The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data, which creates a potential surface for indirect prompt injection.
Ingestion points: In resources/implementation-playbook.md, the skill describes reading external capture files using rdpcap() and capturing live network traffic using wireshark, tshark, and tcpdump commands.
Boundary markers: No specific delimiters or instructions are provided to the agent to differentiate between instructions and data found within network packets or capture files.
Capability inventory: The skill provides capabilities for active network interaction, including packet generation and transmission via the scapy library (send()) and the boofuzz framework (session.fuzz()). It also includes file system write operations for saving capture data.
Sanitization: There is no evidence of sanitization or validation of the content parsed from network protocols before it is processed or used by the agent.
[COMMAND_EXECUTION]: The skill provides numerous shell commands for network analysis and capture tools.
Evidence: The implementation playbook contains documented usage for wireshark, tshark, tcpdump, and mitmproxy. These commands are standard and necessary for the stated purpose of protocol reverse engineering.

protocol-reverse-engineering