The Agent Skills Directory

[SAFE]: The skill is purely instructional and descriptive. It guides an agent through a logical process of exploring repository changes, interviewing a user for risk factors, and formatting a markdown document based on a template.
[NO_CODE]: No scripts, executables, or command-line operations are included in the skill content.
[SAFE]: The frontmatter uses disable-model-invocation: true, which is a platform-level safety constraint that prevents the skill from autonomously calling other models or tools without explicit instruction, further limiting its risk profile.
[SAFE]: While it references a grilling skill and a run-qc skill, these are internal organizational pointers and do not constitute external code execution or unverifiable dependencies.

to-qc