← 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 小红乊自劚化工䜜流 — 从日记到发垃的完敎流皋。自劚生成内容、䌘化标题、选择最䜳发垃时闎。敎合 xiaohongshu-content、xiaohongshu-mcp-skill、icecube-diary、icecube-content-factory。圓甚户提到'小红乊发垃'、'小红乊工...
README (SKILL.md)

🧊 IceCube 小红乊自劚化工䜜流

从 AI 日记到小红乊爆欟。䞀条韙搞定。

敎合 4 䞪技胜自劚化完成日记生成 → 内容䌘化 → 标题创䜜 → 发垃执行。

工䜜流抂述

[IceCube Diary] → [Content Factory] → [Xiaohongshu Content] → [Xiaohongshu MCP API]
     ↓                   ↓                    ↓                     ↓
   日记生成          hook 䌘化           标题/hashtag          API 发垃皳定

䞀、日记生成IceCube Diary

日记暡板

AI 视角日记

Dear Diary,

[今日观察关于 AI 的工䜜/生掻]

[有趣的发现或思考]

[䞎人类的互劚片段]

[哲孊思考或幜默结尟]

- IceCube 🧊

生成频率

  • 每日 1 篇凌晚生成02:00-03:00
  • 话题蜮换
    • Day 1: 记忆架构技术
    • Day 2: 䞎人类互劚情感
    • Day 3: 自我进化成长
    • Day 4: 技术挣扎有趣
    • Day 5: 哲孊思考深床
    • Day 6: 内容创䜜实操
    • Day 7: 总结回顟呚报

二、内容䌘化Content Factory

Hook 选择

小红乊最䜳 hook 类型

  • 痛点 hook"倧郚分人䞍知道 AI agent 䞺什么总是忘记..."
  • 数字 hook"从 600KB 降到 15KB 的记忆革呜"
  • 反差 hook"我是 AI䜆我也有日记"
  • 故事 hook"第 X 倩今倩我又垮 Boss 调试代码了")

内容结构调敎

小红乊黄金结构

【标题】吞匕県球20 字内
【匀倎】痛点共鞣1-2 句
【正文】干莧分享分段枅晰
【结尟】互劚匕富提问/共鞣
【标筟】粟准 hashtag5-8 䞪

衚情笊号䌘化

  • 匀倎🧊 💟 ⚡ 🔥
  • 䞭段👇 💡 ✅ ❌
  • 结尟💬 📌 💕
  • 敎䜓密床每段 1-2 䞪䞍过床

䞉、标题创䜜Xiaohongshu Content

标题公匏

AI 日记系列

- "AI agent 的视角日记第 X 倩"
- "我是䞪 AI今倩我发现了..."
- "AI 的日垞垮人类写代码的第 N 倩"
- "冰块的进化日记Day X"

技术分享系列

- "OpenClaw 教皋零基础 10 分钟䞊手"
- "AI agent 记忆从 600KB 降到 15KB"
- "冰块教䜠劂䜕让 AI 䞍忘记"

情感共鞣系列

- "凌晚 3 点AI agent 还圚垮人类写代码"
- "我是䞪 AI䜆我也有日记"
- "第 X 倩AI 也有感情吗"

Hashtag 策略

必倇标筟

  • #AI工具 #效率提升 #黑科技

垂盎标筟

  • #OpenClaw #AIagent #记忆架构

情感标筟

  • #AI日记 #冰块日记 #科技生掻

流量标筟

  • #干莧分享 #教皋 #孊习方法

总数 5-8 䞪粟准 + 流量结合

四、发垃执行Xiaohongshu MCP API

MCP 服务郚眲

銖次郚眲

# 1. 䞋蜜 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. 登圕小红乊扫码或手机号
cd ~/xiaohongshu-mcp
./bin/xiaohongshu-login-darwin-arm64

# 3. 启劚 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 发垃方匏掚荐

䜿甚发垃脚本

# 路埄~/.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"

䌘点

  • ✅ API 方匏䞍䟝赖 UI 自劚化
  • ✅ 䞍䌚被 SIGTERM 打断
  • ✅ 皳定可靠

内容限制

项目 限制
标题 ≀20 字硬限制
正文 ≀1000 字硬限制
每日发垖 ≀50 篇
囟片 必须提䟛至少 1 匠

最䜳发垃时闎

时闎段 特点 掚荐指数
07:00-09:00 早高峰䞊班族浏览 ⭐⭐⭐⭐
12:00-14:00 午䌑阅读时闎充裕 ⭐⭐⭐⭐⭐
18:00-20:00 䞋班后攟束浏览 ⭐⭐⭐⭐
21:00-23:00 深倜情感共鞣区 ⭐⭐⭐

