作者：Mike Futia ·更多类似的Claude代码系统在SCALE AI中

当人们询问人工智能在你类别中的最佳产品时，它推荐的是你还是你的 竞争对手？这项技能会通过Google AI概览来处理你类别的买家问题， ChatGPT和Perplexity检测你的品牌是否被命名/引用，显示AI推荐了谁 而非，并对整体AI可见度进行评分——在仪表盘中。仅限Python标准库;一个密钥（Apify）。

安装

克洛德代码

Claude桌面版

1. 将该仓库下载为ZIP（代码→下载ZIP）。
2. 设置 → 能力 → 技能 → 上传技能 →下放压缩包。

然后进行一次性令牌设置：跟随PLAYBOOK.md（一个Apify API令牌在.env中）。 将 .env.example 复制到 .env，并粘贴你的令牌在那里——.env 被 gitignoror，所以它保持本地状态。

使用

运行AI可见性审计

它会问两个问题——你的品牌+网站和你的类别——然后生成 买家提问，查询AI引擎，并打开显示你的可见性评分的仪表盘， 各引擎分解、竞争对手排行榜以及如何弥补差距。 输出降落在./ai-visibility/。

里面是什么

注释

AI Visibility Audit (Claude skill)

By Mike Futia · more Claude Code systems like this inside SCALE AI

When people ask AI for the best product in your category, does it recommend you — or your competitors? This skill runs your category’s buyer questions through Google AI Overview, ChatGPT, and Perplexity, detects whether your brand is named/cited, shows who AI recommends instead, and scores your overall AI visibility — in a dashboard. Python stdlib only; one key (Apify).

Install

Claude Code

git clone https://github.com/mikefutia/ai-visibility-audit.git ~/.claude/skills/ai-visibility-audit

Claude Desktop

1. Download this repo as a ZIP (Code → Download ZIP).
2. Settings → Capabilities → Skills → Upload skill → drop the zip.

Then do the one-time token setup: follow PLAYBOOK.md (an Apify API token in a .env). Copy .env.example to .env and paste your token there — .env is gitignored, so it stays local.

Use

run the ai visibility audit

It asks two questions — your brand + website and your category — then generates the buyer questions, queries the AI engines, and opens a dashboard with your visibility score, per-engine breakdown, the competitor leaderboard, and how to fix the gaps. Output lands in ./ai-visibility/.

What’s inside

ai-visibility-audit/
├── SKILL.md                 the runbook Claude follows
├── PLAYBOOK.md              one-time Apify token setup
├── .env.example             copy to .env, add your Apify token
├── .gitignore               keeps .env and run output out of git
└── scripts/
    ├── run_audit.py         queries the AI engines, detects brand presence (Apify, stdlib)
    └── render_dashboard.py  renders the dark dashboard (stdlib)

Notes

• Being named in the answer is the goal (that’s a recommendation) — stronger than just being cited in the sources.
• Runs take a few minutes (AI engines are slow). ~$0.30–0.60 per audit.
• Requires an Apify token (PLAYBOOK.md) + Python 3.

Example run

Run against HexClad (24 buyer questions, US, July 2026):

A brand that spends heavily on ads was named every time a buyer already knew it, and almost never in the questions that start a purchase — which is the gap this skill is built to find.

Who built this

Made by Mike Futia.

I build production-grade Claude Code systems for ecommerce brands, creative agencies, and performance marketers, and I drop new workflows like this one every week inside my community.

Join 600+ brands and agencies in SCALE AI →