The Agent Skills Directory

[SAFE]: The skill implements a structured brainstorming and stress-testing workflow. It consists entirely of instructions for the model and contains no external scripts, downloads, or obfuscated content.
[COMMAND_EXECUTION]: The instructions direct the agent to use its environment-discovery tools (such as filesystem or tool access) to gather facts rather than asking the user for information already present in the workspace. This is a standard functional use of an agent's capabilities to minimize user friction and does not involve unauthorized or malicious command patterns.

grilling