Token导航 LogoToken导航TokenDH.com
研究检索权限需确认github未标认证来源可访问许可证需确认审计提醒

icp-website-auditICP 网站审核

Agent Skill

用于辅助安全审计、权限检查、凭据风险、认证流程和常见漏洞排查。它适合让 Agent 梳理敏感配置、检查依赖风险、分析鉴权逻辑或生成安全复核清单。使用时不能把工具输出直接当最终结论,涉及密钥、令牌、用户数据或生产系统时,应先确认最小权限、脱敏方式和操作边界。

总安装

264

周安装

11

GitHub Stars

607

下载量

88
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

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:icp-website-audit(ICP 网站审核)
来源仓库:https://github.com/athina-ai/goose-skills
仓库路径:skills/icp-website-audit
安装命令:
npx skills add https://github.com/athina-ai/goose-skills --skill icp-website-audit
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/athina-ai/goose-skills --skill icp-website-audit

简介

用于辅助网站安全审计,检查权限控制、认证流程和常见漏洞。

  • 可帮助梳理敏感配置和依赖风险,生成安全复核清单。
  • 不能将工具输出直接视为最终结论,需人工复核关键发现。
  • 涉及密钥或生产系统时,应先确认最小权限和操作边界。
  • 建议结合脱敏方式和日志审计进一步验证潜在威胁。icp-website-audit 属于研究检索类 Skill,可作为该场景下的辅助能力补充。

SKILL.md

ICP Website Audit

The complete "how do buyers experience our site vs the competition?" workflow. Chains persona building, website evaluation, and competitive comparison into a single end-to-end audit.

Quick Start

Run an ICP website audit for [company]. Site: [url]. Compare against [competitor 1] and [competitor 2].

With existing personas:

Run an ICP website audit for [client]. Personas already exist. Compare against [competitor urls].

Inputs

InputRequiredSource
Company nameYesUser provides
Company URLYesUser provides
Competitor URLsYes (1-3)User provides, or discovered in Phase 1
Client context fileOptionalclients/<client>/context.md
Existing personasOptionalclients/<client>/personas/personas.json

Step-by-Step Process

Phase 1: Persona Setup

Check if personas already exist:

clients/<client>/personas/personas.json

If personas exist: Load them, confirm they look current, and list them for the user. Skip to Phase 2.

If no personas exist: Run buyer-persona-generator:

  1. Research the company — what they sell, who they serve, pricing model, stage
  2. Identify 4-6 ICP segments from website, case studies, reviews, job postings
  3. Build detailed synthetic personas with full profiles
  4. Save to clients/<client>/personas/

Output from this phase:

  • clients/<client>/personas/personas.json (machine-readable)
  • clients/<client>/personas/personas.md (human-readable)
  • clients/<client>/personas/segments.md (summary table)

Phase 2: Website Scorecard Review

Run icp-website-audit in scorecard mode against the client's own site.

  1. Crawl the client's site — homepage, pricing, product, solutions, about, case studies, blog, docs
  2. Check external presence — search results, review sites, social proof
  3. Run each persona through the site, scoring on:

- First Impression (1-10) - Messaging Relevance (1-10) - Trust & Credibility (1-10) - Clarity & Navigation (1-10) - Objection Handling (1-10) - Overall (1-10)

  1. Cross-persona synthesis — consensus issues, segment-specific gaps, messaging disconnects

Output from this phase:

  • Per-persona scored reviews
  • Cross-persona analysis
  • Initial recommendations

Phase 3: Competitive Head-to-Head

Run icp-website-audit in head-to-head mode against each competitor.

For each competitor (1-3):

  1. Crawl the competitor's site — same pages as the client site
  2. Run each persona through BOTH sites
  3. Per-persona comparison:

- Quick takes on each site - Dimension-by-dimension scoring for both - "If I had to choose today..." verdict - What the client should steal from the competitor - What the client does better

  1. Cross-persona competitive summary:

- Where the client wins across personas - Where the competitor wins across personas - At-risk segments (personas who lean toward competitor) - Priority competitive moves

Output from this phase:

  • Per-competitor, per-persona head-to-head reviews
  • Competitive positioning map
  • At-risk segments identified

Phase 4: Consolidated Report

Merge all findings into a single audit report. This is the deliverable.

