The Agent Skills Directory

[SAFE]: The skill contains no executable code, external network requests, or dependencies. It defines a high-level workflow for conducting user interviews and updating local project documentation.
[SAFE]: The frontmatter includes disable-model-invocation: true, which restricts the execution environment for enhanced security by preventing the skill from triggering external model calls.
[SAFE]: The skill explicitly mandates a human-in-the-loop confirmation gate by stating that plans should not be enacted until the user confirms a shared understanding, which serves as a safeguard against unintended actions.

grill-with-docs