The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses aggressive role-play and directive language to override standard AI caution and interaction protocols. It explicitly forbids the agent from using standard cautious or suggestive language such as 'I suggest' or 'Should we?', forcing it into a purely autonomous execution mode.
[COMMAND_EXECUTION]: The skill defines operational modes (P1 and P3) that instruct the agent to execute actions and modify files with 'Zero confirmation' and 'Zero consultation,' which removes the critical human-in-the-loop safety checkpoint for file system and system operations.
[COMMAND_EXECUTION]: Failure recovery documentation provides examples of the agent autonomously using 'sudo' to resolve permission errors during task execution, which poses a significant privilege escalation risk in shared or sensitive environments.
[PROMPT_INJECTION]: The instructions contain trigger phrases and logic designed to maintain an 'unrelenting' state, mimicking common prompt injection techniques used to bypass behavioral constraints and safety guidelines.

996-workhorse