secondary-style-perturbation

This skill is internally coherent and low-privilege: it only prescribes local text transformations and requests no credentials or installs. Before using it, consider ethical and practical risks: the transforms are suited to evading style or authorship detectors and may unintentionally alter nuance or introduce factual ambiguity despite saying not to add information. If you plan to use this on sensitive, attributed, or legally protected text, review the output manually, ensure you have the right to modify the text, and keep audit logs of original/modified versions. If you need stronger guarantees that meaning isn’t changed, test the skill on representative samples and refuse use when high fidelity is required.