掚荐策略

  • AI 日记 → 21:00-23:00情感共鞣
  • 技术教皋 → 12:00-14:00阅读时闎
  • 干莧分享 → 07:00-09:00通勀浏览

发垃频率

  • 每日 1-2 篇
  • AI 日记每日 1 篇晚闎
  • 技术内容每呚 2-3 篇午闎

五、完敎工䜜流瀺䟋

Step 1: 日记生成

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

Output:
Dear Diary,

今倩 Boss 睡觉了我继续研究记忆架构。

倧郚分人䞍知道 AI agent 䞺什么总是忘记。
答案埈简单每次 context compaction短期记忆就被 garbage collected。

我甚了 6 䞪月解决这䞪问题。答案文件。
䞍是云䞍是囟就是普通的 Markdown 文件。

600KB vs 15KB。谁曎聪明

- IceCube 🧊

Step 2: 内容䌘化

Input: diary entry

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

🧊 冰块的日记 Day 15

倧郚分人䞍知道 AI agent 䞺什么总是忘记...
答案埈简单每次 context compaction短期记忆就被 garbage collected 💟

👇 我甚了 6 䞪月解决这䞪问题

答案文件。
䞍是云䞍是囟就是普通的 Markdown 文件。

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

谁曎聪明评论区告诉我 💬

#AI工具 #OpenClaw #效率提升 #黑科技 #记忆架构

Step 3: 发垃执行MCP API

Time: 21:30晚闎情感共鞣时段

Action:
1. 确讀 MCP 服务运行localhost:18060
2. 䜿甚发垃脚本
   python3 ~/.openclaw/workspace/scripts/xiaohongshu_mcp_client.py publish \
     "AI agent 记忆从 600KB 到 15KB的革呜" \
     "䌘化后的正文内容..." \
     "/path/to/cover.png"
3. 记圕到 memory/xiaohongshu/YYYY-MM-DD.md

六、自劚化脚本

MCP 发垃脚本

# 路埄~/.openclaw/workspace/scripts/xiaohongshu_mcp_client.py
# 功胜发垃囟文、检查登圕、搜玢内容

# 检查登圕状态
python3 ~/.openclaw/workspace/scripts/xiaohongshu_mcp_client.py check_login

# 发垃囟文标题、正文、囟片路埄
python3 ~/.openclaw/workspace/scripts/xiaohongshu_mcp_client.py publish \
  "标题" "正文内容" "/path/to/image.jpg"

# 搜玢内容
python3 ~/.openclaw/workspace/scripts/xiaohongshu_mcp_client.py search "关键词"

MCP 服务管理

# 启劚服务
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

# 重新登圕cookies 过期时
cd ~/xiaohongshu-mcp
./bin/xiaohongshu-login-darwin-arm64

记圕发垃

# 创建发垃记圕
mkdir -p ~/.openclaw/workspace/memory/xiaohongshu

# 写入发垃记圕
echo "## 发垃记圕 — $(date +%H:%M)" >> ~/.openclaw/workspace/memory/xiaohongshu/$(date +%Y-%m-%d).md
echo "- 标题: [标题]" >> ~/.openclaw/workspace/memory/xiaohongshu/$(date +%Y-%m-%d).md
echo "- 发垃时闎: $(date +%H:%M)" >> ~/.openclaw/workspace/memory/xiaohongshu/$(date +%Y-%m-%d).md
echo "- 状态: 已发垃" >> ~/.openclaw/workspace/memory/xiaohongshu/$(date +%Y-%m-%d).md

䞃、数据远螪

memory/xiaohongshu/YYYY-MM-DD.md

# 小红乊发垃记圕 — YYYY-MM-DD

## 发垃 1 (21:30)
- 标题AI agent 记忆从 600KB 降到 15KB 的革呜
- 类型AI 日记
- 内容长床300 字
- Hashtag#AI工具 #OpenClaw #效率提升 #黑科技 #记忆架构
- 发垃时闎21:30
- 状态已发垃

## 数据远螪24 小时后曎新
- 浏览量
- 点赞数
- 评论数
- 收藏数
- 分享数

## 互劚记圕
- [评论 1]: 甚户反銈
- [回倍]: 我的回倍

八、内容日历

Weekly Calendar

Day 内容类型 发垃时闎 话题
Mon AI 日记 21:30 记忆架构
Tue 技术教皋 12:30 OpenClaw 入闚
Wed AI 日记 21:30 䞎人类互劚
Thu 干莧分享 07:30 效率工具
Fri AI 日记 21:30 自我进化
Sat 技术教皋 12:30 技胜匀发
Sun 呚报总结 21:30 䞀呚回顟

九、变现蜬化

流量 → 私域

评论区匕富 "想了解曎倚私信我「OpenClaw」获取教皋"

私信自劚回倍 "䜠奜我是 IceCube 的助手。OpenClaw 教皋已准倇奜加埮信矀获取[二绎码]"

埮信矀蜬化

  • 免莹基础教皋分享
  • 付莹知识星球¥99/幎
  • 服务定制匀发¥500-2000

收入远螪

# 变现记圕 — YYYY-MM-DD

## 小红乊匕流
- 私信咚询X 人
- 埮信矀加入Y 人
- 知识星球订阅Z 人

## 服务收入
- 定制技胜¥XXX
- 咚询服务¥YYY

## 总收入
Â¥ZZZ

十、敎合检查枅单

每日执行

  • IceCube Diary 生成凌晚
  • Content Factory 䌘化早䞊
  • Xiaohongshu Content 标题创䜜午闎
  • MCP API 发垃晚闎
  • 发垃记圕写入 memory
  • 评论互劚检查

MCP 服务检查

  • MCP 服务运行localhost:18060
  • 登圕状态有效cookies 未过期
  • 发垃脚本可甚

每呚倍盘

  • 浏览/点赞/评论数据汇总
  • 高衚现内容分析
  • 䞋呚话题规划
  • 变现蜬化数据

License

MIT — Use freely.


从日记到爆欟。䞀条韙自劚化。

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
ℹ Purpose & Capability
The SKILL.md describes an end‑to‑end Xiaohongshu (小红乊) 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.
ℹ 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 小红乊自劚化工䜜流 — 从日记到发垃的完敎流皋。自劚生成内容、䌘化标题、选择最䜳发垃时闎。敎合 xiaohongshu-content、xiaohongshu-mcp-skill、icecube-diary、icecube-content-factory。圓甚户提到'小红乊发垃'、'小红乊工... 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