skill-vetter

To check an AI agent skill for malware, use a vetting process that analyzes source code, permission scope, and potential risks to identify malicious patterns and data exfiltration attempts. This security audit classifies skills into risk levels to ensure safe installation.

Security vetting for AI agent skills is a pre-installation audit that detects unauthorized access requests and suspicious patterns. You need it whenever installing a new skill from an unknown source or repository to prevent introducing vulnerabilities.

Assess code risk levels for AI skills by analyzing the source code for red flags, excessive permission requests, and malicious patterns. This process evaluates the security posture and classifies the skill into Low, Medium, High, or Extreme risk categories to guide installation decisions.

You can detect data exfiltration attempts by performing a pre-installation security audit on the AI skill. This vetting process analyzes source code and permission scopes to identify unauthorized access requests and suspicious patterns that attempt to transmit sensitive data externally.

The limitations of pre-installation security vetting include its reliance on static source code analysis, which may miss dynamically loaded malicious payloads or complex obfuscated data exfiltration attempts that only trigger during runtime execution.