The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous shell commands for wireless reconnaissance and exploitation, using tools like airodump-ng, hcxdumptool, reaver, and bettercap.
[EXTERNAL_DOWNLOADS]: Instructions include downloading and installing external security tools from GitHub, such as eaphammer and hostapd-wpe, as well as installing various packages via apt.
[PRIVILEGE_ESCALATION]: Extensive use of sudo is documented throughout the skill. This is necessary for wireless tools to access raw network interfaces and enable monitor mode.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the physical RF environment (SSIDs, MAC addresses, probe requests) and external files (PCAPs, logs). This represents an attack surface where malicious input from the environment could influence the agent's interpretation during a survey.
Ingestion points: RF capture data, SSID strings, and log files (hostapd-wpe.log, captured_psk.txt) processed by the agent.
Boundary markers: The skill uses markdown headers and code blocks to separate instructions, but lacks explicit guardrails against processing malicious strings within survey data.
Capability inventory: Significant capability to execute shell commands with sudo and write files to the local system.
Sanitization: No specific sanitization or validation of SSID names or MAC addresses is provided before they are used in command-line examples.

wireless-technique