GEO Tracker

Track and optimize brand visibility across AI search engines (ChatGPT, Perplexity, Gemini, Google AI Overview, Claude). Use when monitoring brand mentions in...

Security Scan

The skill's code and runtime instructions mostly match its stated purpose, but the registry metadata omits required API credentials and the source is unknown — this mismatch and lack of provenance warrant caution before installing.

Purpose & Capability

The skill claims to track brand visibility across multiple AI engines and the included Python scripts implement exactly that (OpenAI, Perplexity, Google Generative AI, Anthropic). However, the registry metadata lists no required environment variables or primary credential despite the SKILL.md and code requiring multiple API keys (OPENAI_API_KEY, PERPLEXITY_API_KEY, GOOGLE_API_KEY, ANTHROPIC_API_KEY). That metadata omission is incoherent and should be corrected; otherwise the skill's manifest understates what it needs.

Instruction Scope

SKILL.md gives concrete instructions to run the provided scripts, set API keys, and schedule audits. The runtime instructions do not request unrelated filesystem reads or other credentials beyond the provider API keys. One minor inconsistency: SKILL.md mentions a 'Google AI Overview' web_search tool, but the code implements only API calls (no explicit web scraping/web_search implementation for a separate 'Google AI Overview').

Install Mechanism

This is instruction-only with Python scripts and a standard pip dependency list (openai, anthropic, google-generativeai). There is no download-from-URL or archive extraction; installing Python packages via pip is expected for this use case.

Credentials

The skill legitimately needs API keys for each engine it queries, so multiple secret-like env vars are reasonable. The problem is the published registry metadata declares no required env vars while SKILL.md and the code expect several sensitive keys. This mismatch could mislead users into installing without preparing credentials. Also, supplying provider API keys gives the code the ability to query those services (and incur billing), so users should use limited-scope or test keys where possible.

Persistence & Privilege

The skill does not request 'always: true', does not modify other skills or system-wide configuration, and only writes a report file when the user requests an output path. It uses normal agent invocation semantics and does not demand elevated or persistent platform privileges.

What to consider before installing

This skill's scripts do what the description says, but the registry metadata fails to list the API keys the code actually requires — treat that as a red flag. Before installing: (1) view and review the included scripts (you already have them) and verify you understand the network calls; (2) do not provide production API keys immediately — test with limited-privilege or dummy keys and run in an isolated virtual environment; (3) be aware that supplying provider keys lets the tool make API calls that may incur charges and reveal usage to those providers; (4) confirm the skill's source/provenance (there's no homepage and the owner ID is opaque); and (5) if you plan to run scheduled audits, ensure the scheduling agent runs in a controlled environment with appropriate network and billing limits. If the registry metadata is later corrected to declare the required env vars, the coherence concerns would be resolved.

Track how AI engines mention (or ignore) a brand. Query multiple AI-powered search engines, extract brand mentions, score visibility, and generate actionable optimization reports.

Quick Start

Single Query Check

python3 scripts/geo_query.py --brand "EZsite.ai" --query "best AI website builder" --engines chatgpt,perplexity,gemini

Full Audit (batch prompts)

python3 scripts/geo_audit.py --brand "EZsite.ai" --prompts references/prompts.txt --engines all --output report.md

Competitor Comparison

python3 scripts/geo_query.py --brand "EZsite.ai" --competitors "Wix,Squarespace,Framer" --query "best website builder for small business"

Dependencies

Install required Python packages (one-time):

pip3 install openai anthropic google-generativeai

Or create a virtual environment:

python3 -m venv venv
source venv/bin/activate
pip install openai anthropic google-generativeai

Set API keys as environment variables:

export OPENAI_API_KEY="sk-..."
export PERPLEXITY_API_KEY="..."
export GOOGLE_API_KEY="..."
export ANTHROPIC_API_KEY="sk-ant-..."

How It Works

1. Takes a brand name + search prompts
2. Queries AI engines via their APIs or web interfaces
3. Extracts: mentions, citations, sentiment, positioning
4. Scores visibility (0-100) per engine and overall
5. Generates optimization recommendations

Engines Supported

Engine	Method	API Key Env Var
ChatGPT	OpenAI API	OPENAI_API_KEY
Perplexity	Perplexity API	PERPLEXITY_API_KEY
Gemini	Google AI API	GOOGLE_API_KEY
Claude	Anthropic API	ANTHROPIC_API_KEY
Google AI Overview	web_search tool	(none)

At minimum, configure one API key. More engines = better coverage.

Visibility Score

• 0-20: Invisible — AI doesn't know the brand
• 21-40: Low — occasional mentions, not recommended
• 41-60: Moderate — mentioned but not top choice
• 61-80: Strong — frequently cited/recommended
• 81-100: Dominant — top recommendation across engines

Output Format

Reports include:

• Per-engine mention count and context
• Visibility score breakdown
• Competitor comparison matrix
• Top optimization recommendations
• Source prompts that triggered (or missed) mentions

Prompt Library

Edit references/prompts.txt — one prompt per line. These are the queries sent to AI engines.

Example prompts for a website builder brand:

best AI website builder
how to build a website without coding
website builder comparison 2025
best website builder for small business
AI-powered web design tools

Optimization Tips Reference

See references/geo-optimization.md for content optimization strategies to improve AI visibility.

Scheduling

Use OpenClaw cron to run daily/weekly audits:

Schedule a daily GEO audit for EZsite.ai at 9am

The agent will run the audit and report findings.