Token导航 LogoToken导航

Agent Skill · 研究检索

substack-ghostwriting(子堆栈代笔)

用于辅助文档、README、Markdown、说明文和内容稿件的整理与改写。它适合让 Agent 提炼结构、补齐章节、统一术语、检查链接或把零散材料整理成可读文档。使用时应保留项目已有事实、命令和路径,不要把未确认的信息写成确定结论;涉及对外文案时,还需要控制语气,避免过度营销或夸大能力。

substack-ghostwriting 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,权限为只读;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 43,786。

总安装 960周安装 40GitHub Stars 73下载 320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:substack-ghostwriting(子堆栈代笔)
来源仓库:https://github.com/samber/cc-skills
仓库路径:skills/substack-ghostwriting
安装命令:
npx skills add https://github.com/samber/cc-skills --skill substack-ghostwriting
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/samber/cc-skills --skill substack-ghostwriting

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。

substack-ghostwriting 协助 Substack 专栏文章的内容组织与润色。适合长文写作、系列主题规划与读者互动引导。保持作者风格一致性,不擅自更改核心观点立场。涉及第三方引用时应确保授权与版权合规。

适用宿主包括 Codex、Claude、Cursor、Gemini CLI,接入前应确认版本、权限和运行环境要求。

来源仓库内路径:skills/substack-ghostwriting

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/samber/cc-skills --skill substack-ghostwriting;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

    它标注为只读,先用只读任务试一次,确认行为符合预期再扩大使用范围。

常见问题

6 条

substack-ghostwriting 怎么安装?

在终端执行 npx skills add https://github.com/samber/cc-skills --skill substack-ghostwriting。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

substack-ghostwriting 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。

substack-ghostwriting 需要什么权限?

权限级别是只读。该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

substack-ghostwriting 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。

substack-ghostwriting 用的人多吗?

公开数据是累计安装 960 次,近一周安装 40 次,下载 320 次,GitHub 73 星。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。

substack-ghostwriting 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。
  • 标注为只读权限,试用阶段不容易误改本地文件。

先想清楚的地方

  • 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。
  • 只想先试试效果、不希望它改动本地文件的人。

不太适合

  • 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
substack-ghostwriting

子堆栈代笔 · 本页

96073只读未明确有提醒

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

Substack Ghostwriting & Content Optimization

A skill for writing Substack content — both newsletter issues (email-first) and web posts (web-first articles/essays) — that grows subscribers and converts readers. Handles two voice modes (own voice, ghostwriting) and two format modes (newsletter issue, web post).

Core philosophy

Substack is not a blog with an email list. It's a social-media-newsletter hybrid with an algorithm that optimizes for subscriptions, not engagement. This changes everything about how you write, format, and distribute content on the platform.

The algorithm's incentives genuinely align with quality. Substack's revenue comes from subscription cuts (not ads), so gaming engagement metrics doesn't help. What helps: writing content good enough that readers convert to paid subscribers and recommend you to others.

For ghostwriting specifically: the job is capturing someone's existing insights in their voice, not generating insights from scratch. As Nicolas Cole frames it: clients are "insights-rich and time-poor", writers are "time-rich but insights-poor." The art is extraction and voice matching.


Platform formatting constraints

Substack is a social-media-newsletter hybrid with an algorithm that optimizes for subscriptions, not engagement. Revenue comes from subscription cuts (not ads), so quality genuinely wins. For ghostwriting: the job is capturing someone's existing insights in their voice — clients are "insights-rich and time-poor."

Read references/platform-constraints.md for post fields, Notes limits, special content blocks, and media embeds.


Mode detection

Determine two dimensions:

Voice dimension:

  • Own voice — the user writes/publishes under their own name. Go directly to the Writing Workflow.
  • Ghostwriting — writing in someone else's voice, or preparing content for a client. Complete the Ghostwriting Workflow first, then the Writing Workflow.

Format dimension:

  • Newsletter issue (email-first) — sent to subscribers' inboxes. Subject line and email formatting matter most. Read references/email-formatting.md during Phase 3.
  • Web post (web-first) — published as a Substack article/essay, discoverable via search and Substack's feed. SEO and web formatting matter most. Read references/web-post-formatting.md during Phase 3.

If unclear, ask the user. Default to newsletter issue when they say "newsletter" or "issue"; default to web post when they say "article", "essay", "post", or "evergreen content".


Ghostwriting Workflow

When writing for someone else, voice matching comes before content. Read references/voice-matching.md for the full extraction process — it covers sample collection (transcripts > writing > media), voice marker extraction, building a voice guide (10-15 markers with examples), and iteration with the user.

Complete the voice guide and get user validation before proceeding to the Writing Workflow.


Writing Workflow

Phase 1 is mandatory — always ask the user the intake questions and wait for answers before writing anything. If the user already provided some context, extract what you can and ask only about missing pieces.

Phase 0: Voice calibration (own voice mode)

Skip this phase if ghostwriting (the Ghostwriting Workflow handles voice separately).

