pyjail
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
- [Dynamic Execution] (HIGH): The skill provides payloads for traversing Python's class hierarchy to access restricted modules and builtins via
__subclasses__.- [Command Execution] (HIGH): Includes explicit code to execute shell commands usingos.systemand thebreakpoint()function to gain unauthorized system access.- [Privilege Escalation] (HIGH): The primary purpose of the skill is to 'escape' restricted environments, which constitutes a direct attempt at privilege escalation.- [Obfuscation] (MEDIUM): Teaches the evasion of security filters through string concatenation and character encoding (chr) to bypass keyword-based blocklists.
Recommendations
- AI detected serious security threats
Audit Metadata