The Agent Skills Directory

Indirect Prompt Injection Considerations: The skill is designed to process external market intelligence and contact data from ZoomInfo. As with any skill that ingests external content to generate summaries or narrative research, there is a theoretical surface for indirect prompt injection if the source data contains adversarial instructions. The skill mitigates this by emphasizing the use of structured identifiers (IDs) and canonical identity resolution before performing research tasks.
Modular Skill Loading: The instructions specify loading and executing logic from a relative local path (../user-context/SKILL.md). This modular dependency is a standard practice for complex agent configurations, allowing for shared context across different functional components.
Data Integrity and Privacy: The skill includes explicit guardrails against guessing sensitive professional details like email addresses or phone numbers. It instructs the agent to rely solely on connector-surfaced data, which aligns with best practices for data accuracy and privacy.

zoominfo