diff --git a/CLAUDE.md b/CLAUDE.md
index f75bbae..8e2ad32 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -222,7 +222,51 @@ servers:
 - 控速依据是 **GPU 温度**（DCGM），目标机的机箱风扇，写 `ipmitool raw 0x3a 0x01`
 - 运行时状态（模式 / 管控 GPU / 分配关系 / 审计）**全部落 SQLite**，配置文件只是首次运行的种子
 
-**实时数据源：三个外部 exporter（自备组件，不随仓库提供，装法不限）**
+### 控制台模块地图（`app/`）
+
+| 模块 | 职责 |
+|---|---|
+| `main.py` | 入口：单进程 = 控制回路 + API + 静态托管；`STATIC_DIR` 来自 `runtime.py` |
+| `api.py` | REST + WebSocket（`/api/status`、`/api/curve/preview`、`/ws` 等） |
+| `controller.py` | 控制回路（默认 15s 一轮，可配） |
+| `curve.py` | 分段曲线 + 滞回（升温立即生效，降温须跌出滞回带） |
+| `ipmi.py` | `raw 0x3a 0x01` 命令族与占空比编码 |
+| `sensors.py` | DCGM / ipmi_exporter / node_exporter / nvidia-smi 兜底 / Prometheus 历史 |
+| `safety.py` | 安全护栏（退出回退 BMC 自动档 + atexit + 心跳） |
+| `store.py` | SQLite（`data/fan-console.db`，唯一权威数据源） |
+| `config.py` | pydantic 配置模型；`load_config` 缺文件时优雅降级为默认值 |
+| `runtime.py` | **资源定位中枢**：源码运行根 = `app/` 包目录；PyInstaller 冻结运行根 = exe 所在目录。config / DB / static / heartbeat 四处路径全部从这里取——**改路径相关代码时只动这里，别在各自模块里再写 `__file__` 推导** |
+
+### 打包与 CI/CD
+
+**CI**（`.github/workflows/ci.yml`，推 main / PR / 打 `v*` tag 触发）：
+
+1. `test-and-build`：后端单测（Python 3.13）→ 前端 `npm ci && npm run build`（含 vue-tsc 全量类型检查）→ 打部署包 `gpu-fan-console-app.tgz`（成员路径 `app/...`，排除 `app/data`）挂 Artifacts
+2. `build-console` / `build-legacy`（非 PR 才跑）：PyInstaller 双平台可执行；分发压缩包由 `packaging/build_console.py` 内置生成（zipfile/tarfile，**不要**在 workflow 里用 tar/PowerShell 打——Windows runner 的 cp1252 控制台编不了中文输出，打包脚本必须 `sys.stdout.reconfigure(encoding="utf-8")`）
+3. `release`（打 tag 才跑）：`gh release create` 自动发版附全部产物；**必须显式 `--repo "$GITHUB_REPOSITORY"`**（该 job 无 checkout，gh 推断不出仓库）
+
+**冻结（PyInstaller）相关**：
+
+- 打包入口 `entry_console.py`（控制台 onedir）、`packaging/build_console.py` / `packaging/build_legacy.py`（旧脚本 onefile）
+- 冻结后**可变资源必须落 exe 旁**（`_MEIPASS` 每次启动清空）：config.yaml / `data/` / heartbeat 由 `runtime.py` 定位到 exe 旁；旧脚本 `fancontroller(.once).py` 的 `_app_directory()` 同理
+- 本机已实测：控制台 exe（API/静态页/SQLite 落位）、旧脚本 exe（缺配置报错、空配置正常退出）
+
+### 部署
+
+- 部署方式、踩坑（tar 路径对齐 / upload 分片 base64 / 远端 rm 黑名单）、回滚、验证清单 → **[`DEPLOY.md`](./DEPLOY.md)**
+- 一键脚本 `./deploy.sh`：**`CONN=<ssh连接名>` 必填**，可选 `REMOTE_DIR` / `SERVICE` / `URL`；
+  脚本打包时排除 `app/data` 与 `app/config.yaml`
+- 服务器直连 GitHub 可用时，更快的更新方式：直接 `curl -L` Release 页的
+  `gpu-fan-console-app.tgz`，免去本地构建与分片上传（v2.0 实测可行）
+- ⚠️ 升级已配置的机器时，解压必须 `--exclude='app/config.yaml'`——仓库里的
+  config.yaml 是**模板**（示例地址），覆盖会毁掉现场配置
+
+### 仓库约定
+
+- **公开仓库，严禁出现真实内网 IP / 主机名 / 凭据**。示例地址一律用 RFC 5737
+  文档段（192.0.2.0/24 等）；写文档、注释、测试样例时同样遵守
+- `config.yaml` 按模板维护；`README_EN.md` 与中文版结构对齐
+- 实时数据源：三个外部 exporter（自备组件，装法不限）：
 
