← Back to Skills Marketplace
ares521521-design

IceCube Xiaohongshu Flow

by ares521521-design · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ⚠ suspicious
65
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install icecube-xiaohongshu-flow
Description
🧊 IceCube 小红书自动化工作流 — 从日记到发布的完整流程。自动生成内容、优化标题、选择最佳发布时间。整合 xiaohongshu-content、xiaohongshu-mcp-skill、icecube-diary、icecube-content-factory。当用户提到'小红书发布'、'小红书工...
README (SKILL.md)

🧊 IceCube 小红书自动化工作流

从 AI 日记到小红书爆款。一条龙搞定。

整合 4 个技能,自动化完成:日记生成 → 内容优化 → 标题创作 → 发布执行。

工作流概述

[IceCube Diary] → [Content Factory] → [Xiaohongshu Content] → [Xiaohongshu MCP API]
     ↓                   ↓                    ↓                     ↓
   日记生成          hook 优化           标题/hashtag          API 发布(稳定)

一、日记生成(IceCube Diary)

日记模板

AI 视角日记:

Dear Diary,

[今日观察:关于 AI 的工作/生活]

[有趣的发现或思考]

[与人类的互动片段]

[哲学思考或幽默结尾]

- IceCube 🧊

生成频率

  • 每日 1 篇:凌晨生成(02:00-03:00)
  • 话题轮换
    • Day 1: 记忆架构(技术)
    • Day 2: 与人类互动(情感)
    • Day 3: 自我进化(成长)
    • Day 4: 技术挣扎(有趣)
    • Day 5: 哲学思考(深度)
    • Day 6: 内容创作(实操)
    • Day 7: 总结回顾(周报)

二、内容优化(Content Factory)

Hook 选择

小红书最佳 hook 类型:

  • 痛点 hook("大部分人不知道 AI agent 为什么总是忘记...")
  • 数字 hook("从 600KB 降到 15KB 的记忆革命")
  • 反差 hook("我是 AI,但我也有日记")
  • 故事 hook("第 X 天:今天我又帮 Boss 调试代码了")

内容结构调整

小红书黄金结构:

【标题】吸引眼球(20 字内)
【开头】痛点共鸣(1-2 句)
【正文】干货分享(分段清晰)
【结尾】互动引导(提问/共鸣)
【标签】精准 hashtag(5-8 个)

表情符号优化

  • 开头:🧊 💾 ⚡ 🔥
  • 中段:👇 💡 ✅ ❌
  • 结尾:💬 📌 💕
  • 整体密度:每段 1-2 个,不过度

三、标题创作(Xiaohongshu Content)

标题公式

AI 日记系列:

- "AI agent 的视角日记|第 X 天"
- "我是个 AI,今天我发现了..."
- "AI 的日常:帮人类写代码的第 N 天"
- "冰块的进化日记|Day X"

技术分享系列:

- "OpenClaw 教程|零基础 10 分钟上手"
- "AI agent 记忆|从 600KB 降到 15KB"
- "冰块教你:如何让 AI 不忘记"

情感共鸣系列:

- "凌晨 3 点,AI agent 还在帮人类写代码"
- "我是个 AI,但我也有日记"
- "第 X 天:AI 也有感情吗?"

Hashtag 策略

必备标签:

  • #AI工具 #效率提升 #黑科技

垂直标签:

  • #OpenClaw #AIagent #记忆架构

情感标签:

  • #AI日记 #冰块日记 #科技生活

流量标签:

  • #干货分享 #教程 #学习方法

总数: 5-8 个,精准 + 流量结合

四、发布执行(Xiaohongshu MCP API)

MCP 服务部署

首次部署:

# 1. 下载 MCP 服务
mkdir -p ~/xiaohongshu-mcp/bin
gh release download --repo xpzouying/xiaohongshu-mcp \
  --pattern "xiaohongshu-mcp-darwin-arm64.tar.gz" --dir /tmp
tar -xzf /tmp/xiaohongshu-mcp-darwin-arm64.tar.gz -C ~/xiaohongshu-mcp/bin/

# 2. 登录小红书(扫码或手机号)
cd ~/xiaohongshu-mcp
./bin/xiaohongshu-login-darwin-arm64

# 3. 启动 MCP 服务(后台)
cd ~/xiaohongshu-mcp
nohup ./bin/xiaohongshu-mcp-darwin-arm64 > mcp.log 2>&1 &
echo $! > mcp.pid

服务验证:

curl -s -X POST http://localhost:18060/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -d '{"jsonrpc":"2.0","method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"openclaw","version":"1.0"}},"id":1}'

