resend-inbound-emails

Pass

Audited by Gen Agent Trust Hub on Mar 8, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill implements an AI-powered email personalization feature that uses inbound email content as context for generating replies, creating a surface for indirect prompt injection.
  • Ingestion points: Inbound emails are processed in assets/api/webhook-resend-route.ts where text and HTML bodies are extracted.
  • Boundary markers: The system uses assets/lib/inbox/reply-parser.ts to strip quoted content, which provides some structural separation, but lacks explicit delimiters or instructions to the AI to ignore embedded commands within the processed email body.
  • Capability inventory: The skill has the capability to send outbound emails via the Resend API (assets/api/inbox-send-route.ts).
  • Sanitization: Content is escaped for HTML notifications in assets/lib/inbox/inbound-notification.ts, though the raw content is utilized as context for the personalization engine.
  • [EXTERNAL_DOWNLOADS]: The skill interacts with the Resend API for email operations and domain management.
  • Fetches and manages configuration from Resend's official API endpoints using the resend library.
  • This interaction is necessary for the core functionality of the skill and uses well-known technology services.
  • [EXTERNAL_DOWNLOADS]: The skill optionally utilizes Redis for real-time Server-Sent Events (SSE).
  • Connects to a Redis instance for pub/sub operations to update the frontend instantly when new messages arrive.
  • Documented as an optional feature requiring the standard ioredis package.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 8, 2026, 05:02 PM