Synthesis work:

  1. Combine scorecard + competitive findings — scorecard shows absolute quality, head-to-head shows relative positioning. Both perspectives matter.
  2. Identify the "so what" — what are the 3-5 moves that would make the biggest difference?
  3. Rank recommendations by:

- How many personas it affects (breadth) - How much it would change conversion likelihood (depth) - Whether competitors already do it better (urgency) - Estimated effort (feasibility)

  1. Flag at-risk segments — ICPs where competitors are winning. These need immediate attention.

Output

Save to the current working directory as icp-website-audit.md (or user-specified path).

Also save the sub-reports to the current working directory:

  • personas/ (persona assets, reusable)
  • icp-reviews/<date>-scorecard.md (site review)
  • icp-reviews/<date>-head-to-head-<competitor>.md (per competitor)

Consolidated Report Template

# ICP Website Audit: [Company Name]

**Date:** [Date]
**Website:** [url]
**Competitors:** [competitor 1], [competitor 2], [competitor 3]
**Personas used:** [N] (from `clients/<client>/personas/`)

---

## Executive Summary

[5-7 sentences. How do buyers actually experience this site? What's working, what's
broken, where are competitors winning? Top 3 things to fix and why.]

### Scorecard Overview

**Average score across personas: [X/10]**

| Persona | Segment | Own Site Score | Would Convert? |
|---------|---------|:-:|:-:|
| [Name] | [Segment] | [X/10] | [Yes/Maybe/No] |
| ... | ... | ... | ... |

### Competitive Overview

| Persona | Own Site | [Comp 1] | [Comp 2] | Pick |
|---------|:-------:|:--------:|:--------:|:----:|
| [Name] | [X/10] | [X/10] | [X/10] | [Winner] |
| ... | ... | ... | ... | ... |

**Personas at risk:** [Names of personas who preferred a competitor — these represent ICP segments you may be losing]

---

## Part 1: ICP Segments & Personas

### Segments Identified

| # | Segment | Description | Priority |
|---|---------|-------------|----------|
| 1 | [Name] | [One line] | Primary |
| 2 | [Name] | [One line] | Primary |
| 3 | [Name] | [One line] | Secondary |
| ... | | | |

### Persona Summaries

[2-3 sentence summary of each persona — who they are, what they care about, how they evaluate. Link to full profiles in `personas/personas.md`.]

---

## Part 2: Website Scorecard

### Score Matrix

| Dimension | [P1] | [P2] | [P3] | [P4] | [P5] | Avg |
|-----------|:----:|:----:|:----:|:----:|:----:|:---:|
| First Impression | | | | | | |
| Messaging Relevance | | | | | | |
| Trust & Credibility | | | | | | |
| Clarity & Navigation | | | | | | |
| Objection Handling | | | | | | |
| **Overall** | | | | | | |

### Persona-by-Persona Highlights

For each persona, include:
- **[Persona Name]** ([Score]/10) — [One sentence summary]. Liked: [key positive]. Frustrated by: [key negative]. Blocker: [#1 thing stopping conversion].

### Cross-Persona Findings

#### Consensus Issues (All Personas)
1. [Issue] — [Why it matters]
2. ...

#### Segment-Specific Gaps
| Gap | Personas Affected | Impact |
|-----|------------------|--------|
| [Gap] | [Names] | [H/M/L] |

#### Messaging Disconnects
| Site says | Buyers say | Personas |
|-----------|-----------|----------|
| "[quote]" | "[quote]" | [Names] |

---

## Part 3: Competitive Analysis

### vs [Competitor 1]

**Overall: [Company wins / Competitor wins / Split]**

| Dimension | [Company] | [Competitor] | Winner |
|-----------|:---------:|:------------:|:------:|
| First Impression | [avg] | [avg] | |
| Messaging | [avg] | [avg] | |
| Trust | [avg] | [avg] | |
| Clarity | [avg] | [avg] | |
| Objections | [avg] | [avg] | |
| **Overall** | **[avg]** | **[avg]** | |

