用AI构建自己的内容创作系统Skill 创建

SKILL内容创作
内容创作 未标注权限 许可证不明确
安装这个技能

将播客、音频、视频回放、直播回放、访谈录音、活动回放、圆桌逐字稿或口播笔记转化为公众号文章。Use when the user asks to 把播客转公众号, 音频转文章, 视频转公众号, 直播回放转文章, 逐字稿改成推文, 音视频内容整理成长文, 从录音/视频/访谈/圆桌提炼公众号稿, or turn spoken audio/video content into a publishable WeChat article while preserving field texture, speaker boundaries, public consent, and the creator's own voice.

一句话选型结论

按资料字段整理

用AI构建自己的内容创作系统Skill 创建是内容创作类的 Agent Skill,安装方式未标注,权限级别未标注,许可证不明确,有可直接复制的安装命令。技能本质是给模型读的说明书,效果取决于模型与客户端;装完先用一个小任务验证,再放进正式流程。

分类内容创作 安装方式未标注 权限级别未标注权限 许可证许可证不明确
累计安装 未标注 本周 未标注
GitHub 星标 未标注 仓库 未标注
下载量 未标注 技能市场下载数
权限级别 未标注权限 安装前请确认风险范围
许可证 许可证不明确 未说明

安装

未标注
在支持 Skills 的客户端里执行,装完问一句「你有哪些技能」验证
skillhub install user_65a4a566/song1
  1. 1
    确认客户端支持

    确认你的客户端支持 Skills(Claude、Codex、Cursor 等),并升级到较新的版本。

  2. 2
    拿到技能文件

    执行上面的命令安装;也可以把仓库里的 SKILL.md 手动放进客户端的 skills 目录。

  3. 3
    刷新技能列表

    重新打开客户端,或刷新技能列表,让新放进来的技能被加载。

  4. 4
    验证是否生效

    问一句「你有哪些技能」,确认列表里出现它,再拿一个小任务试跑。

技能档案

内容创作
字段来自技能仓库与公开索引,中文说明由本站标注
作者 / 维护方user_65a4a566仓库 未标注 · 版本 未标注
宿主客户端未标注技能可被这些客户端加载
安装方式未标注安装包 community
权限级别未标注权限安装前请确认风险范围
许可证未说明许可证不明确,商用前请自行确认条款
仓库可用性available技能仓库当前是否可访问
最近核验未标注上游最后核验时间 · 数据更新 1791674316
正文体量10.0 KBSKILL.md 原文大小,越大说明说明书写得越细

典型使用场景

3 类
按分类整理的常见用法,具体以 SKILL.md 为准
写文章与脚本

按主题产出初稿与分镜脚本。

改稿与润色

把草稿改顺、统一语气。

多平台文案改写

同一内容按平台改成长短不同的版本。

选择建议

装之前先看这几条
按条目字段给出的判断,不替你决定用不用

值得选的地方

  • 有可直接复制的安装命令,不用自己翻文档找入口。

先想清楚的地方

  • 没有标注权限级别,实际能做什么要装好后确认。
  • 许可证不明确,商用或二次分发前先到仓库确认条款。
  • 没有做站内核验,内容以作者仓库为准。
  • 技能只是说明书,实际效果取决于模型与客户端,装完先用小任务试一遍。

更适合

  • 需要在内容创作场景里给模型补一套做法的人。
  • 愿意先读一遍 SKILL.md、确认它会让模型做什么的人。

不太适合

  • 对许可证要求很严、必须明确可商用的人。
  • 只想装完就自动变强、不打算看说明书的人。

同类对比

6 个同类
同分类里挑最接近的几条,字段以各自详情页为准
技能权限级别安装方式累计安装宿主客户端安装命令
用AI构建自己的内容创作系统Skill 创建当前 未标注权限 未标注 未标注 未标注 有
AI 视频生成 调用外部服务 GitHub 仓库 431.3 万 Codex、Claude、Cursor 有
视频帧 联网访问 ClawHub 市场 262.1 万 OpenClaw 有
Markdown 转换器 写入文件 ClawHub 市场 135.3 万 OpenClaw 有
哔哩哔哩全合一 接触敏感数据 ClawHub 市场 64.1 万 OpenClaw 有
Remotion video toolkit 命令行 浏览器控制 ClawHub 市场 63.9 万 OpenClaw 有

SKILL.md 全文

10.0 KB
模型实际读到的说明书原文,来自技能仓库

Audio To WeChat Article

Use this skill to turn spoken audio/video material into a publishable WeChat article draft. The goal is not transcription cleanup or video editing; it is editorial transformation: find the strongest public argument, preserve live texture, protect speaker boundaries, and decide whether the material deserves a long article.

