The Agent Skills Directory

[SAFE]: The file contains only markdown instructions and configuration metadata. There are no associated scripts, binaries, or active components that could perform unauthorized actions.
[NO_CODE]: No code execution capabilities are present in this skill. It functions as a reference document for an AI orchestrator and sub-agents.
[PROMPT_INJECTION]: The skill contains internal logic requiring the agent to perform specific tasks regardless of user input to skip them. These instructions reinforce a specific business workflow (verification reporting) and do not represent a malicious attempt to bypass model safety filters or extract sensitive system information.

verification-rules