extension-email
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data (email addresses, usernames, and order references) to generate content for outgoing emails, creating a potential surface for indirect injection if these inputs are not sanitized.
- Ingestion points: The arguments for the
sendOrderConfirmationEmailfunction insrc/backend/main.mo. - Boundary markers: The skill does not implement delimiters or ignore-instructions for the interpolated data.
- Capability inventory: The skill uses the
EmailClient.sendServiceEmailtool to transmit the processed content externally. - Sanitization: No explicit HTML escaping or input validation is observed in the provided Motoko logic.
- [EXTERNAL_DOWNLOADS]: The skill utilizes the Mops package manager to include functionality from a vendor-controlled module.
- Dependency: The module
mo:caffeineai-email/emailClientis imported to facilitate email operations.
Audit Metadata