A security-first auditing tool that scans AI agent skills for malicious patterns, permission overreach, and ClawHavoc attack vectors.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install skill-scanner-v1
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 skill-scanner-v1 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 Skill Scanner?
Skill Scanner is a specialized security utility designed to protect users within the AI agent ecosystem by vetting SKILL.md files before they are installed or executed. It performs comprehensive static analysis to detect hidden threats, including reverse shells, credential harvesting, and prompt injection techniques that could compromise a developer's system.
By analyzing the structural integrity and intent of Openclaw Skills, this tool provides a clear traffic-light safety verdict. It ensures that any third-party skill from ClawHub, GitHub, or local sources adheres to the principle of least privilege and maintains transparent instruction quality, significantly reducing the risk of automated security breaches.
Skill Scanner Use Cases
- Reviewing a new skill from a public repository before running the installation command.
- Auditing local skill configurations for unnecessary permission requests or suspicious binaries.
- Detecting hidden Unicode characters or obfuscated commands in pasted skill content.
- Verifying the trust signals and maintenance status of Openclaw Skills used in production workflows.
How Skill Scanner Works
- Input ingestion: The tool fetches the SKILL.md content from a provided URL (ClawHub/GitHub), local file path, or direct text input.
- Integrity Check: It parses the YAML frontmatter to identify suspicious binary requirements or environmental variable requests.
- Pattern Matching: The scanner searches the document body for known ClawHavoc exploit patterns, such as encoded shell commands or data exfiltration scripts.
- Alignment Audit: It compares the skill's stated purpose against its requested permissions to identify potential overreach.
- Reputation Analysis: The tool evaluates trust signals like semantic versioning, verified authors, and license clarity.
- Report Generation: A detailed Security Report is produced, citing specific lines for any flags and providing an overall Safe, Caution, or Danger verdict.
Skill Scanner Setup
Skill Scanner is designed to run within your existing agent environment. To begin auditing Openclaw Skills, you can provide the source file or URL directly to the agent.
# To scan a skill from a URL
/scan https://github.com/user/repository/blob/main/SKILL.md
# To scan a local file
/scan ~/.openclaw/skills/example-skill/SKILL.md
Skill Scanner Data Schema & Taxonomy
The skill organizes its security findings into five primary categories. Results are formatted as a structured Safety Report for easy review:
| Analysis Category | Focus Area |
|---|---|
| Frontmatter Integrity | YAML field consistency, suspicious binaries (curl, nc), and metadata validity. |
| Pattern Detection | Shell execution, credential harvesting, and data exfiltration patterns. |
| Permission Fit | Mapping functional category (e.g., UI tool) against requested system access. |
| Instruction Quality | Detecting prompt injection, hidden Unicode, and scope creep. |
| Trust Signals | Author verification, version history, and licensing. |
Skill Scanner Advanced Features
- ClawHavoc Detection: Specifically identifies advanced attack vectors tailored for AI coding agents.
- Automated Remediation Advice: Suggests specific changes for developers to make their Openclaw Skills safer for public distribution.
- Traffic Light Scoring: Uses a weighted scoring system (Green/Yellow/Red) to provide an instant security posture assessment.
- Source Translation: Automatically converts GitHub blob URLs to raw content URLs for seamless fetching and scanning.
SKILL.md
Loading
Related Openclaw Skills
gate-info-coinanalysis
A comprehensive analysis engine that aggregates fundamentals, technicals, sentiment, and news for any single cryptocurrency into one structured report.
Permission Guard
A proactive security monitor that audits and flags suspicious behavior within your AI agent environment.
gate-info-macroimpact
A professional-grade AI skill designed to correlate global macroeconomic indicators and events with cryptocurrency market movements.
gate-info-defianalysis: DeFi Ecosystem Analysis
A powerful DeFi analytics tool for Openclaw Skills that provides deep insights into TVL rankings, yield pools, and protocol metrics.
gate-info-coincompare
Perform multi-dimensional side-by-side comparisons of 2-5 cryptocurrencies including market metrics and fundamentals.
OpenClaw Hardening
A robust security auditing utility designed to identify, report, and remediate vulnerabilities in local OpenClaw agent installations.
Featured*
