Token导航 LogoToken导航

Agent Skill · 研究检索

hook-methodology(钩子方法)

hook-methodology 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。

hook-methodology 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,权限为需要联网;许可证为 MIT,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 50,054。

总安装 630周安装 26GitHub Stars 93下载 206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
MIT

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:hook-methodology(钩子方法)
来源仓库:https://github.com/dv0x/creative-ad-agent
仓库路径:skills/hook-methodology
安装命令:
npx skills add https://github.com/dv0x/creative-ad-agent --skill hook-methodology
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。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/dv0x/creative-ad-agent --skill hook-methodology

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。

hook-methodology 用于查找、检索和筛选相关信息,支持关键词驱动的内容定位。适用于需要快速获取候选结果的研究型工作流场景。通过 npx 命令从 creative-ad-agent 仓库安装使用。安装前需确认是否涉及联网权限及外部命令执行能力。建议结合原始文档了解具体检索范围和结果过滤机制。

来源仓库内路径:skills/hook-methodology

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/dv0x/creative-ad-agent --skill hook-methodolog…;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

    它需要需要联网,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

hook-methodology 怎么安装?

在终端执行 npx skills add https://github.com/dv0x/creative-ad-agent --skill hook-methodology。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

hook-methodology 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。

hook-methodology 需要什么权限?

权限级别是需要联网。该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

hook-methodology 是免费的吗?

来源标注的许可证是 MIT。本站不代收任何费用,也不托管安装包;能否商用、要不要署名,以许可证原文为准。

hook-methodology 用的人多吗?

公开数据是累计安装 630 次,近一周安装 26 次,下载 206 次,GitHub 93 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。

hook-methodology 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。
  • 3 个目录同时收录,说明这个 Skill 流传比较广。
  • 许可证明确为 MIT,商用和分发边界好判断。
  • 第三方扫描已通过,常见的高风险行为没有命中。

先想清楚的地方

  • 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。
  • 需要许可证清楚、并且有第三方扫描结果的团队项目。

不太适合

  • 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
hook-methodology

钩子方法 · 本页

63093需要联网MIT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

Research-First Hook Methodology

Every hook must be traceable to a specific research element. No invention—only discovery.

Core Philosophy

Research first. Every hook has a SOURCE.

The best hooks aren't constructed from templates.
They're extracted from what the brand has already proven, said, or delivered.

If you can't point to where it came from in the research,
it's invention. Invention is weak. Discovery is strong.

The 10 Hook Types

Before extracting, know what you're looking for:

Attention Hooks (Stop the Scroll)

TypePurposeExample Pattern
QuestionChallenge assumption"What's the biggest mistake you're making?"
Surprising StatUnexpected number"90% of banks reject self-employed applicants"
Pattern InterruptBreak expectationsUnexpected statement that contradicts normal
ControversialChallenge beliefQuestion something everyone accepts
Direct AddressCall out situation"Self-employed? Bank said no?"

Desire Hooks (Drive Clicks)

TypePurposeExample Pattern
Social ProofCredibility through numbers"16,000+ Canadians switched and saved"
Problem-SolutionName pain, offer reliefPain statement → Solution hint
ContrastShow gap"[Bad thing] vs. [Better thing]"
FOMO/UrgencyScarcity, pressureLimited time, competitive pressure
CuriosityTease hidden info"What your bank isn't telling you"

Step 1: EXTRACT (Read EVERY Section)

Read agent/files/research/{brand}_research.md

You MUST extract from EVERY section. No exceptions.


FROM: The Offer

Extract:

  • Core promise (what they sell)
  • Key numbers (rates, coverage, speed, capacity)
Example:
"AI-powered mortgage brokerage, 350+ lenders, rates from 4.19%, 24-48 hour approvals"

FROM: Key Value Props

Extract:

  • Each differentiator as a potential hook angle
  • What makes them different from competitors
Example:
- AI-Powered Intelligence
- Rapid Approval (24-48 hours)
- 350+ institutional + 150+ private lenders
- 24+ specialized loan products

FROM: Proof Points

Extract:

  • Specific stats (ratings, reviews, credentials)
  • Real deal examples with numbers
  • Certifications, licenses, awards
Example:
- 4.8/5 rating, 856 reviews
- FSRA Licensed #M20003023
- $547.5K purchase closed in 24 hours (Oshawa)
- $2M construction deal (Vaughan)

FROM: Products/Services ← CRITICAL

Every product is a potential hook for a specific audience.

