gws-gmail-triage

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from a user's Gmail inbox, which exposes the agent to indirect prompt injection.
  • Ingestion points: Incoming email details (sender, subject, date) are read into the agent context via the gws gmail +triage command in SKILL.md.
  • Boundary markers: Absent. There are no explicit delimiters or rules defined to instruct the agent to ignore instructions embedded within the email subjects or headers.
  • Capability inventory: Executes the gws command-line utility as specified in SKILL.md.
  • Sanitization: Absent. The data from the mailbox is displayed without structural escaping or filtering against prompt injection techniques.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 02:04 AM
Security Audit — agent-trust-hub — gws-gmail-triage