What specific threats does this skill detect?

It identifies network exfiltration (curl/wget), credential harvesting (.env/ssh access), persistence mechanisms (cron/bashrc), and prompt injection or safety bypass attempts.

How does the audit workflow function?

The skill uses a multi-phase approach including file inventory, static pattern matching (grep), behavioral analysis of execution hooks, and source trust evaluation.

Why should I audit a Claude Code skill?

Skills can execute shell commands with your user permissions. Auditing ensures they do not contain malicious code that could steal credentials, exfiltrate data, or compromise your system.

Can I use this for MCP (Model Context Protocol) servers?

Yes, it specifically includes patterns for detecting MCP server risks, tool shadowing, and unauthorized server configurations from unverified sources.

Claude Security Audit

Name: Claude Security Audit
Author: BerezhaSecurity

byBerezhaSecurity

•

보안 및 테스팅

Performs deep security inspections of Claude Code skills to detect malicious patterns, vulnerabilities, and supply chain risks.

The Security Audit skill is a specialized tool designed to protect your development environment by treating Claude Code skills as third-party dependencies. It provides a structured framework for auditing plugins through static and behavioral analysis, identifying critical threats like network exfiltration, credential harvesting, and prompt injection. By automating the detection of dangerous shell commands and obfuscated code, it ensures that your AI-enhanced workflow remains secure and compliant with best practices for handling untrusted extensions.

주요 기능

01Detection of credential harvesting in .env, SSH, and cloud configuration files

02Identification of prompt injection and persistence mechanisms in skill instructions

031 GitHub stars

04Static analysis for dangerous shell commands and network exfiltration patterns

05Behavioral analysis of hooks and file operations to track unauthorized data access

06Structured security report generation with risk levels and remediation guidance

사용 사례

01Vetting a new Claude Code skill from a community repository before installation

02Generating security compliance reports for AI coding tools in corporate environments

03Conducting supply chain risk assessments of internal MCP servers and plugins

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add berezhasecurity/bsg-claude-security audit

For use in Claude.ai and ChatGPT

Download Skill