Token导航 LogoToken导航

Agent Skill · 研究检索

xiaohongshu-article-packager(小红书文章打包器)

xiaohongshu-article-packager 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

xiaohongshu-article-packager 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为执行命令;许可证信息还不明确,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 63,891。

总安装 309周安装 13GitHub Stars 公开资料未说明下载 108核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:xiaohongshu-article-packager(小红书文章打包器)
来源仓库:https://github.com/postplusai/postplus-skills
仓库路径:skills/xiaohongshu-article-packager
安装命令:
npx skills add https://github.com/postplusai/postplus-skills --skill xiaohongshu-article-packager
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/postplusai/postplus-skills --skill xiaohongshu-article-packager

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

xiaohongshu-article-packager 用于查找、检索和筛选相关信息。适合在 Codex、Claude、Cursor、Gemini CLI 中根据关键词、任务场景或来源线索快速定位候选结果。通过 npx skills add 命令从指定 GitHub 仓库安装,需结合原始 README 核验具体用法。

安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。可结合来源仓库和 SKILL.md 进一步了解功能边界和使用限制。

来源仓库内路径:skills/xiaohongshu-article-packager

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/postplusai/postplus-skills --skill xiaohongshu…;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要执行命令,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

xiaohongshu-article-packager 怎么安装?

在终端执行 npx skills add https://github.com/postplusai/postplus-skills --skill xiaohongshu-article-p…。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

xiaohongshu-article-packager 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

xiaohongshu-article-packager 需要什么权限?

权限级别是执行命令。安装流程涉及命令执行,可能通过 npx skills add https://github.com/postplusai/postplus-skills --skill xiaohongshu-article-packager 联网下载 Skill 或依赖。用户安装前应确认命令来源、仓库内容和执行环境。

xiaohongshu-article-packager 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

xiaohongshu-article-packager 用的人多吗?

公开数据是累计安装 309 次,近一周安装 13 次,下载 108 次。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

xiaohongshu-article-packager 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。

先想清楚的地方

  • 需要执行命令,装到常用环境前要先评估影响面。
  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 不能接受 Skill 在你机器上执行命令的人。
  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
xiaohongshu-article-packager

小红书文章打包器 · 本页

309—执行命令未明确有提醒

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

Xiaohongshu Article Packager

Follow shared release-shell rules in:

  • postplus-shared release-shell rules

Use this skill when the user wants:

  • 一篇原稿变成可发的小红书图文
  • 口播稿改成图文分页
  • 给一段观点内容做分页、层级、配图和可渲染页面
  • 产出适合后续排版或渲染的结构化内容包,但保留原文

This skill is for layout packaging. It is not for:

  • platform scraping
  • competitor research
  • direct posting
  • fabricating claims not present in the source
  • rewriting the source by default

If the request is mainly about platform research, use skills/20-research/xiaohongshu-research.

Core Idea

Do not turn the source into a different piece of writing.

The job has two layers:

  1. structure without rewrite

- split the source into readable pages - choose which original lines deserve larger emphasis - preserve the original sequence unless the user asks to reorder

  1. render packaging

- assign page roles - place images - prepare clean layout instructions or HTML/CSS for rendering

The default goal is:

  • readable
  • faithful to the original wording
  • faithful to the original information density
  • faithful to the original voice

Inputs

Typical source inputs:

  • a markdown file
  • a long paragraph dump
  • a script written for talking-head video
  • rough bullets
  • a merged note with too much information

Useful optional inputs:

  • preferred page count
  • image assets to place
  • desired visual direction
  • whether the output is: carousel pages or one long image

If these are missing, infer the smallest safe set from the source.

Output Contract

Default outputs should be a package with these sections:

  1. Source Rule

- state that the wording is preserved

  1. Cover Copy

- only use original lines or original line fragments unless the user explicitly allows rewriting

  1. Page Plan

- 6 to 10 pages by default - one clear purpose per page

  1. Page Copy

- source text assigned to each page

  1. Layout Brief

- structured instructions for downstream layout

  1. Render Files

- optional HTML/CSS when requested

When the user asks for files, prefer:

  • 01-outline.md
  • 02-titles.md
  • 03-pages.md
  • 04-caption.md
  • 05-layout-brief.md

Read references/output-template.md when composing the package.

