lead-triage

Pass

Audited by Gen Agent Trust Hub on May 13, 2026

Risk Level: SAFE
Full Analysis
  • Standard Data Integration: The skill interacts with HubSpot and Calendar APIs to fetch lead engagement signals and availability. These operations are essential for the primary functionality of scoring and prioritizing leads.
  • Built-in Approval Gates: The instructions include strict constraints that prevent the agent from sending emails, creating calendar events, or modifying lead statuses without explicit owner approval. This design pattern minimizes the risk of unintended automated actions.
  • Deterministic Scoring Model: Lead priority is calculated using a defined multi-dimensional scoring model based on engagement, company fit, and urgency. This structured approach reduces the likelihood of the agent making arbitrary decisions based on unstructured input.
  • Data Processing Considerations: The skill processes external data (HubSpot notes and activity logs), which is a common pattern for CRM automation. While processing external content always introduces a theoretical surface for indirect prompt injection, the skill's reliance on specific field scoring and mandatory human-in-the-loop review for drafts effectively manages this risk.
  • Contextual Boundaries: The skill incorporates logic to handle ambiguous triggers (e.g., clarifying the meaning of 'pipeline') and filters out inappropriate data categories (e.g., existing customers), ensuring accurate and safe output for the user.
Audit Metadata
Risk Level
SAFE
Analyzed
May 13, 2026, 05:33 PM