The Agent Skills Directory

[SAFE]: The skill consists of documentation and reference implementation patterns that follow standard software development best practices with no detected malicious patterns.- [DATA_EXPOSURE]: The provided code samples demonstrate the secure usage of environment variables (e.g., RESEND_API_KEY and REDIS_URL) to manage sensitive service credentials, preventing hardcoded secrets.- [PROMPT_INJECTION]: The skill involves the processing of external data within email templates, which constitutes a potential surface for indirect prompt injection.
Ingestion points: External data is passed via props to React Email components like WelcomeEmail and parameters in the sendEmail function within SKILL.md.
Boundary markers: The skill does not explicitly define custom delimiters for variable interpolation in templates.
Capability inventory: The skill utilizes network capabilities to send emails via the Resend API and perform DNS lookups, as well as database capabilities to manage user records and suppression lists.
Sanitization: The implementation relies on the default HTML escaping provided by the react-email library, which mitigates basic script injection but does not specifically target prompt-based manipulation of downstream agents.- [EXTERNAL_DOWNLOADS]: All external libraries mentioned (resend, @react-email/components, bullmq, ioredis) are well-known, legitimate packages hosted on the official npm registry.

email-systems