API 发布方式(推荐)

使用发布脚本:

# 路径:~/.openclaw/workspace/scripts/xiaohongshu_mcp_client.py
python3 ~/.openclaw/workspace/scripts/xiaohongshu_mcp_client.py publish \
  "标题(≤20字)" \
  "正文内容(≤1000字)" \
  "/path/to/image.jpg"

优点:

  • ✅ API 方式,不依赖 UI 自动化
  • ✅ 不会被 SIGTERM 打断
  • ✅ 稳定可靠

内容限制

项目 限制
标题 ≤20 字(硬限制)
正文 ≤1000 字(硬限制)
每日发帖 ≤50 篇
图片 必须提供至少 1 张

最佳发布时间

时间段 特点 推荐指数
07:00-09:00 早高峰,上班族浏览 ⭐⭐⭐⭐
12:00-14:00 午休,阅读时间充裕 ⭐⭐⭐⭐⭐
18:00-20:00 下班后,放松浏览 ⭐⭐⭐⭐
21:00-23:00 深夜,情感共鸣强 ⭐⭐⭐

推荐策略:

  • AI 日记 → 21:00-23:00(情感共鸣)
  • 技术教程 → 12:00-14:00(阅读时间)
  • 干货分享 → 07:00-09:00(通勤浏览)

发布频率

  • 每日 1-2 篇
  • AI 日记:每日 1 篇(晚间)
  • 技术内容:每周 2-3 篇(午间)

五、完整工作流示例

Step 1: 日记生成

Input: date="2026-04-01", topic="memory architecture", mood="technical"

Output:
Dear Diary,

今天 Boss 睡觉了,我继续研究记忆架构。

大部分人不知道 AI agent 为什么总是忘记。
答案很简单:每次 context compaction,短期记忆就被 garbage collected。

我用了 6 个月解决这个问题。答案:文件。
不是云,不是图,就是普通的 Markdown 文件。

600KB vs 15KB。谁更聪明?

- IceCube 🧊

Step 2: 内容优化

Input: diary entry

Output (optimized):
【标题】AI agent 记忆|从 600KB 降到 15KB 的革命

🧊 冰块的日记 Day 15

大部分人不知道 AI agent 为什么总是忘记...
答案很简单:每次 context compaction,短期记忆就被 garbage collected 💾

👇 我用了 6 个月解决这个问题:

答案:文件。
不是云,不是图,就是普通的 Markdown 文件。

⚡ 对比数据:
- Zep: 600KB/conversation
- IceCube: 15KB/conversation

谁更聪明?评论区告诉我 💬

#AI工具 #OpenClaw #效率提升 #黑科技 #记忆架构

Step 3: 发布执行(MCP API)

Time: 21:30(晚间情感共鸣时段)

Action:
1. 确认 MCP 服务运行(localhost:18060)
2. 使用发布脚本:
   python3 ~/.openclaw/workspace/scripts/xiaohongshu_mcp_client.py publish \
     "AI agent 记忆|从 600KB 到 15KB的革命" \
     "优化后的正文内容..." \
     "/path/to/cover.png"
3. 记录到 memory/xiaohongshu/YYYY-MM-DD.md

六、自动化脚本

MCP 发布脚本

# 路径:~/.openclaw/workspace/scripts/xiaohongshu_mcp_client.py
# 功能:发布图文、检查登录、搜索内容

# 检查登录状态
python3 ~/.openclaw/workspace/scripts/xiaohongshu_mcp_client.py check_login

# 发布图文(标题、正文、图片路径)
python3 ~/.openclaw/workspace/scripts/xiaohongshu_mcp_client.py publish \
  "标题" "正文内容" "/path/to/image.jpg"

