gws-events
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill relies on the
gwscommand-line utility to perform operations against the Google Workspace Events API, including resource inspection withgws schemaand event management usinggws events. - [INDIRECT_PROMPT_INJECTION]: The skill ingests external event data from Google Workspace, which serves as a potential surface for indirect instructions if the event content contains malicious text.
- Ingestion points: Workspace event streams fetched via
gws events message streamandgws events tasks subscribeas specified in SKILL.md. - Boundary markers: None present; the instructions do not define specific delimiters to isolate external event content from agent instructions.
- Capability inventory: The skill includes capabilities to create, patch, reactivate, and delete Workspace event subscriptions.
- Sanitization: No explicit data validation or sanitization of incoming event content is described in the skill configuration.
Audit Metadata