 | 数据 | exporter | 项目地址 | 指标 | 默认端口 |
 |---|---|---|---|---|
@@ -234,5 +278,3 @@ servers:
 （仅 `/api/history` 用）可配；DCGM 不可用时 `sensors.py` 降级 `nvidia-smi` 兜底。
 ipmi_exporter in-band 读 `/dev/ipmi0` 需要 root。安装/自检详见 [`DEPLOY.md`](./DEPLOY.md) 2.1 节。
 
-- 部署方式、踩坑、回滚、验证清单 → **见 [`DEPLOY.md`](./DEPLOY.md)**，一键脚本 `./deploy.sh`
-
diff --git a/DEPLOY.md b/DEPLOY.md
index 3d25161..2feeaac 100644
--- a/DEPLOY.md
+++ b/DEPLOY.md
@@ -11,8 +11,8 @@
 
 ```bash
 # 在项目根目录（Windows Git Bash）
-./deploy.sh              # 全量：后端 + 前端，自动重启服务并自检
-./deploy.sh --static-only  # 只更新前端（不改后端、不重启服务）
+CONN=<连接名> ./deploy.sh   # 全量：后端 + 前端，自动重启服务并自检
+CONN=<连接名> ./deploy.sh --static-only  # 只更新前端（不改后端、不重启服务）
 ```
 
 下面是人话版说明，出问题时看这里。
@@ -139,7 +139,7 @@ curl -s http://192.0.2.10:8765/api/status | head -c 300
 后端没动时不用重启进程 —— `StaticFiles` 每次请求读盘，`index.html` 也不会被缓存：
 
 ```bash
-./deploy.sh --static-only
+CONN=<连接名> ./deploy.sh --static-only
 ```
 
 > ⚠️ **别手动只拷 `index-xxxx.js`**。文件名带**内容 hash**，源码改一行文件名就变。
@@ -226,7 +226,7 @@ find <dir> -type f ! -name '新文件1' ! -name '新文件2' -delete
 # 代码回滚
 git log --oneline -5
 git checkout <上一个好提交>
-./deploy.sh
+CONN=<连接名> ./deploy.sh
 
 # 数据回滚（SQLite 有备份时）
 agentsshcli exec <连接名> "systemctl stop gpu-fan-console"
diff --git a/README.md b/README.md
index 1135a10..29c76ac 100644
--- a/README.md
+++ b/README.md
@@ -1,4 +1,4 @@
-[English Readme](./README_EN.md)（英文版目前只覆盖旧版脚本，尚未同步新控制台）
+[English Readme](./README_EN.md)
 
 # python-ipmitool
 
@@ -119,8 +119,8 @@ python -m unittest discover -s app/tests -t . -v
 **日常更新一条命令**：
 
 ```bash
-./deploy.sh                 # 全量：后端 + 前端，自动重启服务并自检
-./deploy.sh --static-only   # 只更新前端（不改后端、不重启服务）
+CONN=<连接名> ./deploy.sh   # 全量：后端 + 前端，自动重启服务并自检
+CONN=<连接名> ./deploy.sh --static-only  # 只更新前端（不改后端、不重启服务）
 ```
 
 首次部署、踩坑记录、回滚与验证清单见 **[DEPLOY.md](./DEPLOY.md)**。
@@ -246,7 +246,7 @@ Linux 长期运行建议配置为 systemd 服务（`/etc/systemd/system/fancontr
 
 - [DEPLOY.md](./DEPLOY.md) —— GPU 风扇控制台部署手册（一键脚本、踩坑、回滚、验证清单）
 - [CLAUDE.md](./CLAUDE.md) —— 仓库架构说明（新旧两套的实现细节、加新机型的方法）
-- [README_EN.md](./README_EN.md) —— 英文说明（仅旧版脚本）
+- [README_EN.md](./README_EN.md) —— 英文说明
 
 ## 贡献与反馈
 
diff --git a/README_EN.md b/README_EN.md
index ce884eb..cd21433 100644
--- a/README_EN.md
+++ b/README_EN.md
@@ -1,213 +1,232 @@
 [中文文档](./README.md)
 
