auto-logging-email-interactions

Pass

Audited by Gen Agent Trust Hub on Jun 24, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection because it ingests and processes untrusted external data from Gmail to perform write operations in a CRM.
  • Ingestion points: The skill fetches recent sent and received emails from Gmail, which is untrusted external content.
  • Boundary markers: The instructions do not define boundary markers or delimiters to help the agent distinguish between the skill's instructions and the content of the emails being summarized.
  • Capability inventory: The skill utilizes search_contacts (read access) and interactions.log (write access) to modify the Evergreen CRM database.
  • Sanitization: There is no explicit sanitization, validation, or filtering of the email content before it is summarized and logged to the CRM.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 24, 2026, 12:27 AM
Security Audit — agent-trust-hub — auto-logging-email-interactions