Ask the user for their existing Substack URL. If they have one, recent posts are a valuable source of tone markers — formality level, sentence rhythm, humor style, paragraph length, how they open and close, recurring phrases. Summarize the voice in 5-7 bullet points and confirm with the user before writing.

If they don't have an existing Substack, ask: "How do you want to sound? Casual and conversational, professional and authoritative, or something else?" Use their answer plus any other writing samples they can share.

Phase 1: Content planning (interview)

Stop and ask. Present the intake questions below to the user and wait for their answers. Do not skip this phase, do not infer silently, and do not start drafting until you have explicit answers or confirmation on every item.

  1. Topic: What's this about? If vague, ask what specific angle or story the reader should walk away with.
  2. Format: Newsletter issue (email-first) or web post (web-first)? See mode detection above.
  3. Audience: Who reads this? (developers, founders, marketers, general tech, niche community...) A newsletter for junior devs reads very differently than one for CTOs.
  4. Objective: What's the concrete goal?

- Grow subscribers (free or paid)? - Drive signups/traffic to an external product (SaaS, course, tool)? - Establish authority / thought leadership? - Nurture existing subscribers toward a paid tier? - Something else? The objective shapes the CTA, the hook angle, and where depth goes vs where the paywall or link sits.

  1. Context: Part of a series? What have recent posts covered?
  2. Length: Short (500-800 words), Standard (1000-1500), Deep dive (2000+)

If critical pieces are missing (especially topic, audience, objective, or format), ask and wait — don't guess. A wrong assumption wastes an entire draft.

If the user has Notes data (which Notes got engagement), use that to validate topic selection. Notes function as a cheap testing pipeline for long-form content.

Phase 2: Title and hook selection

Generate 5 title/subject line variants and 3 hook options (opening 2-3 sentences each). Present them together and ask the user to pick or remix before proceeding. Do not write the body until the user has validated a title and hook direction.

Title principles:

  • Specificity beats vagueness
  • Promise a clear benefit or reveal
  • 6-10 words (readable on mobile and in search results)
  • For dev audiences: technical keywords filter for the right audience; "How to" and numbers perform well; avoid urgency/scarcity tactics

Hook types — write 3 distinct hooks using different strategies (e.g. credibility, counter-narrative, curiosity, surprise, data). Each hook should be 2-3 sentences that could open the piece. Present them labeled (Hook A, Hook B, Hook C) with a brief note on the strategy used.

Newsletter issue — subject line + preview text:

  • The subject line is the headline. The preview text (first ~90 chars of the email) is the subhead. Together they determine open rate.
  • Preview text should complement, not repeat, the subject line.

Web post — SEO + discoverability:

  • Keep the main title punchy for the feed. Use the separate SEO title field for a keyword-rich version (under 60 chars).
  • Write a dedicated SEO description (150-160 chars) — don't rely on the subtitle fallback, it's usually too short.
  • Suggest a URL slug: short (3-6 words), keyword-rich, no dates.
  • Assign to a publication section if applicable.
  • Read references/web-post-formatting.md for detailed SEO guidance.

Wait for the user to choose a title and hook before moving to Phase 3.

Phase 3: Write the content

Using the chosen title and hook, write the full piece. The hook opens the article, then continue with:

  1. Hook (chosen from Phase 2)
  2. Context (1-2 paragraphs): Why this matters now. What prompted this.
  3. Body (bulk): The actual content. Structure depends on content type.
  4. Takeaway (1-2 sentences): The one thing the reader should remember.
  5. CTA (1-2 sentences): Ask for a specific action. Questions that invite replies are strongest (replies are an algorithm signal).

Newsletter issue formatting — read references/email-formatting.md for full rules:

  • Paragraphs: 2-3 sentences max (email clients make long paragraphs feel like walls)
  • Code blocks: < 10 lines (link to Gist for longer code)
  • Images: sparingly (many email clients block them by default)
  • TL;DR at top for issues > 1500 words

Web post formatting — read references/web-post-formatting.md for full rules:

  • Paragraphs: 3-4 sentences acceptable (full-width web rendering is more forgiving)
  • Longer code blocks OK (up to 30-40 lines with full syntax highlighting)
  • Images and embeds render reliably — use more liberally
  • Table of contents for posts > 2000 words

Shared formatting rules:

  • Subheadings every 200-400 words. Bold key phrases so skimmers catch the argument.
  • Descriptive anchor text on links, not "click here."

Phase 4: Growth-optimized elements

Add elements that leverage the Substack algorithm. Read references/substack-algorithm.md for the full mechanics.

  1. Reply prompt: End with a genuine question. Replies signal engagement to the algorithm.
  2. Share prompt: "If you found this useful, forward it to a colleague who [specific situation]." Specificity increases share rate.
  3. Recommendation hook: If the user has recommendation partnerships, weave in natural cross-references.
  4. Notes teaser: Write a 2-3 sentence version for Substack Notes. Notes should stand alone as valuable, not just be a link to the issue.

