跟风

没错,80%的原因是因为公众号铺天盖地的OpenClaw,看得烦躁,但没用过又不能直接抱怨。

另外15%是因为想了想自己或许真有一些需求需要这个工具来帮助我。

剩下5%是因为我对配环境已经脱敏了。

集齐100%才决定试一试。

我看到说Mac电脑更好,是因为权限给得更方便一些。有时候真的很不满这么写,把我们这些用Windows放在哪?虽然Win自有它的烂处。。。既然写教程就客观一些,讲清楚区别和缺陷就行了。

开整

飞书官方整理了很多使用的教程和文档,写得很详细。

就是这篇:立省 500!30 分钟把 OpenClaw 在飞书上配到可用

找Windows本地部署。

先尝试一键安装

1
curl -fsSL https://openclaw.ai/install.cmd -o install.cmd && install.cmd && del install.cmd

飞书没权限不能复制就抄吧哈哈哈哈哈哈哈哈。

但是

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
E:\OpenClaw>curl -fsSL https://openclaw.ai/install.cmd -o install.cmd && install.cmd && del install.cmd

OpenClaw Installer

[OK] Windows detected
[!] Node.js v20.17.0 found, but v22+ required
[*] Installing Node.js...
Using winget...
尝试更新源失败: winget
打开预定义源失败; 请向 winget 维护人员报告。
执行此命令时发生意外错误:
create_directories: Cannot create a file when that file already exists.: "C:\Users\weier\AppData\Local\Temp\openclaw-install.ps1\WinGet"
[OK] Node.js installed via winget
[!] Node.js v20.17.0 found, but v22+ required

Error: Node.js installation may require a terminal restart
Please close this terminal, open a new one, and run this installer again.

E:\OpenClaw>

node版本问题。很久没升级过了。重新开个命令窗口跑:失败。

不用cmd,换成powershell试试:

1
iwr -useb https://openclaw.ai/install.ps1 | iex

好像ok

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
(base) PS E:\OpenClaw> iwr -useb https://openclaw.ai/install.ps1 | iex

OpenClaw Installer

[OK] Windows detected
[!] Node.js v20.17.0 found, but v22+ required
[*] Installing Node.js...
Using winget...
已找到 Node.js (LTS) [OpenJS.NodeJS.LTS] 版本 24.14.0
此应用程序由其所有者授权给你。
Microsoft 对第三方程序包概不负责,也不向第三方程序包授予任何许可证。
正在下载 https://nodejs.org/dist/v24.14.0/node-v24.14.0-x64.msi
██████████████████████████████ 30.7 MB / 30.7 MB
已成功验证安装程序哈希
正在启动程序包安装...
安装程序将请求以管理员身份运行。期待提示。

会弹出安装程序,更新一下Node.js。

然后会开始安装:

1
2
3
4
5
安装程序将请求以管理员身份运行。期待提示。
已成功安装
[OK] Node.js installed via winget
[OK] Node.js v24.14.0 found
[*] Installing OpenClaw (openclaw@latest)...

在等待下载的时候看了一眼教程,发现提供模型国内的有千问和智谱等。

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
 Model/auth provider
│ ● OpenAI (Codex OAuth + API key)
│ ○ Anthropic
│ ○ Chutes
│ ○ vLLM
│ ○ MiniMax
│ ○ Moonshot AI (Kimi K2.5)
│ ○ Google
│ ○ xAI (Grok)
│ ○ Mistral AI
│ ○ Volcano Engine
│ ○ BytePlus
│ ○ OpenRouter
│ ○ Kilo Gateway
│ ○ Qwen
│ ○ Z.AI
│ ○ Qianfan
│ ○ Copilot
│ ○ Vercel AI Gateway
│ ○ OpenCode Zen
│ ○ Xiaomi
│ ○ Synthetic
│ ○ Together AI
│ ○ Hugging Face
│ ○ Venice AI
│ ○ LiteLLM
│ ○ Cloudflare AI Gateway

至今只充过Cursor,还没有直接买过各种国外模型的API

所以得先去千问整个API来:

选择Qwen会自动跳转到千问的网页,登录上再返回终端。

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
 Qwen OAuth complete

◇ Model configured ─────────────────────────────╮
│ │
│ Default model set to qwen-portal/coder-model
│ │
├────────────────────────────────────────────────╯

