hubspot
Pass
Audited by Gen Agent Trust Hub on Jul 15, 2026
Risk Level: SAFE
Full Analysis
- Human-in-the-Loop Write Verification: The skill explicitly requires the agent to present a field-level draft of proposed changes to the user and obtain explicit approval before executing any write operations via the CRM connector. This prevents unauthorized or accidental modifications to CRM data.
- Property and Schema Validation: The instructions mandate confirming property labels and internal names before querying or writing. This ensures that the agent interacts with the CRM schema accurately and avoids errors or unexpected data alignment issues.
- Data Ingestion Considerations: The skill processes external data from HubSpot records. To manage risks associated with processing external content, the workflow incorporates several checkpoints, including bounded candidate lookups and mandatory review phases for any derived actions.
- Controlled Connector Boundaries: The skill is restricted to specific, identified HubSpot tool surfaces for identity, property discovery, and record management, ensuring it operates within a defined and expected scope of capability.
Audit Metadata