-# Python IPMI Fan Controller
-
-A Python script to monitor server CPU temperature via IPMI and adjust fan speeds according to predefined temperature ranges.
-
-## Cross-Platform Support (Windows & Linux)!
-
-## Compatible Servers
-
-The following models have been tested and are confirmed to work. More models are pending testing; contributions are welcome!
+# python-ipmitool
+
+Server fan control via IPMI. The repository contains two parts:
+
+| Component | Location | Status | Description |
+|---|---|---|---|
+| **GPU Fan Console** | `app/` + `frontend/` | Current main project | Web console that regulates chassis fans in a closed loop by GPU temperature (FastAPI + Vue3) |
+| **Fan Control Scripts** | `fancontroller.py` etc. | Legacy, still usable | Command-line scripts, CPU-temperature-based fan control (Dell 730 etc.) |
+
+## GPU Fan Console
+
+### What it does
+
+Target platform: **ASRock Rack EPYCD8** (BMC firmware 2.20). Passive cards like the
+Tesla T10 rely entirely on chassis fans for cooling, so the console reads GPU
+temperature, computes a duty cycle from an editable step curve with hysteresis,
+and writes it to the BMC via `ipmitool raw 0x3a 0x01`.
+
+- **Data sources**: GPU temperature from a local DCGM exporter (`:9400`), fan speed
+  from ipmi_exporter (`:9290`), CPU temperature from node_exporter (`:9100`, requires
+  `--collector.hwmon`) — all read directly from the exporters; historical trends go
+  through Prometheus. Falls back to `nvidia-smi` when DCGM is unavailable
+- **Control curve**: piecewise curve + hysteresis band (rising temperature applies
+  immediately; falling temperature must leave the hysteresis band before downshift,
+  avoiding the "helicopter effect"). Editable in the UI with a **live preview**
+  (`/api/curve/preview` runs the real algorithm)
+- **Three modes**: auto (curve) / manual (fixed duty from the UI) / BMC auto
+- **GPU ↔ fan-slot assignments**: each fan slot is bound to a GPU; the highest
+  temperature of the bound GPU drives the speed. Unassigned slots go back to the BMC
+- **State lives in SQLite**: mode, curve, assignments and audit log all persist in
+  `app/data/fan-console.db`, the single source of truth; `app/config.yaml` is only
+  a first-run seed
+- **Safety net (three layers)**:
+  1. On shutdown / crash / SIGTERM, `SafetyGuard` hands all managed fan slots back to BMC auto
+  2. In-process `atexit` second layer
+  3. **Independent heartbeat watchdog** (systemd timer, every 2 min): if the heartbeat
+     expires it force-writes `8×0x00` — covering even SIGKILL / power loss
+- **Dashboard**: GPU cards (temperature / SM clock), fan speeds, separate temperature
+  and fan-speed trend charts, audit log
+
+### Repository layout
+
+```
+app/
+  main.py          # Entry: one process = control loop + API + static hosting
+  api.py           # REST + WebSocket routes
+  controller.py    # Control loop (15 s per tick)
+  curve.py         # Piecewise curve + hysteresis
+  ipmi.py          # raw 0x3a 0x01 command family
+  sensors.py       # DCGM / ipmi_exporter / nvidia-smi / Prometheus readers
+  safety.py        # Safety guard
+  store.py         # SQLite persistence
+  runtime.py       # Resource paths (source run = app/ dir; frozen = exe dir)
+  config.yaml      # Configuration (template with example addresses; seed for first run)
+  deploy/          # systemd units (main service + watchdog) and watchdog script
+  data/            # fan-console.db (generated at runtime, never overwrite)
+  static/          # Frontend build output (npm run build lands here)
+  tests/           # Unit tests
+frontend/          # Vue3 + TS + Vite + Tailwind (Dashboard / Fans / Settings)
+deploy.sh          # One-click deploy script
+DEPLOY.md          # Deployment manual (pitfalls, rollback, verification checklist)
+```
+
+### Run locally
+
+```bash
+# 1) Build the frontend (output goes straight into app/static/)
+cd frontend && npm install && npm run build && cd ..
+
+# 2) Install backend dependencies
+pip install -r app/requirements.txt
+
+# 3) Start (one process serves pages + API + control loop)
+python -m uvicorn app.main:app --host 0.0.0.0 --port 8765
+# Open http://127.0.0.1:8765
+```
+
+Frontend development with hot reload:
+
+```bash
+cd frontend && npm run dev    # vite on :5173, /api proxied to 127.0.0.1:8765
+```
+
+Run tests:
+
+```bash
+python -m unittest discover -s app/tests -t . -v
+```
+
+### API overview
+
+| Method | Path | Description |
+|---|---|---|
+| GET | `/api/status` | Current state (mode, managed slots, temperatures, duty) |
+| GET | `/api/gpus` / `/api/fans` | GPU / fan lists |
+| GET/PUT | `/api/curve` | Read/write the control curve |
+| POST | `/api/curve/preview` | Curve preview (real algorithm) |
+| GET | `/api/history` | Historical trends (from Prometheus) |
+| GET/PUT | `/api/assignments` | GPU ↔ fan-slot assignments |
+| POST | `/api/mode` | Switch auto / manual / bmc-auto |
+| POST | `/api/manual` | Manually set duty cycle |
+| POST | `/api/restore-auto` | Hand everything back to BMC auto |
+| GET | `/api/audit` | Audit log |
+| GET | `/api/health` | Health check |
+| WS | `/ws` | Live push |
+
+### Deployment
+
+Production target is Linux + systemd (in-band `/dev/ipmi0` access requires root).
+**Daily updates are one command**:
+
+```bash
+CONN=<ssh-connection> ./deploy.sh                 # full: backend + frontend, restart + self-check
+CONN=<ssh-connection> ./deploy.sh --static-only   # frontend only (no backend change, no restart)
+```
+
+First-time deployment, pitfalls, rollback and the verification checklist live in
+**[DEPLOY.md](./DEPLOY.md)**. Last-resort fallback: send
+`ipmitool raw 0x3a 0x01 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00` to the BMC to hand
+all fan slots back to BMC auto (can be done from the BMC's own management address,
+no host OS required).
+
+### CI & releases (GitHub Actions)
+
+Every push / PR runs backend tests, frontend type-checked build and produces a
+deployable bundle artifact. Pushes to main additionally build **standalone
+executables for Windows and Linux** (PyInstaller); pushing a `v*` tag automatically
+publishes a GitHub Release with all artifacts. See
+[.github/workflows/ci.yml](./.github/workflows/ci.yml).
+
+---
+
+## Fan Control Scripts (Legacy CLI)
+
+Interface-free command-line version: monitors CPU temperature and adjusts fan speeds
+per predefined ranges. Works on Windows and Linux. Fan speed and temperature readings
+are unified through Prometheus while write commands remain per-model, with
+multi-server, multi-threading, daily log rotation and optional e-mail alerting
+(off by default).
+
+### Compatible servers
 