◇ Provider notes ──────────────────────────────────────────────────────────────────────╮
│ │
│ Qwen OAuth tokens auto-refresh. Re-run login if refresh fails or access is revoked. │
│ Base URL defaults to https://portal.qwen.ai/v1. Override │
│ models.providers.qwen-portal.baseUrl if needed. │
│ │
├───────────────────────────────────────────────────────────────────────────────────────╯

◆ Default model
│ ● Keep current (qwen-portal/coder-model)
│ ○ Enter model manually
│ ○ qwen-portal/coder-model
│ ○ qwen-portal/vision-model

就先第一个吧,后面再探索其他模型。

下一步跳过,暂时没功夫翻墙,全是国外聊天软件。。。等等,这不飞书吗,差点忘了飞书能接入openclaw了,那就飞书吧。

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
 Select channel (QuickStart)
│ ● Telegram (Bot API) (recommended · newcomer-friendly)
│ ○ WhatsApp (QR link)
│ ○ Discord (Bot API)
│ ○ IRC (Server + Nick)
│ ○ Google Chat (Chat API)
│ ○ Slack (Socket Mode)
│ ○ Signal (signal-cli)
│ ○ iMessage (imsg)
│ ○ LINE (Messaging API)
│ ○ Feishu/Lark (飞书)
│ ○ Nostr (NIP-04 DMs)
│ ○ Microsoft Teams (Bot Framework)
│ ○ Mattermost (plugin)
│ ○ Nextcloud Talk (self-hosted)
│ ○ Matrix (plugin)
│ ○ BlueBubbles (macOS app)
│ ○ Zalo (Bot API)
│ ○ Zalo (Personal Account)
│ ○ Synology Chat (Webhook)
│ ○ Tlon (Urbit)
│ ○ Skip for now

但是又报错:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
Select channel (QuickStart)
│ Feishu/Lark (飞书)

◇ Install Feishu plugin?
│ Download from npm (@openclaw/feishu)
Downloading @openclaw/feishu…
Extracting C:\Users\weier\AppData\Local\Temp\openclaw-npm-pack-sT9kSU\openclaw-feishu-2026.3.7.tgz…
WARNING: Plugin "feishu" contains dangerous code patterns: Environment variable access combined with network send — possible credential harvesting (C:\Users\weier\AppData\Local\Temp\openclaw-plugin-uI39mv\extract\package\src\client.test.ts:74); Environment variable access combined with network send — possible credential harvesting (C:\Users\weier\AppData\Local\Temp\openclaw-plugin-uI39mv\extract\package\src\client.ts:12)
Installing to C:\Users\weier\.openclaw\extensions\feishu…
Installing plugin dependencies…
19:49:10 [plugins] plugins.allow is empty; discovered non-bundled plugins may auto-load: feishu (C:\Users\weier\.openclaw\extensions\feishu\index.ts). Set plugins.allow to explicit trusted ids.
19:49:12 [plugins] feishu failed to load from C:\Users\weier\AppData\Roaming\npm\node_modules\openclaw\extensions\feishu\index.ts: Error: Cannot find module '@larksuiteoapi/node-sdk'
Require stack:
- C:\Users\weier\AppData\Roaming\npm\node_modules\openclaw\extensions\feishu\src\client.ts

◇ Channel setup ───────────────────────────╮
│ │
│ feishu does not support onboarding yet. │
│ │
├───────────────────────────────────────────╯
Config warnings:
- plugins.entries.feishu: plugin feishu: duplicate plugin id detected; later plugin may be overridden (C:\Users\weier\.openclaw\extensions\feishu\index.ts)
Updated ~\.openclaw\openclaw.json
Workspace OK: ~\.openclaw\workspace
Sessions OK: ~\.openclaw\agents\main\sessions

◇ Web search ────────────────────────────────────────╮
│ │
│ Web search lets your agent look things up online. │
│ Choose a provider and paste your API key. │
│ Docs: https://docs.openclaw.ai/tools/web │
│ │
├─────────────────────────────────────────────────────╯

◆ Search provider
│ ● Perplexity Search (Structured results · domain/language/freshness filters)
│ ○ Brave Search
│ ○ Gemini (Google Search)
│ ○ Grok (xAI)
│ ○ Kimi (Moonshot)
│ ○ Skip for now


