sf-service-models-api
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides defensive coding patterns for Salesforce Apex. It includes explicit warnings against using
eval, SOQL injection, or HTML injection with model-generated content. - [SAFE]: The skill emphasizes the use of the Einstein Trust Layer for PII masking and provides examples of local redaction using regex as a defense-in-depth measure.
- [SAFE]: The code patterns use standard Salesforce
aiplatform.ModelsAPIclasses. There are no external dependencies, remote script executions, or unauthorized network operations. - [SAFE]: Input handling includes recommendations for deterministic context capping and failure-closed validation strategies to prevent model hallucinations from affecting record data integrity.
Audit Metadata