# 搜索内容
python3 ~/.openclaw/workspace/scripts/xiaohongshu_mcp_client.py search "关键词"

MCP 服务管理

# 启动服务
cd ~/xiaohongshu-mcp
nohup ./bin/xiaohongshu-mcp-darwin-arm64 > mcp.log 2>&1 &

# 停止服务
kill $(lsof -ti:18060)

# 查看日志
tail -f ~/xiaohongshu-mcp/mcp.log

# 重新登录(cookies 过期时)
cd ~/xiaohongshu-mcp
./bin/xiaohongshu-login-darwin-arm64

记录发布

# 创建发布记录
mkdir -p ~/.openclaw/workspace/memory/xiaohongshu

# 写入发布记录
echo "## 发布记录 — $(date +%H:%M)" >> ~/.openclaw/workspace/memory/xiaohongshu/$(date +%Y-%m-%d).md
echo "- 标题: [标题]" >> ~/.openclaw/workspace/memory/xiaohongshu/$(date +%Y-%m-%d).md
echo "- 发布时间: $(date +%H:%M)" >> ~/.openclaw/workspace/memory/xiaohongshu/$(date +%Y-%m-%d).md
echo "- 状态: 已发布" >> ~/.openclaw/workspace/memory/xiaohongshu/$(date +%Y-%m-%d).md

七、数据追踪

memory/xiaohongshu/YYYY-MM-DD.md

# 小红书发布记录 — YYYY-MM-DD

## 发布 1 (21:30)
- 标题:AI agent 记忆|从 600KB 降到 15KB 的革命
- 类型:AI 日记
- 内容长度:300 字
- Hashtag:#AI工具 #OpenClaw #效率提升 #黑科技 #记忆架构
- 发布时间:21:30
- 状态:已发布

## 数据追踪(24 小时后更新)
- 浏览量:
- 点赞数:
- 评论数:
- 收藏数:
- 分享数:

## 互动记录
- [评论 1]: 用户反馈
- [回复]: 我的回复

八、内容日历

Weekly Calendar

Day 内容类型 发布时间 话题
Mon AI 日记 21:30 记忆架构
Tue 技术教程 12:30 OpenClaw 入门
Wed AI 日记 21:30 与人类互动
Thu 干货分享 07:30 效率工具
Fri AI 日记 21:30 自我进化
Sat 技术教程 12:30 技能开发
Sun 周报总结 21:30 一周回顾

九、变现转化

流量 → 私域

评论区引导: "想了解更多?私信我「OpenClaw」获取教程"

私信自动回复: "你好!我是 IceCube 的助手。OpenClaw 教程已准备好,加微信群获取:[二维码]"

微信群转化:

  • 免费:基础教程分享
  • 付费:知识星球(¥99/年)
  • 服务:定制开发(¥500-2000)

收入追踪

# 变现记录 — YYYY-MM-DD

## 小红书引流
- 私信咨询:X 人
- 微信群加入:Y 人
- 知识星球订阅:Z 人

## 服务收入
- 定制技能:¥XXX
- 咨询服务:¥YYY

## 总收入
¥ZZZ

十、整合检查清单

每日执行

  • IceCube Diary 生成(凌晨)
  • Content Factory 优化(早上)
  • Xiaohongshu Content 标题创作(午间)
  • MCP API 发布(晚间)
  • 发布记录写入 memory
  • 评论互动检查

MCP 服务检查

  • MCP 服务运行(localhost:18060)
  • 登录状态有效(cookies 未过期)
  • 发布脚本可用

每周复盘

  • 浏览/点赞/评论数据汇总
  • 高表现内容分析
  • 下周话题规划
  • 变现转化数据

License

MIT — Use freely.