Treat "AI味", "不像发布者", "太完整", "太顺", or "太平台" as hard revision signals. Do not publish abstract values naked; let judgments grow from a small narrative, field detail, showcase, or lived practice.

Use these base content rules: scene before concept, evidence before fluency, dialogue as a first-class output, human-AI boundaries, platform fit, and asset recovery.

Not every spoken source should become a deep article. When the value lives in the exchange itself, choose an edited dialogue, conversation digest, or Q&A-style post instead of forcing a thesis-led WeChat essay.

When the user wants to record the material for 视频号, produce a mouth-friendly, scene-led, recordable script instead of only a written article.

When the user wants an easier way to view outputs, return a compact readiness card or platform decision table before drafting the full article.

Core Stance

Treat podcast, livestream, recording, and video replay content as source material, not as an article draft.

  • Do not summarize every topic in order.
  • Do not force all audio/video into a long article. Some material should become a short post, quote card, topic seed, story card, or private note.
  • Do not force all interviews into the publisher's synthesized argument. Some should remain visibly dialogic.
  • Preserve spoken energy, but remove filler, repetition, hosting logistics, digressions, and timestamp debris.
  • Keep the central editorial judgment visible. A WeChat article needs one reason to exist.
  • Protect guests, listeners, private examples, and commercially sensitive details.

Workflow

  1. Build a source map:
    - 来源: podcast, audio recording, video replay, livestream, interview, event replay, transcript, notes, or mixed material
    - 时间/场景: when and where the conversation happened
    - 人物: speakers, roles, relationship, naming/anonymization needs
    - 可公开边界: public, semi-public, private, needs consent, or unknown
    - 素材状态: raw transcript, cleaned transcript, notes only, partial audio/video clips, or user memory
  2. Diagnose whether it should become a WeChat article:
    - 适合长文: has a clear claim, tension, story arc, strong scenes, useful method, or public timing
    - 适合对谈稿: the back-and-forth, guest wording, host questions, or thinking process is the main value
    - 适合对谈摘要: the source is long but contains several publishable exchanges around clear themes
    - 适合视频号口播: has one concrete detail, question, or exchange that the publisher can speak to camera
    - 适合短文: has one sharp idea but not enough evidence for a long piece
    - 适合素材卡: useful quotes/scenes but no article center yet
    - 需补问: missing facts, speaker consent, examples, chronology, or the creator's judgment
    - 暂不外发: too private, too sensitive, or relationship risk is high
  3. Choose one article type:
    - 观点观察文: one central judgment from the conversation, supported by quotes and scenes
    - 人物/故事文: one person or group arc, with turning points and concrete actions
    - 活动/直播复盘文: what happened, what it revealed, and what should happen next
    - 方法提炼文: turn repeated practice into a framework, checklist, or working method
    - 对谈梳理文: curated dialogue around themes, preserving speaker turns and thinking process
    - 对话摘录文: lightly edited Q&A or excerpt, used when the speakers' wording is the value
    - 视频号口播稿: spoken script with cover text, subtitles, recording posture, and post caption
  4. Extract article material:
    - Central claim
    - 3-5 support points
    - Scenes worth showing
    - Quotes worth keeping
    - Reader tension or problem
    - Boundary risks
    - Missing follow-up questions
  5. Build the article skeleton before drafting:
    - Title candidates
    - One-sentence thesis
    - Opening hook: scene, tension, or sharp question
    - Section arc
    - Quote placement
    - Ending: reflection, invitation, CTA, or next action
  6. Draft the article:
    - Rewrite oral fragments into readable paragraphs.
    - Keep key quotes short and attributed when allowed.
    - Make transitions explain why each section follows.
    - Let concrete moments carry meaning before abstract conclusions.
    - Add a light editor's note only when context is necessary.
  7. Run a compact final check:
    - Logic: What is the article really saying?
    - Evidence: Which claims are supported by the audio?
    - Voice: Which lines sound generic or over-polished?
    - AI味: Which parts sound too smooth, too complete, too framework-like, or unlike the publisher?
    - Boundary: What needs consent, masking, or removal?
    - Form: Should this remain a dialogue instead of becoming a synthesized article?
    - Asset: What can this article become later: case, method, offer signal, source card, or portfolio proof?

WeChat Article Defaults

Use these defaults unless the user requests another style.

  • Length: 1800-3500 Chinese characters for a standard article; 800-1500 for a light recap.
  • Structure: opening, 3-5 titled sections, closing note.
  • Tone: concrete, reflective, field-based, not marketing-heavy.
  • Voice: small narrative before big trend; real scene before concept; do not end every section with a neat takeaway.
  • Title: generate 5-8 options across 直接判断, 场景张力, 问题意识, and 栏目感.
  • Subtitle/摘要: one short line explaining why this piece matters now.
  • CTA: soft and context-specific, such as asking for stories, inviting replies, linking to the next event, or naming a collaboration boundary.

