gws-shared
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill documents interactions with external Google Workspace APIs, creating a surface for indirect prompt injection where data retrieved from the API (such as document content or metadata) could contain malicious instructions for the agent.
- Ingestion points: Data returned by the
gwsCLI tool from various Google Workspace services (e.g., Sheets, Drive). - Boundary markers: The skill does not define specific prompt boundary markers, as it is a documentation reference for CLI syntax.
- Capability inventory: The
gwsbinary provides capabilities to read, write, upload, and delete resources across Google Workspace services. - Sanitization: The documentation explicitly recommends using the
--sanitizeflag to screen responses through Model Armor for PII and content safety.
Audit Metadata