The Agent Skills Directory

[SAFE]: No security issues or malicious patterns detected. The skill is composed of instructional markdown and reference files that define a workflow for analyzing and rewriting prompts. It does not perform sensitive file operations, network requests, or code execution.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user-supplied prompts as its primary function, which creates an inherent attack surface for indirect prompt injection. However, the risk is mitigated because the skill lacks executable capabilities (such as shell access or network tools) that could be exploited. Furthermore, the skill's core function—dissolving personas into explicit capabilities—is a defensive engineering practice that helps prevent common prompt injection techniques.

chisel