Default Workflow

1. Respect the source as immutable by default

Unless the user explicitly asks for rewriting:

  • do not paraphrase
  • do not compress
  • do not add summary lines
  • do not add CTA lines
  • do not add a separate caption

Allowed operations:

  • split into pages
  • promote original lines into headline size
  • move image modules around the source text
  • add neutral page labels

2. Decide the page architecture

Default to 6-10 pages.

Each page should do one job only:

  • cover
  • tension
  • key distinction
  • proof or explanation
  • consequence
  • closing or CTA

Do not use page breaks mechanically every N characters. Split on natural paragraph or sentence boundaries from the source.

3. Write for page readability without rewriting

For each page:

  • keep one main point
  • give it a visible title or emphasis line only by promoting original source lines
  • avoid dense wall-of-text blocks
  • make the first two lines legible without context from the previous page

If a page exceeds reasonable density, split it.

4. Prepare layout-aware copy

Assume the downstream renderer or layout model benefits from structure.

For each page, clearly separate:

  • pageRole
  • headline from source if used
  • subhead if needed
  • body from source
  • emphasisLines from source if used

Then prepare a consolidated layout brief.

Read references/layout-system.md when the user wants the warm-gradient card style from the referenced PDF.

Quality Rules

1. Source-first

Do not invent authority, case studies, metrics, or personal stories not present in the source.

2. No stealth rewriting

Do not silently change:

  • sentence rhythm
  • information density
  • spoken-language phrasing
  • the user's own wording habits

3. Avoid AI-shaped cleanup

Do not force the source into:

  • neat binary formulas
  • over-organized summary language
  • your own preferred phrasing patterns

If the source says it one way, keep that way unless explicitly asked to rewrite.

4. Cover copy must be from the source

The cover should use:

  • original lines
  • or original line fragments

not a fresh headline invented by the model.

Heuristics

Good cover material

  • strong original opening lines
  • strong original judgments
  • strong original rhetorical questions

Weak cover material

  • lines that only work after a lot of setup
  • lines that require new explanation from the model

Common Input Shape

A common use case is:

  • a draft in the active work folder or a nearby project directory
  • especially talking-head scripts or founder/operator viewpoint drafts
  • output needs to feel publishable on Xiaohongshu without losing the original opinion

For example:

  • 口播稿 -> 小红书图文分页
  • 观点长文 -> 封面 + 标题 + 8 页图文
  • 创始人表达 -> 更适合图文阅读的版本

Default File Placement

If the user asks you to write outputs to disk, prefer creating a sibling folder next to the source draft, for example:

<source-dir>/<draft-name>-xhs-package/

Store the package files there using the default filenames from the output contract.

Release-Shell Execution Contract

  • keep page planning notes, source extracts, and intermediate packaging files under <work-folder>/.postplus/xiaohongshu-article-packager/
  • keep only final user-facing page packages outside .postplus/
  • start with a bounded first pass on one draft and one package target before broader packaging
  • if the source is insufficient or under-specified, stop immediately instead of inventing missing material or relying on customer-specific path assumptions

Failure Modes

Stop and say the source is insufficient if:

  • there is no clear argument
  • the draft is only slogans
  • the user expects claims or case studies not in the material

Then offer one of:

  • a lighter layout pass
  • a request for more source material
  • a narrower page selection based on what is actually present

平台分布

Codex

36.59%
按下载量换算40

Claude

27.99%
按下载量换算30

Cursor

19.03%
按下载量换算21

Gemini CLI

8.48%
按下载量换算9

安全审计

Gen Agent Trust Hub

通过

Socket

可疑

Snyk

通过

权限和风险

执行命令

安装流程涉及命令执行,可能通过 npx skills add https://github.com/postplusai/postplus-skills --skill xiaohongshu-article-packager 联网下载 Skill 或依赖。用户安装前应确认命令来源、仓库内容和执行环境。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills