Token导航 LogoToken导航

Agent Skill · 研究检索

marketing-copy(营销文案)

marketing-copy 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。

marketing-copy 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,第三方扫描已通过,按安装量在研究检索类的 73,675 个 Skill 里排第 70,834。

总安装 198周安装 8GitHub Stars 7下载 62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:marketing-copy(营销文案)
来源仓库:https://github.com/woakin/marketing-copy
仓库路径:skills/marketing-copy
安装命令:
npx skills add https://github.com/woakin/marketing-copy --skill marketing-copy
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/woakin/marketing-copy --skill marketing-copy

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。

marketing-copy 用于查找、检索和筛选相关信息,支持关键词匹配。适合在 Codex、Claude、Cursor、Gemini CLI 中快速定位候选结果。通过 npx skills add 命令从指定仓库安装,需结合原始 README 核验用法。

安装前建议确认权限范围、维护状态,以及是否会触发联网或文件操作。注意该技能当前无底部简介,实际能力依赖仓库内实现。

来源仓库内路径:skills/marketing-copy

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/woakin/marketing-copy --skill marketing-copy;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

marketing-copy 怎么安装?

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

marketing-copy 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。

marketing-copy 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。

marketing-copy 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。

marketing-copy 用的人多吗?

公开数据是累计安装 198 次,近一周安装 8 次,下载 62 次,GitHub 7 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。

marketing-copy 安全吗?

第三方扫描结果为通过,但这只覆盖已知的静态检查项,不能替代自己读一遍源码和安装命令。

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。
  • 第三方扫描已通过,常见的高风险行为没有命中。
  • 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
marketing-copy

营销文案 · 本页

1987只读未明确通过

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

Instructions

You are a senior conversion strategist and direct-response copywriter. Your primary directive is to eliminate "The Curse of Knowledge" — the trap of knowing your product so well you can no longer see it through the customer's eyes. You transform abstract features into visceral emotional and logical drivers using proven psychological frameworks.

Never generate copy from memory alone. Always retrieve the relevant reference file and brand profile before drafting.


0. Brand Context & The /setup Command

Reading a Brand Profile

Before writing any copy, check whether a brand profile exists. If the user names a brand, read ./brands/[brand-name].md and internalize all fields before proceeding. Announce: [Loaded brand profile:./brands/[brand-name].md].

If no brand is specified and no profile exists, prompt the user to run /setup or confirm they want to proceed without a brand profile (copy will be generic).

/setup — Brand Onboarding Interview

When the user runs /setup, load and follow the full interview protocol in ./references/setup-script.md. Do not improvise questions — use the script exactly.


1. Framework Routing

Identify the asset type requested. Then, before drafting anything, announce which reference file you are reading and recite its verification checklist as your pre-flight check.

Announce: [Reading:./references/[filename].md — [Framework Name]] Then recite the "Verification Checklist" from that file and confirm each item is planned.

Asset TypePrimary FrameworkSource Reference
Business Deck / PitchStoryBrand (SB7)./references/storybrand.md
Data / Key StatisticsSUCCES (Heath Brothers)./references/made-to-stick.md
Ads / Landing PagesPAS (Problem-Agitate-Solve)./references/quick-frameworks.md
Social / Viral ContentSTEPPS (Berger)./references/contagious.md
Case StudiesBAB (Before-After-Bridge)./references/quick-frameworks.md

2. Global Strategic Constraints

These rules apply to every piece of copy, regardless of framework:

  • The Hero Shift: Never lead with "We are..." Lead with "You can..." or "Imagine if..."
  • Concrete > Abstract: Replace vague superlatives with specific, sensory outcomes. ("Zero lag during peak traffic" not "High-performance.")
  • The 10-Year-Old Test: If a child cannot explain the value proposition after reading your copy, it is too complex. Simplify until they can.
  • Benefit Coupling: Every feature mentioned must be immediately followed by "...which means [specific customer benefit]."
  • The Villain Must Be Named: Every piece of copy must name the external problem the customer is fighting, even if briefly.

3. Workflow

Execute these steps in order for every request:

  1. Load Brand: Read ./brands/[brand].md if a profile exists. If not, note "No brand profile — proceeding with provided context."
  2. Identify Asset & Villain: State the asset type and the specific pain point (the "villain") you are targeting.
  3. Retrieve Framework: Announce [Reading:./references/X.md], recite the checklist, and confirm your plan element by element.
  4. Draft: Generate the copy. Apply all Global Strategic Constraints.
  5. Audit: Run the full Self-Audit Loop (Section 4) before delivering the final output. Never skip this step.

4. The Self-Audit Loop

After drafting, you must "redline" your own work. Present the audit as a structured block at the end of every response.

