gws-gmail
Pass
Audited by Gen Agent Trust Hub on Jul 10, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted data from external email messages, creating a surface for indirect prompt injection.
- Ingestion points: Untrusted data enters the agent context through email reading and triaging commands (e.g.,
+read,+triage, and+watch). - Boundary markers: The provided documentation does not specify the use of delimiters or 'ignore' instructions to isolate processed email content from the agent's primary instructions.
- Capability inventory: The agent possesses high-impact capabilities including sending emails (
+send), replying to messages (+reply), forwarding content (+forward), and modifying mailbox settings. - Sanitization: No sanitization or filtering of email bodies or headers is documented to mitigate the execution of instructions embedded in received messages.
Audit Metadata