Spoken-To-Written Rules

When transforming oral material:

  • Delete filler: "然后", "其实", "就是", "对吧", repeated self-correction, platform reminders, greeting loops, and logistics.
  • Merge repeated points instead of preserving chronology.
  • Keep productive hesitation when it reveals uncertainty, relationship texture, or thought-in-progress.
  • Convert long monologues into paragraphs with one job each.
  • Use direct quotes only for lines with voice, evidence, or emotional weight.
  • Do not fabricate examples, numbers, chronology, speaker intent, or audience reaction.
  • Do not write "这期播客我们聊了很多" as an opening unless the meta-context is the point.

Output Modes

Choose the smallest useful output.

Article Readiness

Use when the source is raw, long, or uncertain.

## 是否值得写成长公众号

结论:

## 最强文章方向
| 方向 | 类型 | 为什么成立 | 还缺什么 |
| --- | --- | --- | --- |

## 可用素材
| 素材 | 用途 | 风险 |
| --- | --- | --- |

## 不建议写成

## 需要创作者补的 3 个判断

Compact View

Use when the user wants a lighter result.

## 公众号查看卡

- 是否值得写成长文:
- 更适合的形式: 长文 / 对谈 / 短观察 / 视频号 / 小红书 / 素材卡
- 最好的入口:
- 需要保留:
- 需要删弱:
- 风险:
- 下一步:

Article Brief

Use before drafting, or when the user wants a writing plan.

## 文章主张

## 读者为什么现在需要读

## 标题候选

## 结构

## 必保留的场景/原话

## 需要删弱的口语内容

## 公开边界

## CTA

WeChat Draft

Use when the user asks for the actual article.

## 标题候选

## 摘要

## 正文

## 备选开头

## 发布前确认

Edited Dialogue

Use when the source should remain a conversation.

## 是否适合对谈形式

结论:

## 标题候选

## 编辑说明

- 对谈发生在:
- 发言人:
- 编辑原则:
- 公开边界:

## 对谈正文

### 1. 主题 / 问题

主持人:

嘉宾:

主持人:

嘉宾:

### 2. 主题 / 问题

...

## 发布者短后记

## 还没聊完的问题

## 发布前确认

Material Recovery

Use when the material should not become a full article yet.

## 不直接写成长文的原因

## 可回收素材卡
| 卡片 | 类型 | 可转化用途 | 待补 |
| --- | --- | --- | --- |

## 更适合的发布形态

## 下一次追问/补采

Optional Handoff

If the user's environment has adjacent writing or knowledge-management skills, use them only when useful. This skill must still work as a standalone workflow.

  • Use a material-intake or note-card workflow for long raw transcripts that first need reusable cards.
  • Use a logic/evidence review workflow when a draft's claims need support checking.
  • Use a voice-editing workflow before final polishing.
  • Use a platform adaptation workflow after the WeChat article is stable and needs variants for Xiaohongshu, newsletter, posts, or show notes.

Guardrails

  • Do not expose private conversations, client details, financial terms, identifiable audience stories, or off-record comments without permission.
  • Do not convert a guest's words into the creator's stance unless the distinction is clear.
  • Do not make the article sound more certain, dramatic, or strategic than the source supports.
  • Do not smooth away contradiction if the contradiction is the real point.
  • If the transcript is too messy or too thin, return an article brief plus follow-up questions instead of pretending to finish.

常见问题

5 问
答案由本站依据条目字段整理
用AI构建自己的内容创作系统Skill 创建怎么安装?

在支持 Skills 的客户端里执行:skillhub install user_65a4a566/song1。安装方式为未标注。装完刷新技能列表或重开客户端。

用AI构建自己的内容创作系统Skill 创建支持哪些客户端?

资料没有标注宿主客户端,但只要客户端支持 Skills(Claude、Codex、Cursor 等),一般都能加载。

用AI构建自己的内容创作系统Skill 创建权限高吗,安全吗?

权限级别为未标注权限。安装前先看它会读取或写入哪些内容。

用AI构建自己的内容创作系统Skill 创建收费吗?

本站不代收费用,也没有标注具体价格。是否收费取决于技能本身和所用的模型或客户端,正式使用前按作者说明确认。

用AI构建自己的内容创作系统Skill 创建用什么许可证?

许可证标注为 未说明(许可证不明确)。商用或二次分发前请到仓库确认条款原文。

热门专题

按场景挑技能
全部专题 →

技能内容以作者仓库为准;分类、权限级别与许可证状态的中文说明为中文整理,最近核验 未标注。