 | Brand | Model | Compatible | Type Name |
-|:-----:|:-----:|:----------:|:---------:|
-| Dell  | 730XD |     Y      | `dell730` |
-| Dell  | 730   |     Y      | `dell730` |
-
-## How to Use
-
-> On Linux, `ipmitool` must be installed.
-> 
-> For Debian-based systems, install it with `apt install -y ipmitool`.
-> For Red Hat-based systems, use `yum install -y ipmitool`.
-
-1.  **Clone the project**
-
-    ```
-    git clone https://github.com/dongyu6/python-ipmitool.git
-    ```
-
-2.  **Navigate to the project directory**
-
-    ```
-    cd python-ipmitool
-    ```
-
-3.  **Install dependencies**
-
-    ```
-    pip install -r requirements.txt
-    ```
-
-4.  **Copy the template file** `fan_settings.yaml.template` to `fan_settings.yaml`.
-
-    ```bash
-    # Linux/Mac
-    cp fan_settings.yaml.template fan_settings.yaml
-
-    # Windows
-    copy fan_settings.yaml.template fan_settings.yaml
-    ```
-
-5.  **Edit the newly created `fan_settings.yaml`** file. The meaning of each field is as follows. You need to configure the IP addresses and fan speeds yourself.
-
-    > Note: Only IP addresses are supported, not domain names.
-
-    ```yaml
-    # IPMI Fan Controller Configuration
-
-    # true for automatic fan control, false for manual
-    auto: true
-
-    # The interval in seconds for checking temperature and adjusting fan speed
-    interval: 60
-
-    # Number of days to retain log files
-    log_backup_count: 30
-
-    # Path to the ipmitool executable on Windows
-    windows_ipmi_tool_path: ".\\ipmitool\\ipmitool.exe"
-
-    # List of servers to manage
-    servers:
-      - type: dell730                    # Server type
-        ip: "192.168.71.90"              # Server IP address. Set to "local" if running on the target machine
-        user: root                       # IPMI username
-        password: "123123"               # IPMI password
-        temperature_ranges:              # List of temperature ranges and corresponding fan speeds
-          - min_temp: 0                  # Minimum temperature of the range (inclusive)
-            max_temp: 60                 # Maximum temperature of the range (inclusive)
-            fan_speeds: [20, 20, 20, 20, 20, 20]  # List of fan speeds in percent
-          - min_temp: 61
-            max_temp: 80
-            fan_speeds: [25, 25, 25, 25, 25, 25]
-    ```
-
-
-6.  **Run the Project**
-
-    The project offers two execution modes:
-
-    ## Mode 1: Loop Control Mode (Recommended for Long-term Operation)
-
-    The program continuously monitors temperature and adjusts fan speeds. Suitable for running as a background service.
-
-    **Foreground Execution (for debugging)**
-    ```bash
-    # Windows
-    python fancontroller.py
-
-    # Linux
-    python3 fancontroller.py
-    ```
-
-    **Background Execution**
-    ```bash
-    # Windows
-    start /b python fancontroller.py
-
-    # Linux
-    nohup python3 fancontroller.py &
-    ```
-
-    ## Mode 2: One-Shot Execution Mode (Recommended for External Scheduling)
-
-    Executes once and exits after temperature detection and fan adjustment. Suitable for being called by external scheduling tools like cron, systemd timer, etc.
-
-    **Direct Execution**
-    ```bash
-    # Windows
-    python fancontroller_once.py
-
-    # Linux
-    python3 fancontroller_once.py
-    ```
-
-    **Using cron for Scheduled Execution (Linux)**
-    ```bash
-    # Edit crontab
-    crontab -e
-
-    # Execute every 10 minutes
-    */10 * * * * /usr/bin/python3 /path/to/python-ipmitool/fancontroller_once.py
-    ```
-
-    **Using Windows Task Scheduler**
-    ```powershell
-    # Create a task that runs every 10 minutes
-    schtasks /create /tn "IPMI Fan Controller" /tr "python C:\path\to\python-ipmitool\fancontroller_once.py" /sc minute /mo 10
-    ```
-
-### Setup as a systemd Service (Linux Recommended)
-
-For reliable operation on a Linux server, setting up a systemd service is highly recommended. This provides features like auto-start on boot and process supervision.
-
-1.  **Create the Service File**
-
-    Use a text editor (like `nano` or `vim`) to create a new service file:
-    ```
-    sudo nano /etc/systemd/system/fancontroller.service
-    ```
-
-2.  **Paste the Service Configuration**
-
-    Paste the following content into the file. **Note:** You must replace the paths for `User`, `WorkingDirectory`, and `ExecStart` with the actual paths on your server.
-
-    ```ini
-    [Unit]
-    Description=Python IPMI Fan Controller
-    After=network.target
-
-    [Service]
-    Type=simple
-    # If you use a non-root user, ensure they have permissions for ipmitool.
-    User=root
-    # Absolute path to the project directory.
-    WorkingDirectory=/path/to/python-ipmitool
-    # Absolute path to the Python interpreter and the script.
-    ExecStart=/usr/bin/python3 /path/to/python-ipmitool/fancontroller.py
