The Agent Skills Directory

[SAFE]: The skill is composed exclusively of documentation files and contains no executable scripts, binaries, or automated installation processes. All instructions are aimed at human-in-the-loop refactoring tasks.
[SAFE]: The skill provides numerous references to external technical papers, industry talks (InfoQ, YouTube), and established software engineering blogs (Simon Willison, Jacob Kaplan-Moss). These are clearly labeled as informational resources and do not involve dynamic code loading or data exfiltration.
[PROMPT_INJECTION]: The skill includes strong behavioral constraints, such as 'Bias toward deletion' and requirements to load specific mindsets before proceeding. These instructions are functional design principles for the skill's intended refactoring purpose and do not attempt to override or bypass the agent's primary safety guardrails.
[INDIRECT_PROMPT_INJECTION]: The skill operates on user-provided source code, which constitutes a potential injection surface. However, the skill explicitly mandates manual activation and user request, ensuring that all code modifications are subject to user review, which effectively mitigates the risk of automated malicious instruction execution.

reducing-entropy