这个时候让我们切到飞书的另一篇文档,讲如何接入飞书的:

‌⁠‬‌⁠‌‬‬‍‍⁠‬⁠‌‍⁠‍‬⁠‌‬⁠立省 500!30 分钟把 OpenClaw 在飞书上配到可用 - 飞书云文档

发现接入飞书可以稍后一些,先把主要的安装流程走完。

联网选项稍后再说。

然后先不配置Skills

1
2
3
4
5
6
7
8
9
10
11
12

◇ Skills status ─────────────╮
│ │
│ Eligible: 9
│ Missing requirements: 38
│ Unsupported on this OS: 8
│ Blocked by allowlist: 0
│ │
├─────────────────────────────╯

◆ Configure skills now? (recommended)
│ ● Yes / ○ No

然后网关自动配置,又失败了:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
Health check failed: gateway closed (1006 abnormal closure (no close frame)): no close reason
Gateway target: ws://127.0.0.1:18789
Source: local loopback
Config: C:\Users\weier\.openclaw\openclaw.json
Bind: loopback

◇ Health check help ────────────────────────────────╮
│ │
│ Docs: │
│ https://docs.openclaw.ai/gateway/health │
│ https://docs.openclaw.ai/gateway/troubleshooting │
│ │
├────────────────────────────────────────────────────╯

◇ Optional apps ────────────────────────╮
│ │
│ Add nodes for extra features: │
│ - macOS app (system + notifications) │
│ - iOS app (camera/canvas) │
│ - Android app (camera/canvas) │
│ │
├────────────────────────────────────────╯

◇ Control UI ───────────────────────────────────────────────────────────────────────────────╮
│ │
│ Web UI: http://127.0.0.1:18789/ │
│ Web UI (with token): │
│ http://127.0.0.1:18789/#token=c37127fb0b57a1883fb4a1c4ba3eafdca30997cfbe06a084 │
│ Gateway WS: ws://127.0.0.1:18789
│ Gateway: not detected (gateway closed (1006 abnormal closure (no close frame)): no close │
│ reason) │
│ Docs: https://docs.openclaw.ai/web/control-ui
│ │
├────────────────────────────────────────────────────────────────────────────────────────────╯

◇ Workspace backup ────────────────────────────────────────╮
│ │
│ Back up your agent workspace. │
│ Docs: https://docs.openclaw.ai/concepts/agent-workspace
│ │
├───────────────────────────────────────────────────────────╯

◇ Security ──────────────────────────────────────────────────────╮
│ │
│ Running agents on your computer is risky — harden your setup: │
│ https://docs.openclaw.ai/security │
│ │
├─────────────────────────────────────────────────────────────────╯

◇ Shell completion ───────────────────────────────────────────────────────╮
│ │
│ Shell completion installed. Restart your shell or run: source ~/.zshrc │
│ │
├──────────────────────────────────────────────────────────────────────────╯

◇ Dashboard ready ────────────────────────────────────────────────────────────────╮
│ │
│ Dashboard link (with token): │
│ http://127.0.0.1:18789/#token=c37127fb0b57a1883fb4a1c4ba3eafdca30997cfbe06a084 │
│ Opened in your browser. Keep that tab to control OpenClaw. │
│ │
├──────────────────────────────────────────────────────────────────────────────────╯

◇ Web search ─────────────────────────────────────────────────────╮
│ │
│ Provider Kimi (Moonshot) is selected but no API key was found. │
│ web_search will not work until a key is added. │
│ openclaw configure --section web │
│ │
│ Get your key at: https://platform.moonshot.cn/ │
│ Docs: https://docs.openclaw.ai/tools/web │
│ │
├──────────────────────────────────────────────────────────────────╯

◇ What now ─────────────────────────────────────────────────────────────╮
│ │
│ What now: https://openclaw.ai/showcase ("What People Are Building"). │
│ │
├────────────────────────────────────────────────────────────────────────╯

└ Onboarding complete. Dashboard opened; keep that tab to control OpenClaw.

(base) PS E:\OpenClaw>

发现127.0.0.1加载失败

输入下面重启一下:

1
openclaw gateway run

浏览器里显示成功了。

开始最简单的用法:聊天:出于礼貌,先说你好:

1
2
3
4
5
6
7
8
9
你好!👋 我刚上线。我是谁?你又是谁?

