The Agent Skills Directory

[DYNAMIC_CONTEXT_INJECTION]: The skill uses !command`` syntax to execute shell commands (command -v nc and nc -h) automatically when the skill is loaded to check for tool availability and version information.\n- [COMMAND_EXECUTION]: The instructions explicitly describe how to establish reverse shells using various Netcat implementations, including the use of the -e flag and mkfifo pipes to redirect shell access over the network.\n- [DATA_EXFILTRATION]: The skill provides patterns for transferring files and entire directories to remote targets using nc and tar pipes, which can be utilized to move sensitive data out of the environment.\n- [INDIRECT_PROMPT_INJECTION]: The skill exposes an attack surface by interpolating user-provided target and port arguments directly into shell commands. While it mentions validation against a scope file, the documentation lacks explicit sanitization patterns for these external inputs.

netcat