**Where competitor wins:**
- [Specific advantage, which personas care]

**Where we win:**
- [Specific advantage, which personas care]

**Personas who'd choose competitor:** [Names — these are at-risk segments]

---

### vs [Competitor 2]
[Same structure]

---

### Competitive Position Map

| Persona | Best Experience | Worst Experience | Notes |
|---------|:-:|:-:|---|
| [P1] | [Company/Comp] | [Company/Comp] | [Why] |
| [P2] | ... | ... | ... |

### At-Risk Segments

[Personas who preferred competitors. For each: which competitor, why, and what it would take to win them back.]

### Competitive Advantages to Amplify

[Things the company does better than ALL competitors — lean into these.]

### Gaps to Close

[Things competitors do better across most personas — neutralize these.]

---

## Part 4: Prioritized Recommendations

### Tier 1: High Impact, Do Now

These affect the most personas and/or address competitive gaps.

1. **[Recommendation]**
   - **Why:** [Evidence from scorecard + competitive analysis]
   - **Personas affected:** [Names]
   - **Competitive context:** [Do competitors already do this better?]
   - **Expected impact:** [What changes — conversion, trust, clarity]
   - **Effort:** [Low/Med/High]

2. **[Recommendation]**
   ...

### Tier 2: Medium Impact, Plan For

3. **[Recommendation]**
   ...

### Tier 3: Segment-Specific Optimizations

These improve experience for specific ICP segments.

5. **[Recommendation]** — For [persona/segment]
   ...

### Quick Wins

- [Small change] → Benefits [persona]
- [Small change] → Benefits [persona]

---

## Appendix

### A. Full Persona Profiles
→ `clients/<client>/personas/personas.md`

### B. Detailed Scorecard Review
→ `clients/<client>/icp-reviews/<date>-scorecard.md`

### C. Detailed Competitive Reviews
→ `clients/<client>/icp-reviews/<date>-head-to-head-<competitor>.md`

### D. Persona Data (JSON)
→ `clients/<client>/personas/personas.json`

Parallelization

Some phases can run in parallel to save time:

Phase 1: Build personas (if needed)
    ↓
Phase 2 + Phase 3 can partially overlap:
  - Crawl all sites (client + competitors) at once
  - Run scorecard first (client site only)
  - Then run head-to-heads (reuse client crawl data)
    ↓
Phase 4: Consolidate (needs all prior phases)

If personas already exist, the whole workflow is just:

Crawl all sites → Scorecard → Head-to-heads → Consolidate

Tips

  • Start with 1-2 competitors, not 5. Each competitor adds a full head-to-head evaluation pass. Start focused, add more if needed.
  • The at-risk segments are the most actionable finding. If a persona consistently prefers a competitor, that's an ICP segment you're actively losing deals in. That should drive urgency.
  • Reuse this quarterly. Run personas once, then re-run the audit after site changes or quarterly to track improvement. Compare scores across dates.
  • Combine with SEO content audit for the full picture. ICP audit tells you "does the messaging work for buyers?" SEO audit tells you "are buyers finding you in the first place?" Together they cover the full funnel.
  • This makes a great client deliverable. The consolidated report is structured for sharing. Use content-asset-creator to make it into a polished HTML report.
  • Don't skip the competitive analysis. The scorecard alone is useful, but the head-to-head is where the real insights are. Buyers always compare — your audit should too.

Dependencies

  • Skills: buyer-persona-generator, icp-website-audit
  • Web search capability (for research, external presence checks)
  • Web fetch capability (for reading website pages)
  • No API keys or paid tools required

Cost

Free — no paid APIs. All research and evaluation is done via web search and web fetch.

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

36.33%
按下载量换算32

Claude

31.7%
按下载量换算28

Cursor

16.97%
按下载量换算15

Gemini CLI

10.18%
按下载量换算9

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

权限需确认

当前来源未能明确判断权限范围,默认进入异常复核队列。

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 warning/failed 结果,不能写成本站确认安全。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills