Name: Mijia
Author: Kai Wang

據 AlphaMoat 統計,截至 2026-10,Mijia 累計下載 5,066 次、獲 4 個星標,按累計下載量在本次參與排名的 10,914 個 AI 技能中排名第 5,753 位,在「AI 智慧體」分類(1,632 個)中排名第 810 位。

列表預設按累計下載量排序,也可按星標排序;詳情頁排名按累計下載量計算。

累計下載

5066

星標

全站下載量排名

#5,753

全站前 53% · 「AI 智慧體」前 50%

平台	累計下載	星標	評論
ClawHub ↗	3736	3	-
SkillHub ↗	5066	4	—

各平台的計數口徑不同,統計時間不一致。

這個技能是做什麼的

支援查看燈具狀態、開關燈、調節亮度與色溫以及切換照明模式。執行需設定裝置 ID，並透過小米帳號登入。命令會實際改變裝置狀態；執行開關、亮度或模式操作前，應逐項確認使用者授權和目標裝置。文件設定需要 uv、Python >=3.13 與 mijiaapi；核實實際技能路徑及相依套件鎖定檔，設定 MIJIA_LAMP_DID，並完成首次小米掃碼登入。

資料釋出於 2026-10-27

可以用來做什麼

查看已設定燈具狀態
在獲得確認後調整燈光亮度、色溫或模式

什麼時候用

裝置 ID 與小米帳號已設定，且使用者明確確認具體裝置及變更時

示例提示詞

以下是根據技能文件生成的示例,不代表實際執行結果。

先說明控制燈具前需要設定哪些資訊，不執行任何操作。

請列出改變米家燈狀態前需要我確認的裝置和設定。

月度資料趨勢

累計下載5066

星標4

2026-09 起統計方法調整:同名技能按作者分別統計,並覆蓋全部技能。圖中虛線為調整前的月份,與之後的資料不直接比較,也不計算跨越調整的增長。

月份	累計下載	環比	星標
2026-10	5066	—	4
2026-09	3729	—	4
2026-08	3478	+4.4%	4
2026-07	3331	+8.8%	4
2026-06	3062	+4.8%	3
2026-05	2921	—	4
2026-03	2514	—	3

SKILL.md 原文

以下是作者釋出的原始檔案,按純文字顯示,本站未做修改或稽核;其中的連結不可點選,命令請核實後再執行。抓取於 2026-10-23。

3.0 KB · ClawHub v0.1.0

---
name: mijia
description: Control Xiaomi Mijia smart home devices. Use this skill when the user wants to control desk lamps, smart plugs, or other Mijia devices. Supports turning lights on/off, adjusting brightness, setting color temperature, switching modes, and more.
invocable: true
---

# Mijia Smart Home Control

Control Xiaomi Mijia smart devices via the mijiaAPI.

展開全部(共 101 行)

## Setup

Before using this skill, you need to:

1. Install dependencies:
```bash
cd /path/to/mijia-skill
uv sync
```

2. Set your device ID as an environment variable:
```bash
export MIJIA_LAMP_DID="your_device_id"
```

3. First run will prompt for Xiaomi account login via QR code.

## Finding Device IDs

To find your device IDs, use the mijia-api library:

```python
from mijiaAPI import mijiaAPI
api = mijiaAPI()
api.login()
devices = api.get_device_list()
for d in devices:
    print(f"{d['name']}: {d['did']}")
```

## How to Use

Skill path: `~/.clawdbot/skills/mijia`

### Lamp Control Commands

```bash
# Navigate to skill directory
cd ~/.claude/skills/mijia

# Check status
uv run python scripts/lamp_cli.py status

# Turn on/off
uv run python scripts/lamp_cli.py on
uv run python scripts/lamp_cli.py off
uv run python scripts/lamp_cli.py toggle

# Adjust brightness (1-100%)
uv run python scripts/lamp_cli.py brightness 50

# Adjust color temperature (2700-6500K)
uv run python scripts/lamp_cli.py temp 4000

# Set mode
uv run python scripts/lamp_cli.py mode reading    # Reading mode
uv run python scripts/lamp_cli.py mode computer   # Computer mode
uv run python scripts/lamp_cli.py mode night      # Night reading
uv run python scripts/lamp_cli.py mode antiblue   # Anti-blue light
uv run python scripts/lamp_cli.py mode work       # Work mode
uv run python scripts/lamp_cli.py mode candle     # Candle effect
uv run python scripts/lamp_cli.py mode twinkle    # Twinkle alert
```

## Natural Language Understanding

When the user says the following, execute the corresponding command:

| User Says | Command |
|-----------|---------|
| Turn on the light / open lamp | `scripts/lamp_cli.py on` |
| Turn off the light / close lamp | `scripts/lamp_cli.py off` |
| Toggle the light | `scripts/lamp_cli.py toggle` |
| Brighter / more bright | Check status first, then increase by 20-30% |
| Dimmer / less bright | Check status first, then decrease by 20-30% |
| Full brightness / maximum | `scripts/lamp_cli.py brightness 100` |
| Minimum brightness | `scripts/lamp_cli.py brightness 1` |
| Warm light | `scripts/lamp_cli.py temp 2700` |
| Cool light / white light | `scripts/lamp_cli.py temp 6500` |
| Reading mode | `scripts/lamp_cli.py mode reading` |
| Computer mode | `scripts/lamp_cli.py mode computer` |
| Night mode | `scripts/lamp_cli.py mode night` |
| Lamp status / what's the light status | `scripts/lamp_cli.py status` |

## Before Executing

1. Navigate to skill directory: `cd ~/.clawdbot/skills/mijia`
2. Ensure `MIJIA_LAMP_DID` environment variable is set
3. Run with uv: `uv run python scripts/lamp_cli.py <command>`
4. Report the result to the user after execution