Extract:

  • Every product category
  • Unique or surprising offerings
  • Niche products that solve specific problems
Example:
RESIDENTIAL:
- Bank Decline Financing → hook for rejected borrowers
- 120% LTV Owner-Occupied → surprising high leverage
- Non-Resident Mortgages → new Canadians hook
- Rush Closing 24-48 hours → urgency hook

SPECIALIZED:
- BRRR Strategy Loans (72-hour funding) → investor hook
- Fix & Flip Financing → investor hook
- Power-of-Sale Financing → rescue/urgent hook
- Bridge Loans → timing gap hook

COMMERCIAL:
- No Personal Guarantee (up to $1M) → risk-reduction hook
- CMHC Multi-Family (95% LTV) → leverage hook

FROM: Pain Points Addressed

Extract:

  • Each pain point + the brand's solution
  • Verbatim language when available
Example:
- "Slow bank approvals" → 24-48 hour turnaround
- "Bank rejection" → Bank-decline rescue financing
- "Self-employed income" → No-income verification options

FROM: Testimonials

Extract:

  • Exact customer quotes (social proof)
  • Specific results mentioned
  • Emotional language used
Example:
- "Got approved in just 2 days with a great rate" — Sarah M.
- "AI search helped me understand all my options" — John D.

FROM: Brand Voice

Extract:

  • Tone to match in hook writing
  • Words/phrases to use or avoid
Example:
"Confident yet accessible, professional but approachable, results-oriented"

FROM: Their Messaging

Extract:

  • Headlines, taglines they already use (starting points)
  • CTAs they use
  • Key phrases they repeat
Example:
- Headline: "Canada's First Lending Hub"
- Tagline: "Not Just Rates. Real Loans for Every Property Type."
- CTAs: "Apply Now", "Start My Deal", "Get started →"

FROM: ICP Segments (EACH segment separately)

For EACH segment extract:

  • WHO they are
  • PAIN POINTS (verbatim language)
  • MOTIVATIONS (immediate + deeper)
  • LANGUAGE they use
Example - Segment 1 (Complex Residential Borrowers):
WHO: Self-employed, business owners, rejected by banks
PAIN: "bank doesn't understand my income", "rejected multiple times"
MOTIVATION: Feel validated, access same opportunities as W-2 employees
LANGUAGE: "self-employed", "bank said no", "complicated income"

Example - Segment 2 (Real Estate Investors):
WHO: BRRR, fix-and-flip, portfolio builders
PAIN: Banks too slow, need creative financing
MOTIVATION: Build wealth, financial freedom
LANGUAGE: "BRRR", "fix and flip", "bridge loan", "LTV"

Example - Segment 3 (Underserved Homebuyers):
WHO: New Canadians, first-time buyers, urgent situations
PAIN: Don't qualify, feeling excluded
MOTIVATION: Canadian dream, stability, prove they belong
LANGUAGE: "new to Canada", "first home", "credit issues"

Extraction Checklist

STOP. Before writing ANY hooks, verify you have extracted from:

  • The Offer — Core promise, key numbers
  • Key Value Props — ALL differentiators listed
  • Proof Points — ALL stats, ratings, real deals
  • Products/Services — EVERY product category, unique offerings noted
  • Pain Points Addressed — ALL pain + solution pairs
  • Testimonials — ALL customer quotes
  • Brand Voice — Tone noted for calibration
  • Their Messaging — Headlines, taglines, CTAs captured
  • ICP Segment 1 — WHO, PAIN, MOTIVATION, LANGUAGE
  • ICP Segment 2 — WHO, PAIN, MOTIVATION, LANGUAGE (if exists)
  • ICP Segment 3 — WHO, PAIN, MOTIVATION, LANGUAGE (if exists)

If any checkbox is unchecked, go back and extract from that section.


Step 2: MATCH (Research Element → Hook Type)

For each extracted element, identify which hook type it best serves:

Research ElementBest Hook TypeWhy
"4.8/5, 856 reviews"Social ProofCredibility through numbers
"24-48 hours vs. 3-6 weeks"ContrastClear before/after
"bank said no"Direct AddressCustomer's exact words
"$547.5K closed in 24 hours"Surprising StatSpecific, unexpected
"350+ lenders"FOMO/AccessMore options than competitors
"AI-powered"CuriosityTechnology intrigue
"Self-employed rejection"Problem-SolutionPain + relief
"Bank Decline Financing" productDirect AddressNames the situation
"BRRR Strategy Loans" productDirect AddressInvestor terminology

Match at least 3-5 research elements to hook types before constructing.


Step 3: CONSTRUCT (Build Hooks)

For each hook, use this structure:

HOOK TYPE: [Which type from the 10]
SOURCE: [Which research section + specific element]
TARGET: [Which ICP segment, or "General"]
HOOK: [The actual headline — under 10 words ideal]
WHY IT WORKS: [Psychology — pattern interrupt? self-reference? open loop?]

