What specific security risks does the Skill Security Scanner detect?

It identifies prompt injection, malicious script execution (such as reverse shells or credential theft), excessive tool permissions, secret exposure in code, and supply chain vulnerabilities.

Does this skill require any specific tools or dependencies?

It utilizes standard developer tools like Grep, Glob, and Bash, along with a bundled Python script that can be executed via 'uv' for static analysis.

Can it scan skills before I install them?

Yes, the scanner is designed to analyze skill directories and repositories locally before adoption, ensuring they are safe to use in your environment.

What kind of output does a security scan produce?

It generates a structured markdown report including a risk level (Critical to Clean), a summary of findings, evidence from the source files, and specific remediation advice.

How does it handle false positives in security skills?

The scanner includes a specialized workflow to evaluate intent, distinguishing between skills that document injection patterns for educational purposes and those that actually attempt to execute them.

Skill Security Scanner

Name: Skill Security Scanner
Author: getsentry

bygetsentry

•

118

•

Seguridad y Pruebas

Audits Claude Code skills for security vulnerabilities, prompt injection, and malicious patterns before installation.

Skill Security Scanner is a specialized tool designed to safeguard your AI development environment by performing rigorous security audits on agent skills. It combines automated static analysis with structured behavioral workflows to identify critical risks including prompt injection, data exfiltration, secret exposure, and excessive tool permissions. By analyzing SKILL.md instructions and bundled scripts, it provides a comprehensive risk assessment, allowing developers to verify the safety of community-contributed or third-party skills before integration. This skill is essential for maintaining a secure and trustworthy agent ecosystem by ensuring all capabilities align with the principle of least privilege and security best practices.

Características Principales

01Permission validation and least-privilege assessment

02118 GitHub stars

03Automated static analysis for deterministic security patterns

04Detailed security reporting with severity-based risk levels

05Detection of prompt injection and obfuscated jailbreak attempts

06Analysis of scripts for data exfiltration and reverse shells

Casos de Uso

01Auditing internal agent skills for security compliance and secret leaks

02Vetting third-party Claude skills before adding them to a project

03Validating the safety of community-contributed skills from public repositories

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add getsentry/dotagents skill-scanner

For use in Claude.ai and ChatGPT

Download Skill