The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses strong instructional language to define its operation (e.g., 'Never state it for them'), but these are intended behavioral constraints for the grounding process and do not attempt to bypass safety filters.
[DATA_EXFILTRATION]: The skill handles user statements and external notes but explicitly disclaims any network, CLI, or database write operations. No exfiltration patterns or sensitive path accesses were detected.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to process potentially untrusted external data (sources and user input). It addresses this risk by providing a robust framework for verification, using techniques like the 'Gate' and 'Verification' to check statements against actual text before confirmation.
[COMMAND_EXECUTION]: No shell commands, executable scripts, or tool invocations are present. The skill relies entirely on natural language instructions for the AI agent.

grounding