tunneling-and-pivoting

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONPRIVILEGE_ESCALATIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill contains a large volume of command-line instructions for establishing network proxies and tunnels, including SSH port forwarding, SOCKS proxy configuration, and the use of specialized relay tools like socat.
  • [PRIVILEGE_ESCALATION]: Documentation includes commands that require administrative or root privileges, such as using sudo to create TUN interfaces for Ligolo-ng and using netsh for port forwarding on Windows.
  • [DATA_EXFILTRATION]: The skill describes methods for exposing local services to remote hosts (remote port forwarding) and includes socat command examples for transferring files between hosts.
  • [INDIRECT_PROMPT_INJECTION]: The skill outlines workflows where the agent interacts with consoles and outputs from untrusted external hosts (e.g., Ligolo-ng or dnscat2 sessions), which represents an indirect injection surface.
  • Ingestion points: Data received through Ligolo-ng proxy sessions and dnscat2 command channels (SKILL.md).
  • Boundary markers: No explicit delimiters or instructions are provided to help the agent distinguish between tool output and instructions.
  • Capability inventory: The skill utilizes extensive shell execution capabilities and network configuration tools across multiple protocols (SKILL.md).
  • Sanitization: There is no evidence of data validation or sanitization for content received from external pivot agents.
  • [EXTERNAL_DOWNLOADS]: Provides guidance on utilizing various third-party security tools (Chisel, Ligolo-ng, iodine, dnscat2, Neo-reGeorg) that are not part of standard operating system installations.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 12:50 AM
Security Audit — agent-trust-hub — tunneling-and-pivoting