analyze-account-signals
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFE
Full Analysis
- [Controlled Data Ingestion]: The skill is designed to synthesize information from a variety of enterprise sources such as CRM systems, document repositories, and communication tools. This ingestion is moderated by a preflight configuration that maps source categories to specific, authorized connectors.
- [Evidence and Interpretation Separation]: To maintain accuracy and transparency, the instructions mandate that raw evidence from external sources must be kept distinct from the AI's strategic interpretations. This practice helps ensure that users can verify claims against the original source data.
- [Human-in-the-Loop Authorization]: The skill includes explicit instructions to never execute write-back operations (such as CRM updates or posting messages) without a direct user request or approval of a reviewed draft. This prevents autonomous or unintended modifications to external systems.
- [Scoped Monitoring]: Portfolio monitoring is restricted to bounded sets of accounts (e.g., primary-owner scoped or calendar-linked), which prevents the skill from performing broad, unconstrained data retrieval across an entire organization's CRM.
Audit Metadata