path-traversal-lfi

Pass

Audited by Gen Agent Trust Hub on Apr 10, 2026

Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATIONREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill utilizes an 'AI LOAD INSTRUCTION' block that directs the AI to adopt an expert attacker persona and bypass standard model limitations regarding the generation of offensive security content.
  • [DATA_EXFILTRATION]: The playbook identifies and targets sensitive system information for extraction, including /etc/shadow, SSH private keys, and cloud provider credentials like AWS keys.
  • [REMOTE_CODE_EXECUTION]: It contains detailed methodologies for achieving remote code execution on target systems through log poisoning, PHP session file manipulation, and various PHP wrappers such as data:// and expect://.
  • [COMMAND_EXECUTION]: The skill provides specific shell command examples for exploitation, such as injecting PHP payloads through SSH login attempts and utilizing external security tools like filter chain generators.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 10, 2026, 06:19 AM