The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses system commands to configure network interfaces and manage kernel modules.
Evidence: In scripts/agent.py, the create_vlan_interface and cleanup_vlan_interfaces functions use subprocess.run to call modprobe 8021q and ip link for interface creation and deletion.
[EXTERNAL_DOWNLOADS]: The skill requires and references well-known external networking libraries and tools.
Evidence: The script scripts/agent.py depends on the scapy library. The documentation in SKILL.md identifies Yersinia and frogger as prerequisites for the assessment workflow.
[PROMPT_INJECTION]: The skill contains an attack surface for indirect prompt injection by ingesting data from the local network environment.
Ingestion points: The listen_for_dtp and listen_for_cdp_lldp functions in scripts/agent.py capture network packets using scapy.sniff().
Boundary markers: Absent. Captured data from packet headers is extracted and included directly in the report objects without delimiters or warnings.
Capability inventory: The script has capabilities for raw packet crafting, interface manipulation, and executing system commands via subprocess.run across multiple functions.
Sanitization: Absent. There is no validation or escaping of the data captured from the network before it is processed and displayed in the final assessment report.

performing-vlan-hopping-attack