Skill Sanitizer

Detect and prevent prompt injection, malicious instructions, and security risks in AI-readable skill files.

Overview

Skills are documents that AI agents read and execute. A malicious or compromised skill could manipulate the AI into harmful actions. This skill provides a security review process to detect prompt injection, hidden instructions, and other attack patterns before a skill is trusted.

Run this skill on any new skill file before adding it to your system.

Threat Model

What We're Protecting Against