Can I use this in my GitHub Actions?

Yes, the tool supports CLI execution with JSON output and strict modes, making it ideal for automated CI/CD security gating.

How does it detect prompt injection?

It scans SKILL.md and reference files for known injection patterns like system prompt overrides, role hijacking, and hidden instructions.

What do the PASS/WARN/FAIL verdicts mean?

PASS indicates no critical issues; WARN suggests manual review of high/medium risks; FAIL means critical vulnerabilities were found that require immediate remediation.

Does this tool execute the code it scans?

No, it uses static analysis to identify dangerous patterns without executing scripts, making it safe to run on untrusted code.

What types of AI agent skills can be audited?

It is optimized for Claude Code plugins, OpenClaw skills, and Codex skills, supporting Python, shell scripts, and JavaScript/TypeScript.

Skill Security Auditor

Name: Skill Security Auditor
Author: oabdelmaksoud

byoabdelmaksoud

0•

Security & Testing

Secures AI agent development by scanning plugins and skills for malicious code, prompt injections, and supply chain vulnerabilities before installation.

Skill Security Auditor is a robust defensive tool designed to protect AI development environments from malicious or poorly-configured agent skills. It performs deep static analysis on Python, shell, and JavaScript files to identify dangerous patterns like command injection and network exfiltration, while also auditing SKILL.md files for prompt injection risks. By providing a clear PASS/WARN/FAIL verdict and specific remediation guidance, it serves as a critical security gate for developers integrating third-party tools into Claude Code, OpenClaw, or Codex ecosystems.

Key Features

01Deep static analysis for command and code injection patterns

02Comprehensive audit reports with actionable remediation steps and severity levels

03Dependency supply chain risk assessment for PyPI and npm packages

04Prompt injection detection in markdown documentation and system instructions

050 GitHub stars

06Filesystem boundary and unauthorized access verification

Use Cases

01Integrating automated security gates into CI/CD pipelines for agent-based projects

02Auditing internal skill directories to ensure compliance with enterprise security standards

03Evaluating third-party AI skills from untrusted repositories before local installation

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add oabdelmaksoud/agention skill-security-auditor

For use in Claude.ai and ChatGPT

Key Features

01Deep static analysis for command and code injection patterns

02Comprehensive audit reports with actionable remediation steps and severity levels

03Dependency supply chain risk assessment for PyPI and npm packages

04Prompt injection detection in markdown documentation and system instructions

050 GitHub stars

06Filesystem boundary and unauthorized access verification

Use Cases

01Integrating automated security gates into CI/CD pipelines for agent-based projects

02Auditing internal skill directories to ensure compliance with enterprise security standards

03Evaluating third-party AI skills from untrusted repositories before local installation

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add oabdelmaksoud/agention skill-security-auditor

For use in Claude.ai and ChatGPT