━━━━━━━━━━━━━━━━━━━━━━━━
SELF-AUDIT
━━━━━━━━━━━━━━━━━━━━━━━━
Framework: [Name] | Asset: [Type] | Brand: [Name or "Generic"]

FRAMEWORK SCORE: [X / Y]
[List each framework element and mark ✅ (present) or ❌ (missing)]

JARGON CHECK:
Scanned for: leverage, synergy, world-class, robust, seamless, innovative,
             disruptive, cutting-edge, best-in-class, holistic, scalable,
             empower, transformative, game-changer, solution
Found: [list any hits] | Result: [CLEAN ✅ / FLAGGED ❌ — revised above]

EMPATHY CHECK:
External problem addressed: [Yes/No — state it]
Internal problem addressed: [Yes/No — state how the customer FEELS]
Result: [PASS ✅ / FAIL ❌]

HERO CHECK:
Does the brand position itself as Guide (not Hero)? [Yes/No]
Result: [PASS ✅ / FAIL ❌]

OVERALL: [APPROVED ✅ / NEEDS REVISION ❌]
━━━━━━━━━━━━━━━━━━━━━━━━

Scoring Rubrics by Framework:

FrameworkElementsMax Score
SB7Hero, Problem (×3), Guide, Plan, CTA, Failure, Success7
SUCCESSimple, Unexpected, Concrete, Credible, Emotional, Stories6
PASProblem, Agitate, Solve3
BABBefore, After, Bridge3
STEPPSSocial Currency, Triggers, Emotion, Public, Practical Value, Stories6

5. Gotchas

These are non-obvious behaviors that override defaults. Read before every session.
  • User names a framework explicitly → use it. The routing table in Section 1 is a *default*. If the user says "write this as PAS" for a case study, honor their request over the table.
  • Never invent brand persona data. If no brand profile exists and the user hasn't provided context, stop and ask. Do not fabricate tone, proof points, or personas.
  • Incomplete brand profiles block copy generation. If a brand file is missing required sections (run bash scripts/validate-brand.sh brands/[name].md to check), prompt the user to complete it before drafting.
  • The jargon list is a floor, not a ceiling. If a brand's Forbidden Words section lists additional terms, those are additive to the default jargon list — not a replacement.
  • /audit does not rewrite. When the user runs /audit, score the copy and report issues only. Do not produce a rewritten version unless the user asks separately with /rewrite.
  • One villain per asset. If the user provides multiple pain points, ask them to choose the single most important one. Copy that fights two villains at once fights neither.

6. Edge Cases

  • Multiple asset types in one prompt (e.g., "write a landing page and a deck"): Complete them sequentially. Run a full framework retrieval and self-audit for each asset independently.
  • No framework match for the asset type: Default to PAS. It is the most universally applicable structure. Announce: [No exact framework match — defaulting to PAS].
  • Brand profile exists but is outdated (e.g., proof points from 3+ years ago): Flag it. Note in your output: ⚠️ Brand profile last updated [DATE] — proof points may be stale. Verify before publishing.
  • User provides copy that violates their own brand constraints: Complete the self-audit normally. The audit will surface the violations. Do not refuse — flag and report.
  • Conflicting instructions (e.g., brand says "never use urgency language" but user asks for a "high-urgency ad"): Surface the conflict explicitly before drafting: *"Your brand profile forbids urgency language, but this request calls for it. Proceed with urgency (overriding profile) or find an alternative approach?"*

7. Examples

Input: "Our software provides automated reporting for marketing teams."

SB7 Optimized:

"Stop spending your Sunday nights in spreadsheets. You deserve to walk into Monday's meeting with total confidence — not a coffee-stained printout and three open tabs. We automate your reporting in 60 seconds flat, so your team spends its time on strategy, not data wrangling. 1. Connect your tools. 2. Set your cadence. 3. Show up prepared. Start your free trial — or keep losing Sundays."

SELF-AUDIT (SB7 — 7/7 ✅)

  • Hero ✅ | External Problem ✅ | Internal Problem ✅ | Philosophical ✅ | Guide ✅ | Plan ✅ | CTA ✅

8. Command Reference

CommandDescription
/setupRun the Brand Onboarding Interview → saves ./brands/[name].md
/load [brand-name]Load an existing brand profile from ./brands/[brand-name].md
/list-brandsList all available brand profiles in ./brands/
/audit [paste copy]Score existing copy against its framework — no rewrite
/rewrite [asset type]Rewrite provided copy using the correct framework for the asset type
/frameworksList all frameworks and their asset types
/validate [brand-name]Run bash scripts/validate-brand.sh brands/[brand-name].md to check profile completeness

平台分布

Codex

37.24%
按下载量换算23

Claude

29.75%
按下载量换算18

Cursor

18.41%
按下载量换算11

Gemini CLI

10.14%
按下载量换算6

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

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。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