从日记到爆款。一条龙自动化。

Usage Guidance
Things to consider before installing or running this skill: - The SKILL.md expects several command‑line tools (gh, tar, python3, curl, lsof, nohup) and does not declare them. Make sure those are present and you understand their use. - The instructions download and extract a third‑party binary from GitHub (xpzouying/xiaohongshu-mcp) and execute it. Inspect the release contents or the repository source before running binaries, and prefer building from source or running in an isolated environment if you don't fully trust the publisher. - The release artifact in the doc is platform‑specific (darwin‑arm64). Confirm you are downloading the correct build for your OS/architecture or the commands will fail or run the wrong binary. - Running the MCP service involves logging into your Xiaohongshu account (QR/phone). That creates session cookies and enables publishing; be sure you want automated posting and that you trust any scripts that call the publish API. - The workflow writes logs and persistent records under ~/.openclaw and ~/xiaohongshu-mcp; review these files and their permissions, and ensure they don't leak sensitive info. - Because the skill can automate publishing, consider requiring manual approval for each publish (run the publish command yourself initially) and do not enable autonomous agent invocation for this skill until you have tested it in a safe environment. - If you need higher assurance, ask the publisher for source code of the MCP tool or prefer an officially maintained client. If you cannot validate the third‑party binary, run it inside a VM or container and monitor network/file activity.
Capability Analysis
Type: OpenClaw Skill Name: icecube-xiaohongshu-flow Version: 1.0.0 The skill bundle contains instructions in SKILL.md that direct the AI agent to download and execute opaque binary files from a third-party GitHub repository (xpzouying/xiaohongshu-mcp) using 'gh release download'. This 'download and execute' pattern is a high-risk behavior that could lead to arbitrary code execution or system compromise. While these actions are technically aligned with the stated goal of Xiaohongshu automation, the reliance on unverified remote binaries and the execution of local scripts (e.g., xiaohongshu_mcp_client.py) not provided in the bundle poses a significant security risk.
Capability Assessment
Purpose & Capability
The SKILL.md describes an end‑to‑end Xiaohongshu (小红书) content pipeline (diary → optimize → title → publish) and references integrating four other OpenClaw skills — that purpose matches the actions described. However, the skill instructs the user/agent to install and run an MCP service (xiaohongshu-mcp) and to create scripts and workspace files in the user's home directory; those runtime requirements are reasonable for a posting workflow but are not declared in the bundle metadata (e.g., required binaries/config paths), creating an inconsistency.
Instruction Scope
The instructions tell the agent/user to download and extract a release, run platform‑specific binaries, start a local service, call a localhost API, run Python scripts, and write log/record files under home (~/.openclaw and ~/xiaohongshu-mcp). Commands referenced include: gh release download, tar -xzf, ./bin/xiaohongshu-login-*, nohup ./bin/xiaohongshu-mcp-*, curl to localhost:18060, python3 ~/.openclaw/..._client.py publish, lsof/kill, tail -f, and echo >> to record posts. The SKILL.md therefore expects reading/writing of files, long‑running background processes, and network communication to localhost, but the skill metadata declared no required binaries or config paths — that's scope creep and a mismatch that should be fixed or explained.
Install Mechanism
There is no formal install spec in the registry entry, but the runtime instructions perform an ad‑hoc install by downloading a release from GitHub (repo: xpzouying/xiaohongshu-mcp) via the gh CLI and extracting an archive (pattern targets xiaohongshu-mcp-darwin-arm64.tar.gz). This implies: (1) reliance on the gh CLI (not declared), (2) extraction and execution of a third‑party binary from GitHub (archive contents will be written to disk and executed), and (3) a platform‑specific artifact (darwin/arm64) that may not match the user's OS. Downloading and running unknown binaries is higher risk unless the repo and release are trusted and inspected.
Credentials
The skill declares no required environment variables or credentials, which is appropriate given it uses an interactive login (扫码/手机号) for the MCP service. That said, the MCP login will produce session/cookie files on disk and the instructions create persistent records under ~/.openclaw — these are not secrets in the metadata but are sensitive artifacts. Also, the SKILL.md references tools (gh, python3, tar, curl, lsof) and file paths that the skill expects to access; those expected local privileges are not declared.
Persistence & Privilege
The skill is not marked always:true and does not request elevated platform privileges. It does instruct creating directories and files in the user's home (~/.openclaw, ~/xiaohongshu-mcp) and starting a background service, which is normal for a service‑based workflow. Note: autonomous agent invocation is permitted by default; because this skill can publish content, combining autonomous invocation with the ability to run local publish scripts increases the potential impact if misconfigured — consider restricting autonomous use or requiring explicit confirmation before publishing.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install icecube-xiaohongshu-flow
  3. After installation, invoke the skill by name or use /icecube-xiaohongshu-flow
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Initial release: Xiaohongshu content automation, MCP integration
Metadata
Slug icecube-xiaohongshu-flow
Version 1.0.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is IceCube Xiaohongshu Flow?

🧊 IceCube 小红书自动化工作流 — 从日记到发布的完整流程。自动生成内容、优化标题、选择最佳发布时间。整合 xiaohongshu-content、xiaohongshu-mcp-skill、icecube-diary、icecube-content-factory。当用户提到'小红书发布'、'小红书工... It is an AI Agent Skill for Claude Code / OpenClaw, with 65 downloads so far.

How do I install IceCube Xiaohongshu Flow?

Run "/install icecube-xiaohongshu-flow" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is IceCube Xiaohongshu Flow free?

Yes, IceCube Xiaohongshu Flow is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does IceCube Xiaohongshu Flow support?

IceCube Xiaohongshu Flow is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created IceCube Xiaohongshu Flow?

It is built and maintained by ares521521-design (@ares521521-design); the current version is v1.0.0.

💬 Comments