Construction Principles

Benefits over features:

  • "Get approved in 48 hours" not "Fast processing"

Direct address ("You"):

  • Make it personal, use identity markers

Specificity:

  • "350+ lenders" not "many options"
  • Exact numbers from Proof Points

Empathy first:

  • Acknowledge struggle before offering solution

Their language:

  • Use verbatim phrases from ICP sections

Example Construction:

HOOK TYPE: Contrast
SOURCE: Value Props ("24-48 hour approval") + Pain Points ("banks take 3-6 weeks")
TARGET: Complex Residential Borrowers
HOOK: "Banks take weeks. We close in 48 hours."
WHY IT WORKS: Pattern interrupt (unexpected speed) + Tension (bad vs. good)

Step 4: BODY + CTA

Body Formula

  1. Expand the hook's promise (don't repeat it)
  2. Add ONE proof element not in the hook
  3. Bridge to CTA

Keep to 1-2 sentences.

CTA Formula

Action Verb + Implied Outcome

Weak CTAStrong CTA
"Learn more""See your options"
"Apply now""Get your answer"
"Contact us""Talk to someone who gets it"
"Get started""Find your lender"

The CTA should feel like the natural next step, not a demand.


Step 5: VARIETY CHECK

Before finalizing, verify diversity:

Hook Types Used

  • At least 3 different hook types represented
  • Not all questions, not all stats

Research Sections Used

  • Not all hooks from Pain Points
  • Proof Points used in at least 1 hook
  • Products/Services referenced in at least 1 hook
  • Value Props featured in at least 1 hook

ICP Segments Covered

  • If multiple segments exist, hooks address different ones
  • Not all hooks targeting same persona

Emotions Triggered

Cover at least 3 of these territories:

  • Fear / Protection (what they might lose)
  • Hope / Aspiration (what becomes possible)
  • Frustration / Anger (what's unfair)
  • Relief / Comfort (burden lifted)
  • Pride / Identity (being seen)
  • Curiosity / Intrigue (hidden information)

If variety is lacking, return to Step 2 and find different research elements to match.


Output Format

Write to: agent/.claude/skills/hook-methodology/hook-bank/{brand}-{YYYY-MM-DD}.md

# [Brand Name] - Hook Bank
**Date:** YYYY-MM-DD
**Hooks Generated:** [number]

## Brand Colors
- Primary: #hex
- Secondary: #hex
- Accent: #hex

## Research Elements Used
[List which sections were mined for each hook]

---

## Hook 1
**Type:** [Hook type from the 10]
**Source:** [Research section + specific element]
**Target:** [ICP segment or "General"]
**Hook:** "[The headline]"
**Body:** [1-2 sentences]
**CTA:** [Action + outcome]
**Psychology:** [Why it stops scroll + drives click]

## Hook 2
[Same structure]

## Hook 3
[Same structure]

[Continue for each hook — default 3, or as specified by user]

---

## Variety Verification
- Hook Types Used: [list]
- Research Sections Used: [list]
- ICP Segments Covered: [list]
- Emotions Triggered: [list]

Hook Volume

  • Default: Generate 3 hooks
  • User-specified: Generate the number requested

Quality over quantity. Each hook must pass all validation checks.


Validation Checklist (Final Pass)

Before each hook is complete, verify:

  1. Is it TRACEABLE? — Can you point to the exact research element?
  2. Is it OWNED? — Could a competitor use this by changing the logo?
  3. Is it FELT? — Does it name an emotion, not just a fact?
  4. Is it CLEAR? — 3 seconds or less to understand?
  5. Is it THEIRS? — Uses their words, their world?

If any answer is no, revise or replace the hook.


Common Failure Modes

The Skip Trap

Skipping Products/Services or Proof Points because they seem "obvious." Every section has hook potential.

The Generic Trap

Writing hooks that could work for any competitor. Use specific proof points to make it OWNED.

The Template Trap

Starting with a formula and filling in blanks. Start with research elements, then match to types.

The Pain-Only Trap

Only using Pain Points section. Testimonials, Products, and Value Props often contain stronger hooks.

The Invention Trap

Creating claims that aren't in the research. If you can't cite the source, don't use it.


Remember

Mine the research. Every hook must be traceable.

The goal is not to be clever.
The goal is to find what the brand has already proven
and frame it in a way that stops the scroll.

If you can't point to where it came from, start over.

平台分布

Claude Code

27.25%
按下载量换算56

OpenCode

23.26%
按下载量换算48

Cursor

17.18%
按下载量换算35

Antigravity

10.3%
按下载量换算21

Gemini CLI

6.92%
按下载量换算14

windsurf

3.32%
按下载量换算7

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

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。

继续浏览同类 Skills