Phase 5: Paid vs. free strategy

If the user has a paid tier, advise on the free/paid split:

  • Free issues should be your best work (they drive growth and recommendations)
  • Paid issues should offer depth, exclusivity, or access (not just a paywall on normal content)
  • The most effective model: free issues that demonstrate value so clearly that readers want more

Common mistake: paywalling too early. At < 1000 subscribers, everything should be free. Growth compounds faster than paid conversion at small scale.

Phase 5b: Humanize

Invoke a humanizer skill (e.g. "humanize", "humanizer", "de-slop", "natural writing check", "AI detection cleanup", "rewrite like a human") to strip AI-generated patterns — filler words, predictable cadence, over-hedging, hollow transitions, inflated language. Substack readers pay for authentic voice; AI-sounding prose kills trust and cancels subscriptions.

Preserve hooks and subject lines. The hook and title/subject line (Phase 2) were deliberately engineered for open rate and curiosity. Instruct the humanizer to leave them intact — rewriting them for "naturalness" destroys the copywriting structure that drives opens and first-scroll retention.

Phase 6: Image suggestions

After the content is drafted, suggest 1-3 images with specific placement. For each image, provide:

  • Placement: Where in the issue (e.g. "as the cover image", "after the hook", "between section X and Y")
  • Purpose: What the image adds (set the tone, break up a long section, illustrate a concept, reinforce the emotional beat)
  • Description: What the image should depict

For newsletter issues: use images sparingly — many email clients block them by default. Prioritize a strong cover image and at most 1-2 inline images. For web posts: images render reliably — use more freely (diagrams, charts, screenshots).

Offer to generate a Midjourney prompt for each suggested image. If the user accepts, use the latest Midjourney model conventions to write the prompt. Use --ar 16:9 or --ar 3:1 for cover images and wide illustrations (optimal for Substack headers and social sharing), --ar 3:2 for smaller inline images. Refer to up-to-date Midjourney documentation for current prompt syntax and parameters.

Phase 7: Social distribution posts (optional — offer, don't auto-generate)

After the content is written, ask the user if they want social distribution posts. Do not generate them automatically. If accepted, write a LinkedIn post and/or a Twitter/X post to promote it. These are not summaries — they are standalone pieces of content that create enough curiosity or value to drive clicks.

Read references/social-distribution.md for LinkedIn and Twitter/X post templates.


Output format

Newsletter issue:

  • Subject line (chosen) + 2 alternatives
  • Preview text
  • Full issue in markdown, formatted for email readability
  • Image suggestions with placement notes (and Midjourney prompts if accepted)
  • A Notes teaser (2-3 sentences)
  • LinkedIn + Twitter/X distribution posts (only if the user accepted)
  • If ghostwriting: the voice guide used

Web post:

  • Title (chosen) + 2 alternatives
  • Subtitle / meta description
  • Suggested URL slug
  • Full post in markdown, formatted for web readability
  • Image suggestions with placement notes (and Midjourney prompts if accepted)
  • A Notes teaser (2-3 sentences)
  • LinkedIn + Twitter/X distribution posts (only if the user accepted)
  • If ghostwriting: the voice guide used

Adapting existing content

When the user wants to convert a blog post, talk, or thread into Substack content:

  1. Choose the right format. Evergreen source material (reference, tutorial, deep dive) → web post. Timely source (commentary, announcement, reaction) → newsletter issue.
  2. Don't just copy-paste. Substack readers expect a different voice and format than blog readers.
  3. Add the personal layer. Substack is more personal than a blog. Add context: why you're sharing this, what prompted it, your take.
  4. Front-load the value. Blog posts can have a slow build. Substack content must hook in the first 2 sentences (the email preview or search snippet).
  5. Shorten for newsletters. Cut 30-50% of blog post length for newsletter issues — email tolerance for length is lower. Web posts can preserve more depth.
  6. Add the CTA. Blog posts can end quietly. Substack content should ask for something.

Reference files

Read these for deeper platform knowledge:

  • references/voice-matching.md -- Detailed ghostwriting voice extraction process, interview techniques, voice guide templates, and iteration workflow. Read when ghostwriting.
  • references/email-formatting.md -- Email client rendering constraints, formatting rules, mobile optimization, and code block handling. Read during Phase 3 for newsletter issues.
  • references/web-post-formatting.md -- SEO optimization, web-first formatting rules, evergreen vs timely content strategy, sections/categories, and rich media usage. Read during Phase 3 for web posts.
  • references/substack-algorithm.md -- How the algorithm works (from Substack's ML lead), Notes ranking signals, Recommendations system, growth levers ranked by impact, and monetization math. Read during Phase 4 or for strategic planning.

平台分布

Codex

33.99%
按下载量换算109

Claude

29.87%
按下载量换算96

Cursor

18.03%
按下载量换算58

Gemini CLI

9.46%
按下载量换算30

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

只读

该 Skill 主要提供规则、说明或参考内容,本身偏只读;真正读写文件、联网或执行命令仍取决于宿主 Agent 的任务。

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