-    Restart=always
-    RestartSec=3
-
-    [Install]
-    WantedBy=multi-user.target
-    ```
-
-3.  **Reload and Enable the Service**
-
-    Run the following commands to reload the systemd configuration, start the service, and enable it to start on boot.
-
-    ```bash
-    # Reload the systemd configuration
-    sudo systemctl daemon-reload
-
-    # Start the service
-    sudo systemctl start fancontroller.service
-
-    # Check the service status to ensure there are no errors
-    sudo systemctl status fancontroller.service
-
-    # Enable the service to start automatically on boot
-    sudo systemctl enable fancontroller.service
-    ```
-
-4.  **View Logs**
-
-    Once configured as a service, all output (including errors) can be viewed with `journalctl`:
-    ```bash
-    journalctl -u fancontroller.service -f
-    ```
-
-## Contributing and Feedback
-
-Contributions via Issues and Pull Requests are welcome to help improve the project. If you have any questions or suggestions, please provide feedback through GitHub Issues.
+|:---:|:---:|:---:|:---:|
+| Dell | 730XD | Y | `dell730` |
+| Dell | 730 | Y | `dell730` |
+| ASRock Rack | EPYCD8 | Y | `epycd8` |
+
+### How to use
+
+> On Linux install `ipmitool` first: Debian-based `apt install -y ipmitool`,
+> Red Hat-based `yum install -y ipmitool`. On Windows use the bundled
+> `ipmitool/ipmitool.exe`.
+
+```bash
+git clone https://github.com/chennest/python-ipmitool.git
+cd python-ipmitool
+pip install -r requirements.txt
+
+# Copy and edit the config (IP addresses only, no domains; use "local" for in-band)
+cp fan_settings.yaml.template fan_settings.yaml
+```
+
+Key fields of `fan_settings.yaml` (full example in the template file):
+
+```yaml
+auto: true                  # true = auto, false = manual
+interval: 60                # control interval (seconds)
+log_backup_count: 30        # log retention (days)
+windows_ipmi_tool_path: ".\\ipmitool\\ipmitool.exe"
+alert:                      # e-mail alerts (optional, off by default)
+  enabled: false
+  fan_speed_threshold: 10000
+  max_failed_attempts: 3
+  email: { ... }            # SMTP config, multiple recipients, 1 h anti-spam
+prometheus:
+  base_url: "http://<your-prometheus>:9090"
+servers:
+  - type: dell730
+    ip: "192.0.2.10"        # example address, replace with yours
+    user: root
+    password: "your-password"
+    temperature_ranges:     # temperature range → per-fan percentages
+      - { min_temp: 0,  max_temp: 60, fan_speeds: [20, 20, 20, 20, 20, 20] }
+      - { min_temp: 61, max_temp: 80, fan_speeds: [25, 25, 25, 25, 25, 25] }
+```
+
+Two run modes:
+
+```bash
+# Mode 1: loop control (recommended for long-running background service)
+python fancontroller.py
+
+# Mode 2: run once (recommended for cron / systemd timer / Task Scheduler)
+python fancontroller_once.py
+# crontab example: every 10 minutes
+# */10 * * * * /usr/bin/python3 /path/to/python-ipmitool/fancontroller_once.py
+```
+
+For long-running Linux setups configure a systemd service
+(`/etc/systemd/system/fancontroller.service`, `After=network.target` +
+`Restart=always`; logs via `journalctl -u fancontroller -f`).
+
+> Alert triggers: fan speed above the threshold (default 10000 RPM) or consecutive
+> failures reaching the limit (default 3); per-server alert interval is at least
+> 1 hour. Gmail requires an app password. Controller implementation details are in
+> [CLAUDE.md](./CLAUDE.md).
+
+---
+
+## Documentation index
+
+- [DEPLOY.md](./DEPLOY.md) — GPU Fan Console deployment manual (one-click script, pitfalls, rollback, verification checklist)
+- [CLAUDE.md](./CLAUDE.md) — repository architecture (both generations, how to add a new model)
+- [README.md](./README.md) — Chinese documentation
+
+## Contributing
+
+Issues and pull requests are welcome. For questions or suggestions, please open a
+GitHub issue.
 
 ## License
 
-This project is licensed under the MIT License. See the LICENSE file for details.
+This project is licensed under **GPL-3.0** — see the [LICENSE](./LICENSE) file.
+
+## Credits
 
-## Acknowledgements
+[perryclements/r410-fancontroller: Python fan controller for Dell R410 server (GitHub.com)](https://github.com/perryclements/r410-fancontroller)
 
-- [perryclements/r410-fancontroller: Python fan controller for Dell R410 server (GitHub.com)](https://github.com/perryclements/r410-fancontroller)
-- [ipmitool/ipmitool: An open-source tool for controlling IPMI-enabled systems (GitHub.com)](https://github.com/ipmitool/ipmitool)
+[ipmitool/ipmitool: An open-source tool for controlling IPMI-enabled systems (GitHub.com)](https://github.com/ipmitool/ipmitool)
diff --git a/app/__init__.py b/app/__init__.py
index 4467afe..a86e271 100644
--- a/app/__init__.py
+++ b/app/__init__.py
@@ -8,4 +8,4 @@
 没有跨机通信、没有独立 agent —— 它只管部署它的这台机器的风扇。
 """
 
-__version__ = "0.1.0"
+__version__ = "2.0.0"