The Agent Skills Directory

[SAFE]: The skill functions as an instructional persona for designing workflow specifications through interactive dialogue.
[SAFE]: Workspace access is strictly scoped to reading and writing local text files (NOTES.md and workflows/*.md), which is appropriate for its role as a documentation and planning tool.
[SAFE]: The skill configuration explicitly disables model tool invocation (disable-model-invocation: true), which limits the agent's capabilities to text processing and workspace file management, minimizing potential attack surfaces.

loop-me