Token导航 LogoToken导航

Agent Skill · 研究检索

cold-open-creator(冷开创造者)

cold-open-creator 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。

cold-open-creator 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 58,594。

总安装 412周安装 17GitHub Stars 4下载 135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:cold-open-creator(冷开创造者)
来源仓库:https://github.com/cdeistopened/opened-vault
仓库路径:skills/cold-open-creator
安装命令:
npx skills add https://github.com/cdeistopened/opened-vault --skill cold-open-creator
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/cdeistopened/opened-vault --skill cold-open-creator

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。

cold-open-creator 制作 25-35 秒的播客开场白,制造叙事张力悬念。基于故事弧线方法,在冲突峰值处截断以引发听众好奇心。避免总结性或预告性质内容,聚焦于‘如何来到这里’的疑问。需输入具体人物目标、障碍与转折事件作为创作素材来源。

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/cold-open-creator

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/cdeistopened/opened-vault --skill cold-open-cr…;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

cold-open-creator 怎么安装?

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

cold-open-creator 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。

cold-open-creator 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。

cold-open-creator 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。

cold-open-creator 用的人多吗?

公开数据是累计安装 412 次,近一周安装 17 次,下载 135 次,GitHub 4 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。

cold-open-creator 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
cold-open-creator

冷开创造者 · 本页

4124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

Podcast Cold Open Creator

Purpose

Create 25-35 second cold opens that drop listeners into a story arc - someone who wanted something, encountered obstacles, and figured something out. Cut at the tension peak, before the resolution.

Core Philosophy: The cold open is not a summary or trailer. It's a story interrupted at maximum tension. Listeners should finish asking "How did we get here?" and "What happens next?"


The Narrative Snippets Method

The best cold opens follow recognizable story beats. Search for complete story arcs, then cut at the right point.

Story Beats

BeatWhat HappensCold Open Role
1. SetupProtagonist pursuing somethingINCLUDE
2. DisasterSomething disrupts status quoINCLUDE
3. Failed ApproachObvious solution backfiresINCLUDE
4. New InsightA realization changes everythingTEASE or CUT HERE
5. ResolutionInsight applied, worksCUT BEFORE
6. ReflectionUniversal takeawayCUT BEFORE

The Rule: Include beats 1-3, maybe hint at 4. Cut BEFORE beats 5-6.

Signal Phrases to Find Story Arcs

BeatSignal Phrases
Setup"Back then I was..." / "I had been trying to..." / "For years I thought..."
Disaster"Then everything fell apart..." / "I was diagnosed with..." / "The call came and they said no..."
Failed Approach"I tried X but..." / "We did everything we were supposed to..." / "Most people just accept it..."
Insight"Then I realized..." / "That's when it clicked..." / "So what we do instead is..." [CUT]

Detailed example: references/story-arc-example.md


Two Approaches

Approach A: Narrative Snippets (Preferred)

Use when episode contains personal stories with clear arcs:

  • Origin stories, failure-to-success journeys
  • Realization/discovery moments
  • Before/after transformations

Approach B: Scene Selection (Fallback)

Use when episode is philosophical/conversational without clear arcs:

  • Expert interviews about concepts
  • Debates or discussions
  • Advice-heavy episodes

Decision rule: Scan for story arcs first. If beats 1-4 are present, use Approach A. Otherwise, use Approach B.


The Fundamental Constraint

You have ONLY TWO tools:

  1. CUTTING - Delete portions
  2. REARRANGING - Change clip order (non-chronological)

You CANNOT: Add words, paraphrase, create dialogue, add narration, or change meaning.


The 5 Descript Elements

1. The Scene

Find the moment that encapsulates big themes. Something specific that gestures toward the general.

2. The Character

Get them in ACTION, not passively described. Start with them doing or revealing something specific.

3. The Stakes

Make clear why this matters: "Her next paycheck depends on it" / "This changed everything about how I parent"

4. The Conflict/Shift

Find friction, surprise, or expectation-violation. Contradiction hooks work well.

5. The Setup/Tease

End by gesturing toward what's to come. Leave a question unanswered.


Workflow: 4 Steps

Step 1: Find Story Arcs

Scan transcript for complete arcs using signal phrases above. If no clear arcs, scan for:

  • Inflection points: "Then everything changed..."
  • Vulnerability: "I failed college twice..."
  • Contradiction: "We thought X, but actually..."
  • Surprising insights: "5,000 studies showed..."

Step 2: Extract Clips

Pull verbatim clips showing: setup, character, conflict, shift, tease. Document speaker, timestamp, quote, duration (~3-8 sec each).

Step 3: Arrange for Arc (Not Chronological)

Rearrange for maximum impact:

  1. Scene/Setup (2-3 sec)
  2. Character (3-5 sec)
  3. Conflict/Shift (4-6 sec)
  4. Stakes (2-3 sec)
  5. Tease (2-3 sec) - CUT mid-thought

Example reorder: Original: Background -> Struggle -> Revelation -> Action Cold open: Revelation -> Struggle -> Background -> Action [cuts before answer]

Step 4: Quality Control (5 Tests)

TestQuestion
StrangerWould someone with zero context be intrigued?
ItchDoes it create unbearable need to know more?
StakesIs it clear why this matters?
TeaseDoes it hint without giving away?
EmotionDoes listener feel something in first 5 seconds?

Pass 4/5 = good cold open.


Technical Rules

Timing

  • Total: 25-35 seconds
  • Per clip: 3-8 seconds (never under 2, never over 10)
  • Max clips: 5-6 (too many = disorienting)

Audio Quality

  • Complete thoughts only (even if cut mid-sentence)
  • Preserve natural speech flow
  • Use speaker changes for rhythm
  • End on open question or incomplete thought

The Cliffhanger (Non-Negotiable)

Every cold open MUST end with an unresolved moment. 100% verbatim.

Types: Unfinished statement / Unfinished question / Shocking statement with no explanation / Promise of revelation


Output

Use template in references/output-template.md


Common Mistakes

  • Scene Selection Trap: Picking "important" moment that doesn't create curiosity. Fix: Ask "Would a stranger care?"
  • Chronological Edit: Arranging in original order. Fix: Deliberately rearrange for drama.
  • Exposition Trap: Explaining who guest is. Fix: Drop listeners in the middle.
  • Over-Explanation: Too many clips or resolving tension. Fix: Cut ruthlessly.
  • Resolution Error: Ending with answer revealed. Fix: End with question.
  • Choppy Cut: Unnatural edits. Fix: Cut at natural pauses.

Key Insight

A cold open is NOT: A summary, movie trailer, explanation, or highlight reel.

A cold open IS: A scene encapsulating themes, creating immediate curiosity, making NOT listening feel like a mistake.

If listeners finish thinking "I need to hear the rest to understand what's happening," you've succeeded.


Related Skills

  • podcast-clip-selector - Short-form clips (TikTok, Reels)
  • youtube-title-creator - Titles that pair with cold opens

平台分布

Codex

36.86%
按下载量换算50

Claude

31.16%
按下载量换算42

Cursor

20.41%
按下载量换算28

Gemini CLI

9.43%
按下载量换算13

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