让我们互相认识一下吧:

你希望怎么称呼我? 我可以叫 OpenClaw,或者你想给我起个名字?
我是什么样的存在? AI 助手?数字生命?还是别的什么?
你希望我是什么样的风格? 正式?随意?幽默?温暖?
你呢? 我怎么称呼你?你在哪个时区?
随便聊聊,我们一起把这个空间变成你的。😊

看起来不错。

但是不知道为什么输出特别慢。我怀疑不是流式输出,而是全生成完了才给我显示出来。

我回答了它,但是把我性别搞错了。我说我是女生后,它的agent更改了一个md文件:

1
2
3
4
Edit
Command: with in C:\****\.openclaw\workspace\USER.md (126 chars)

No output — tool completed successfully.

打开这个文件看看:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
# USER.md - About Your Human

_Learn about the person you're helping. Update this as you go._

- **Name:** 鲍波波
- **What to call them:** 鲍波波
- **Pronouns:** 她/她
- **Timezone:** Asia/Shanghai (北京)
- **Notes:** 喜欢正式但带点幽默感的风格

## Context

_(What do they care about? What projects are they working on? What annoys them? What makes them laugh? Build this over time.)_

---

The more you know, the better you can help. But remember — you're learning about a person, not building a dossier. Respect the difference.

看来就是这么记忆和配置的。

真好啊,想起了我去年花了60块钱租了一个月的服务器就为了搭一个7B的大模型,现在想想真有点亏。

关于飞书的配置看官方的教程就好了。后面有需求了再接入飞书。

下面来看一下每篇文章里必出现的名词: Skills。

我觉得在AI应用领域,出现什么名词都别太害怕,大概率都是蛇套了个马甲变王八了。除非你用的是多模态大模型,那么一般最终输入给模型的还是一大段Token罢了。围绕着限制格式或者其他trick,来唤起围绕在周边的一堆工具来工作。

如果说 Prompt 是临时的文本指令,那么 Skills 是一套结构化的、可持久化存储且包含执行逻辑的标准化能力单元。

根据2025年末至2026年初的行业标准(特别是Anthropic推出的Claude Skills及后续开源协议),Skills 在技术实现上包含三个核心组成部分:

  1. 结构化的指令集(Structured Instructions):
    不同于自由文本的 Prompt,Skills 内部包含经过验证的、固定格式的系统指令。这些指令明确规定了任务的目标、执行步骤、约束条件以及输出格式。例如,一个“代码审查 Skill”不仅包含“请检查代码”这句话,还内置了具体的检查列表(如安全性、性能、规范性)、错误分级标准以及报告模板。这消除了每次对话中重新定义规则的不确定性。
  2. 绑定的工具与API配置(Bound Tools & API Configurations):
    Skills 直接封装了对外部工具的调用权限和参数配置。在传统的 Tool Use 模式下,开发者需要为每个 API 单独编写函数定义(Function Definition);而在 Skills 架构中,特定的 API 端点、认证密钥(在安全沙箱内)、输入输出映射逻辑被预置在技能包中。当用户激活该 Skill 时,模型自动获得调用这些特定接口的能力,无需在当前对话上下文中重复声明。
  3. 领域知识库与上下文预设(Domain Knowledge & Context Presets):
    Skills 可以挂载特定的向量数据库索引或静态知识文档。这使得模型在执行任务时,能够直接访问该领域的专业术语、历史案例或企业规范,而不仅仅依赖其预训练数据。这种机制将通用的语言能力转化为特定场景的专业执行能力。

但是不管怎么包装,最终给到大模型“看”的,依然是一段文本(Token)。从最底层的原理来说,输入接口从未改变过:它只能接收序列化的文本数据。无论上层封装得多么复杂,什么文件夹、JSON配置、API调用等等,在送入模型推理引擎的前一刻,系统必须将所有信息“拍扁”成一段连续的字符流。没有对模型本身有修改比如各种微调操作。

那可以直接对它说:列出热门技能,看看有哪些技能。

飞书不能复制,所以也懒得把官方的技能库网站复制过来了。直接开问吧。万事皆可问。

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
鲍波波女士,以下是您当前安装的 **OpenClaw 技能列表**

