Screen your prompts with Google Model Armor.
Screens a user prompt through a Google Model Armor template.
When to use it
Use to sanitize user prompts; outbound response safety uses a different command.
Give it a template resource name and prompt content; it returns the result of sanitizing the prompt.
What you provide
Requires the gws CLI to invoke Model Armor prompt sanitization.
Requires reading its authentication, global flag, and security instructions before proceeding; directs generating it if missing.
This skill
Google Model Armor
Sends prompts to Model Armor
../gws-shared/SKILL.md
Creates missing shared skill instructions