODES}" 120
+
+echo "[host] prepare data/model (idempotent; reuse cache)"
+"${SCRIPT_DIR}/30_prepare_data_and_model.sh"
+
+echo "[host] install api deps in head container"
+"${SCRIPT_DIR}/12_install_api_deps.sh"
+
+echo "[host] stop api (best-effort)"
+"${SCRIPT_DIR}/61_stop_api.sh" || true
+
+if [[ "${RESET_DB}" == "1" ]]; then
+  echo "[host] reset api sqlite db in container (best-effort)"
+  docker exec -i "${HEAD_CONTAINER}" bash -lc "rm -f /private/common/db/mvp.sqlite3 /private/common/db/mvp.sqlite3-wal /private/common/db/mvp.sqlite3-shm || true"
+else
+  echo "[host] keep existing api sqlite db (RESET_DB=${RESET_DB})"
+fi
+
+echo "[host] start api (admin token + sftpgo admin password via env)"
+MVP_INTERNAL_TOKEN="${ADMIN_TOKEN}" CONFIG_IN_CONTAINER="${CONFIG_IN_CONTAINER}" SFTPGO_ADMIN_PASSWORD="${SFTPGO_ADMIN_PASSWORD}" "${SCRIPT_DIR}/60_start_api.sh"
+api_wait_ready 60
+
+echo "[host] create user (expect SFTP one-time password in response)"
+create_resp="$(api_curl_admin -H "Content-Type: application/json" -d "{\"user_id\":\"${USER_ID}\"}" "${API_ADDR}/api/v2/users")"
+echo "[host] create_user_resp: ${create_resp}"
+USER_SFTP_PASSWORD="$(printf '%s' "${create_resp}" | python3 -c 'import sys,json; o=json.load(sys.stdin); print((o.get("sftp") or {}).get("password") or "")')"
+if [[ -z "${USER_SFTP_PASSWORD}" ]]; then
+  echo "ERROR: expected sftp.password in create user response (is data.sftpgo.enabled=true?)" >&2
+  exit 1
+fi
+
+echo "[host] issue user token"
+USER_TOKEN="$(api_curl_admin -X POST "${API_ADDR}/api/v2/users/${USER_ID}/tokens" | python3 -c 'import sys,json; print(json.load(sys.stdin)["token"])')"
+echo "[host] user_token_issued: user=${USER_ID}"
+
+echo "[host] (optional) verify SFTP login (best-effort)"
+if command -v sshpass >/dev/null 2>&1 && command -v sftp >/dev/null 2>&1; then
+  tmp_batch="$(mktemp)"
+  cat >"${tmp_batch}" <<EOF
+pwd
+ls
+EOF
+  # NOTE: this just proves auth works; dataset upload can be done via SFTP manually later.
+  sshpass -p "${USER_SFTP_PASSWORD}" sftp -o StrictHostKeyChecking=no -P 2022 "${USER_ID}@127.0.0.1" -b "${tmp_batch}" >/dev/null 2>&1 || true
+  rm -f "${tmp_batch}" || true
+else
+  echo "[host] skip: sshpass/sftp not found; you can test manually with: sftp -P 2022 ${USER_ID}@<host>"
+fi
+
+echo "[host] ensure user dataset paths exist (copy from common if needed; best-effort)"
+echo "[host] copy dataset into user datasets path inside head container (avoid host permission issues)"
+dexec "${HEAD_CONTAINER}" bash -lc "set -euo pipefail; \
+  mkdir -p '/private/users/${USER_ID}/datasets/gsm8k' '/private/users/${USER_ID}/datasets/gsm8k_sft'; \
+  (cp -f /private/common/datasets/gsm8k/train.parquet '/private/users/${USER_ID}/datasets/gsm8k/train.parquet' 2>/dev/null || cp -f /private/datasets/gsm8k/train.parquet '/private/users/${USER_ID}/datasets/gsm8k/train.parquet' 2>/dev/null || true); \
+  (cp -f /private/common/datasets/gsm8k/test.parquet  '/private/users/${USER_ID}/datasets/gsm8k/test.parquet'  2>/dev/null || cp -f /private/datasets/gsm8k/test.parquet  '/private/users/${USER_ID}/datasets/gsm8k/test.parquet'  2>/dev/null || true); \