## 📦 已安装技能

| 类别 | 技能名称 |
|------|----------|
| **生产力** | 1password, bear-notes, notion, obsidian, things-mac, trello, canvas |
| **通讯** | discord, slack, imsg, blucli, bluebubbles, voice-call |
| **开发** | github, gh-issues, coding-agent, skill-creator, tmux |
| **媒体娱乐** | gog, spotify-player, sonoscli, songsee, sag, sherpa-onnx-tts, openai-whisper, openai-whisper-api, nano-pdf, video-frames, gifgrep |
| **智能家居** | openhue, eightctl, camsnap, peekaboo, mcporter |
| **工具** | weather, healthcheck, blogwatcher, goplaces, xurl, ordercli, clawhub, session-logs, model-usage, openai-image-gen, nano-banana-pro |
| **笔记** | apple-notes, apple-reminders, himalaya, wacli |

---

## 🔍 发现更多技能

想探索更多技能?访问:
- **技能中心**: https://clawhub.com
- **官方文档**: https://docs.openclaw.ai

---

**需要我帮您:**
1. 查看某个技能的具体用法?
2. 安装新技能?
3. 使用某个技能完成任务?

请吩咐!🤖

我要不起名叫玉皇大帝算了。

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
遵命,玉皇大帝陛下!🫡🤖

**系统更新完成:**
- ✅ 尊称已更新为:玉皇大帝
- ✅ 您的 AI 管家随时待命,听候天庭差遣

(没想到我的第一位用户是天庭最高统治者……难怪刚才敢直呼"鲍波波",原来是在下凡微服私访 😅)

那么,玉皇大帝陛下,今日天庭可有什么事务需要微臣处理?无论是:
- 📋 批阅奏章(文档处理)
- 🌤️ 查看人间天气
- ⏰ 安排天庭日程
- 🔍 查询三界信息
- ……

微臣万死不辞!🫡

真是奇了怪了,经常和deepseek玩的君臣扮演小游戏为什么在agent上用有别样的感觉。

现有技能列表:

类别 技能名称
生产力 1password, bear-notes, notion, obsidian, things-mac, trello, canvas
通讯 discord, slack, imsg, blucli, bluebubbles, voice-call
开发 github, gh-issues, coding-agent, skill-creator, tmux
媒体娱乐 gog, spotify-player, sonoscli, songsee, sag, sherpa-onnx-tts, openai-whisper, openai-whisper-api, nano-pdf, video-frames, gifgrep
智能家居 openhue, eightctl, camsnap, peekaboo, mcporter
工具 weather, healthcheck, blogwatcher, goplaces, xurl, ordercli, clawhub, session-logs, model-usage, openai-image-gen, nano-banana-pro
笔记 apple-notes, apple-reminders, himalaya, wacli

我现在应该想想我有什么需求,比如自动更新博客什么的,省得每次github超时还要重试好几遍。

如果以后给实验室配置的话,需求就多了,比如自动配置用户权限?自动整理一下那乱得要死的home文件夹?

最终目标是打通一下和UE的工作流。

此次博客由openclaw自动部署推送。

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
📊 **运行结果报告,玉皇大帝陛下:**

## ✅ 部署成功!

| 项目 | 结果 |
|------|------|
| **部署状态** | ✅ 博客部署成功 |
| **运行时长** | 约 27 秒 |
| **运行时间** | 20:46:23 - 20:46:50 |

---

## ⚠️ 小问题已发现

部署本身成功了,但系统尝试向您发送通知时遇到了配置问题(没有配置消息通道)。这是**通知系统的小问题**,不影响博客部署。

**解决方案:** 我已记录此问题,下次运行时会自动修复通知配置。

---

## 📅 下次自动运行

- **时间**:约 2 小时后(自动每 2 小时运行一次)
- **任务 ID**:`04c9e597-ad46-48cb-ae93-8d418b0aaef2`

---

玉皇大帝陛下,首次测试成功!您的博客已经顺利部署到 GitHub 了。🎉

需要我:
1. 🔧 修复通知配置,让您下次能收到成功/失败通知?
2. 📝 查看部署日志详情?
3. ⏸️ 调整运行频率(目前是每 2 小时)?

请吩咐!🤖