The Agent Skills Directory

[SAFE]: The skill incorporates mandatory human-in-the-loop (HITL) confirmation steps for all sensitive operations, including sending emails and updating lead interest levels, ensuring the user retains control.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted lead responses which could contain malicious instructions.
Ingestion points: Lead emails are retrieved from the Instantly Unibox API (GET /emails) as defined in SKILL.md.
Boundary markers: There are no explicit instructions to the model to ignore embedded commands within the lead's email text.
Capability inventory: The agent has the ability to send email replies via POST /emails/reply and modify lead records via POST /leads/update-interest-status.
Sanitization: No automated sanitization of the input email body is performed before the model processes it.
Risk Mitigation: The potential for indirect prompt injection is mitigated by the requirement that every drafted reply must be manually reviewed and confirmed by the user before execution.

inbox-reply