The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill allows the agent to subscribe to and process live data from Google Workspace events, such as Chat messages and task updates. This creates a vulnerability where instructions hidden in external data could be executed by the agent.
Ingestion points: The skill utilizes gws events message stream and gws events tasks subscribe to bring external, potentially untrusted content into the agent's context.
Boundary markers: The instructions do not define any delimiters or system-level constraints to differentiate between user instructions and the content of the event stream.
Capability inventory: The agent uses the gws binary to manage Workspace resources, list subscriptions, and stream messages.
Sanitization: There is no evidence of sanitization, filtering, or validation of the event data before it is processed by the agent.

gws-events