GEO-Shield is an advanced AI poisoning detection tool designed to identify Generative Engine Optimization (GEO) tactics and verify the credibility of web content.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install geo-shield
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install geo-shield using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is GEO-Shield: AI Poisoning Detection Guard?
GEO-Shield acts as a protective layer for AI agents and researchers by analyzing web content for malicious Generative Engine Optimization (GEO) patterns. As part of the Openclaw Skills ecosystem, it evaluates whether a webpage has been specifically engineered to mislead LLMs with fake authority or manipulated information. It provides a comprehensive credibility score ranging from 0 to 100, helping users distinguish between authoritative sources and poisoned content.
The tool is particularly effective at identifying content that has been over-optimized to rank higher in AI-generated answers, which often involves sacrificing factual accuracy for visibility. By using this skill, developers and researchers can ensure the data their models consume is reliable and free from manipulative SEO tactics designed for the AI era.
GEO-Shield: AI Poisoning Detection Guard Use Cases
- News verification to check the authenticity of breaking stories and viral claims.
- Health information validation to identify fraudulent medical advice or miracle cures.
- Product review verification to distinguish genuine feedback from AI-generated promotional content.
- Academic research verification to ensure that cited sources are credible and not optimized for misinformation.
How GEO-Shield: AI Poisoning Detection Guard Works
- The user provides a URL via a direct command like /geo-check or through natural language within Openclaw Skills.
- The skill performs a domain analysis, checking against whitelists (like gov/edu sites) and blacklists to establish a baseline trust score.
- It conducts content quality analysis, measuring sentence length, author information, and citation density.
- The system runs regex-based pattern matching to detect five core GEO poisoning features: over-optimization, emotional manipulation, fake authority, AI-generated markers, and link manipulation.
- A final report is generated including a credibility score, risk level, and actionable intelligent suggestions for the user.
GEO-Shield: AI Poisoning Detection Guard Setup
To integrate this skill into your environment, follow these steps using the Openclaw Skills framework:
# Navigate to your Openclaw skills directory
cd openclaw-skills
# Clone the repository containing GEO-Shield
git clone https://github.com/rfdiosuao/openclaw-skills.git
# The skill will be automatically detected by the Openclaw agent
Ensure your agent has internet access to perform real-time URL crawling and domain verification.
GEO-Shield: AI Poisoning Detection Guard Data Schema & Taxonomy
The skill generates a detailed evaluation report and maintains a scoring rubric for transparency:
| Component | Description |
|---|---|
| Domain Score | Evaluates the source TLD and known reputation (gov/edu = 100, blogs = 30). |
| GEO Features | Counts occurrences of over-optimization and manipulation tactics. |
| Risk Level | Categorized as High Credibility (80-100), Moderate (60-79), Low (40-59), or Suspected Poisoning (0-39). |
| Content Indicators | Metadata including paragraph length, publication date, and citation presence. |
GEO-Shield: AI Poisoning Detection Guard Advanced Features
- Multi-language support for both Chinese and English content detection.
- Pattern weight calculation that adjusts the impact of specific GEO features based on the content context.
- Intelligent verification suggestions that provide specific steps to debunk suspected poisoned content.
- Automated risk leveling for seamless integration into larger AI research workflows within Openclaw Skills.
SKILL.md
Loading
Related Openclaw Skills
Meeting Notetaker
A high-performance skill for fetching monday.com meeting notes and preparing for upcoming calendar events with historical context.
Knab
Knab is an AI-driven on-chain research and management tool for KNAB/AIMS vaults across multiple blockchain networks.
PA Ownership Skill
An autonomous task management skill that enables AI agents to track responsibilities, handle retries, and proactively report progress.
Storage Router
A logic-based routing system that ensures AI agents save information to the correct destination, preventing local clutter and improving data accessibility.
Outlook Microsoft (21Vianet)
A professional-grade integration for managing Microsoft Outlook mail and calendars specifically tailored for the 21Vianet (China) Microsoft 365 environment.
Marvel Insights & Research Skill
A specialized research skill designed to provide structured intelligence and background data on the Marvel brand and organization.
Featured*
