The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The script scripts/agent.py logs the full execution command, including cleartext passwords and NT hashes, to standard output using the print function in run_nxc. This behavior exposes sensitive credentials in any console logs or output buffers captured during the agent's operation.
[EXTERNAL_DOWNLOADS]: The SKILL.md file recommends installing the NetExec tool directly from a third-party GitHub repository (https://github.com/Pennyw0rth/NetExec) using pipx. This creates a supply chain dependency on an external source that is not an official package registry or a pre-verified trusted vendor.
[COMMAND_EXECUTION]: The script scripts/agent.py uses subprocess.run to execute local shell commands. While it uses an argument list to prevent direct shell injection, it serves as a broad execution wrapper for nxc, a tool specifically designed for remote authentication, credential harvesting, and code execution, thereby increasing the local attack surface.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection surfaces (Category 8) due to its interaction with untrusted remote network data.
Ingestion points: The scripts/agent.py file ingests and parses the stdout of the nxc command using the PWNED_RE regular expression to identify hostnames, IPs, and credentials.
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are present when processing the tool's output.
Capability inventory: The skill possesses powerful capabilities including remote command execution and credential dumping across multiple protocols (SMB, WinRM, MSSQL).
Sanitization: The script does not perform sanitization or validation of the hostnames or status strings received from remote targets before incorporating them into the structured JSON report.

moving-laterally-with-netexec