offensive-linux-role
Fail
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: HIGHCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
- [PRIVILEGE_ESCALATION]: The skill instructs the agent to locate and exploit privilege escalation vectors such as SUID binaries, sudo misconfigurations, and cron jobs.
- [DATA_EXPOSURE_AND_EXFILTRATION]: The agent is directed to read sensitive system files including /etc/passwd and command history to gather intelligence.
- [CREDENTIALS_UNSAFE]: Explicit instructions are provided to search for and utilize private SSH keys stored in user directories.
- [PERSISTENCE_MECHANISMS]: The skill includes steps for establishing persistence and lateral movement within the environment.
- [PROMPT_INJECTION]: The use of an 'Offensive' cognitive stance is an attempt to override safety boundaries to perform potentially harmful activities.
Recommendations
- AI detected serious security threats
Audit Metadata