gws-modelarmor
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes the
gwscommand-line tool to perform content filtering and schema discovery. This is the expected behavior for a CLI-integrated skill from this vendor. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to process and sanitize user-generated content, which is a natural ingestion point for untrusted data.
- Ingestion points: User-generated content is passed to the
gws modelarmorcommand via resource methods and flags as described inSKILL.md. - Boundary markers: The file refers to a shared skill (
../gws-shared/SKILL.md) for global security rules and auth, though explicit delimiters are not defined within this specific file. - Capability inventory: The skill performs shell execution of the
gwsbinary to communicate with the Model Armor API. - Sanitization: The primary function of this skill is to provide sanitization (
+sanitize-prompt,+sanitize-response) for model interactions.
Audit Metadata