The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documentation provides numerous shell commands for low-level wireless network manipulation, including disabling network processes (airmon-ng check kill), enabling monitor mode (airmon-ng start), and performing packet injection attacks.
[INDIRECT_PROMPT_INJECTION]: The skill instructions involve capturing and analyzing untrusted external data from wireless networks using tools like airodump-ng and tshark. This data, such as SSID names or probe requests, could contain malicious payloads designed to exploit the agent's processing logic.
Ingestion points: Network reconnaissance and capture commands (e.g., airodump-ng, tshark in SKILL.md).
Boundary markers: None specified for delimiting captured network strings from instructions.
Capability inventory: Extensive shell access and specialized security tool execution for network manipulation and password cracking.
Sanitization: No data validation or sanitization routines are described for processed network traffic.
[EXTERNAL_DOWNLOADS]: The skill includes a reference to a domain (book.hacktricks.xyz) that has been identified as malicious by automated scanners.
[METADATA_POISONING]: There is a logical inconsistency in the skill's documentation. The 'When NOT to Use' section explicitly claims the skill is 'Wi-Fi only' and refers users to a different skill for Bluetooth, yet the body of the skill contains a dedicated section for Bluetooth discovery and attacks.

attacking-wireless-networks