The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill makes extensive use of the sudo command to perform administrative tasks such as changing network interface modes, terminating system processes, and enabling monitor mode across various wireless utilities.
[REMOTE_CODE_EXECUTION]: The instructions direct the user or agent to download and execute scripts from external GitHub repositories belonging to individual researchers rather than established organizations:
https://github.com/s0lst1c3/eaphammer.git is cloned and run via ./eaphammer.
https://github.com/vanhoefm/dragonslayer.git is cloned and run via ./dragonslayer.py.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its processing of untrusted wireless environment data.
Ingestion points: Network traffic is captured and analyzed using airodump-ng, kismet, tshark, and bettercap throughout the SKILL.md file.
Boundary markers: The skill does not provide instructions to establish delimiters or to ignore potential instructions embedded within the captured wireless frames.
Capability inventory: The environment provides high-level system execution capabilities, including sudo access for multiple networking and security tools.
Sanitization: There is no mention of sanitization or filtering applied to the captured data before it is processed by the diagnostic tools.

wireless-security-wifi-pentest