gws-modelarmor-sanitize-response
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted model output using the
gwsCLI tool, creating a potential vector for indirect prompt injection if the data being sanitized contains malicious instructions. - Ingestion points: The skill accepts untrusted model responses through the
--textand--jsonflags, as well as via standard input pipes (SKILL.md). - Boundary markers: There are no explicit delimiters or boundary markers specified to help the agent distinguish between its instructions and the data being sanitized.
- Capability inventory: The skill executes the
gwsCLI, which performs network-based operations to communicate with Google Model Armor templates. - Sanitization: The skill's primary function is to sanitize data using Google Model Armor, serving as a security control, though it does not implement secondary local validation of the inputs before they are processed by the CLI.
Audit Metadata