The Agent Skills Directory

[SAFE]: The skill serves as an instructional framework for static authorization testing and risk modeling. It implements strong security controls by explicitly prohibiting the AI agent from accessing real credentials, calling APIs, or performing any real-system execution. The provided prompts and evaluation cases focus entirely on evidence-based modeling of subject-resource-action relationships. No malicious patterns such as obfuscation, exfiltration, or privilege escalation were identified.

authorization-testing