The Agent Skills Directory

[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (HTML injection) because it interpolates untrusted data from form submissions directly into email templates. An attacker could provide malicious HTML or script tags in fields like 'name' or 'message' which would then be rendered by the recipient's email client.
Ingestion points: LeadNotificationData, ConfirmationData, and QuoteData fields in references/lead-notification.md, references/customer-confirmation.md, and references/quote-result.md.
Boundary markers: No delimiters or protective markers are used to isolate user data from the template structure.
Capability inventory: The skill utilizes the resend library to transmit generated HTML emails to external recipients.
Sanitization: No evidence of HTML escaping, validation, or sanitization of input data was found in the provided code snippets.
[EXTERNAL_DOWNLOADS]: The skill implementation references and requires the installation of the resend Node.js package to handle email delivery. This is a standard dependency for integrating with the Resend email service.

email-templates