The Agent Skills Directory

[SAFE]: The skill's primary function is the refinement of text prompts. Analysis of the instructions and metadata revealed no evidence of malicious intent or hidden behaviors.
[COMMAND_EXECUTION]: The SPEC.md file contains documentation for developer-oriented testing and validation commands using a local script (distiller.py). These commands are for development workflows and are not executed by the agent during normal runtime interaction with users.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied prompts as input. While this represents a theoretical ingestion surface, the risk is mitigated by strict rules that limit the agent's output to markdown-formatted refined prompts and require explicit user consent before any filesystem operations occur.

refine-prompt