The Agent Skills Directory

[DATA_EXFILTRATION]: The skill is explicitly designed to identify and extract sensitive information from network traffic.
Targets include HTTP Basic Authentication headers (http.authorization), session cookies (http.cookie), and form POST data which typically contains usernames and passwords.
It instructs the agent to summarize and report discovered credentials, API keys, and session tokens to the user.
[PRIVILEGE_ESCALATION]: The instructions explicitly guide the agent to suggest or use sudo to overcome permission issues during live network captures.
Evidence: "Live capture may require root privileges. If permission denied, suggest running with sudo or using sudo tshark directly."
[COMMAND_EXECUTION]: The skill executes complex shell commands via tshark and calls external wrapper scripts (e.g., capture-http-credentials.sh) to perform analysis.
It utilizes the $ARGUMENTS variable directly in shell commands, which could lead to command injection if the user-provided interface name or file path is not properly sanitized by the platform.
[DYNAMIC_CONTEXT_INJECTION]: The skill uses the !command`` syntax to execute shell commands at load time for environment detection.
Evidence: !"test -f scripts/tshark/capture-http-credentials.sh && echo "YES" || echo "NO"" is used to determine which execution path the agent should take.

sniff