The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external and potentially untrusted binary data (pcap files or live traffic) using protocol dissectors, creating a surface for indirect injection if a tool vulnerability is exploited by crafted traffic.
Ingestion points: Steps 1, 2, and 4 involve reading data from external sources like cap.pcap or live network interfaces via tshark and Scapy.
Boundary markers: The instructions do not define specific data delimiters, but Step 5 provides strong operational boundaries, recommending physical and logical network isolation (sandbox) and strictly authorized testing for all analysis activities.
Capability inventory: The skill utilizes shell-based tools (tshark) and Python scripts (scapy) to parse and reconstruct protocol data, involving standard file system and network analysis capabilities.
Sanitization: There is no explicit sanitization of the input traffic data, as the primary purpose of the skill is the raw analysis of protocol structures and payloads.

re-ics