model-product-abuse-risk

Installation
SKILL.md

Model Product Abuse Risk

Prefer the least harmful effective control. A high loss estimate does not justify collective punishment, secret confiscation, or inaccessible appeal. Legitimate recovery is first-class. Do not teach an attacker how to bypass the control. A score that cannot be appealed is not a control, it is an unaccountable taking.

Open abuse model and control ladder when actors, assets, or controls need depth. Open decision evidence and adaptive operations when scoring, appeals, or automation need depth.

Installs
2
Repository
sylphxai/skills
GitHub Stars
1
First Seen
Aug 22, 2026
model-product-abuse-risk — sylphxai/skills