+  (cp -f /private/common/datasets/gsm8k_sft/train.parquet '/private/users/${USER_ID}/datasets/gsm8k_sft/train.parquet' 2>/dev/null || cp -f /private/datasets/gsm8k_sft/train.parquet '/private/users/${USER_ID}/datasets/gsm8k_sft/train.parquet' 2>/dev/null || true); \
+  (cp -f /private/common/datasets/gsm8k_sft/test.parquet  '/private/users/${USER_ID}/datasets/gsm8k_sft/test.parquet'  2>/dev/null || cp -f /private/datasets/gsm8k_sft/test.parquet  '/private/users/${USER_ID}/datasets/gsm8k_sft/test.parquet'  2>/dev/null || true)"
+
+echo "[host] submit PPO/GRPO/SFT via API using user dataset paths"
+PPO_TASK_ID="$(submit_taskspec_inline "${USER_TOKEN}" $'workload: ppo\nnnodes: 2\nn_gpus_per_node: 4\ncode_path: /private/common/code/verl/verl_repo\ntrain_file: /private/users/'"${USER_ID}"$'/datasets/gsm8k/train.parquet\nval_file: /private/users/'"${USER_ID}"$'/datasets/gsm8k/test.parquet\nmodel_id: Qwen/Qwen2.5-0.5B-Instruct\ntotal_epochs: 1\ntotal_training_steps: 10\nsave_freq: 10\n')"
+GRPO_TASK_ID="$(submit_taskspec_inline "${USER_TOKEN}" $'workload: grpo\nnnodes: 2\nn_gpus_per_node: 4\ncode_path: /private/common/code/verl/verl_repo\ntrain_file: /private/users/'"${USER_ID}"$'/datasets/gsm8k/train.parquet\nval_file: /private/users/'"${USER_ID}"$'/datasets/gsm8k/test.parquet\nmodel_id: Qwen/Qwen2.5-0.5B-Instruct\ntotal_epochs: 1\ntotal_training_steps: 10\nsave_freq: 10\n')"
+SFT_TASK_ID="$(submit_taskspec_inline "${USER_TOKEN}" $'workload: sft\nnnodes: 1\nn_gpus_per_node: 1\ncode_path: /private/common/code/verl/verl_repo\ntrain_file: /private/users/'"${USER_ID}"$'/datasets/gsm8k_sft/train.parquet\nval_file: /private/users/'"${USER_ID}"$'/datasets/gsm8k_sft/test.parquet\nmodel_id: Qwen/Qwen2.5-0.5B-Instruct\ntotal_epochs: 1\ntotal_training_steps: 10\nsave_freq: 10\n')"
+
+echo "[host] submitted task ids:"
+echo "  ppo=${PPO_TASK_ID}"
+echo "  grpo=${GRPO_TASK_ID}"
+echo "  sft=${SFT_TASK_ID}"
+
+echo "[host] wait for tasks (in submission order)"
+wait_task "${USER_TOKEN}" "${PPO_TASK_ID}"
+wait_task "${USER_TOKEN}" "${GRPO_TASK_ID}"
+wait_task "${USER_TOKEN}" "${SFT_TASK_ID}"
+
+echo "[host] ===== run_all_v30_api.sh done ====="
diff --git a/src/mvp/scripts/run_e2e_v30_cases.sh b/src/mvp/scripts/run_e2e_v30_cases.sh
new file mode 100755
index 0000000..dc56b19
--- /dev/null
+++ b/src/mvp/scripts/run_e2e_v30_cases.sh
@@ -0,0 +1,19 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+
+# Minimal E2E runner for v3.0:
+# - Happy path: Ray + SFTPGo + API + 3 workloads
+# - Leaves retention validation as a manual follow-up (adjust thresholds).
+
+ADMIN_TOKEN="${MVP_INTERNAL_TOKEN:-my-dev-token}"
+USER_ID="${USER_ID:-alice}"
+
+echo "[case] HP-1: run_all_v30_api.sh (Ray + SFTPGo + API + 3 workloads)"
+MVP_INTERNAL_TOKEN="${ADMIN_TOKEN}" USER_ID="${USER_ID}" RESET_DB=1 RESET_SFTPGO=0 "${SCRIPT_DIR}/run_all_v30_api.sh"
+
+echo "[case] NOTE: retention validation (C2) is manual:"
+echo "  - set data.retention.jobs_trash_after_days / jobs_purge_after_days to small values in configs/dev_v30.yaml"
+echo "  - restart API server and wait for janitor"
+