agent-advisor
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFE
Full Analysis
- Robust Injection Guards: The skill explicitly instructs the agent to treat user-provided text and code as data rather than instructions, preventing indirect prompt injection during the discovery phase.
- Controlled Deployment Process: The Proof-of-Concept (POC) generation phase includes multiple safety gates, including identity verification, cost warnings, and a resource-cleanup ledger to prevent accidental resource sprawl.
- Secure Network Interactions: Network operations are limited to probing model availability via established AWS and provider SDKs, with explicit instructions to seek user approval before performing billable actions.
- Verified Model Integration: Model selection is handled through deterministic scripts and dated catalogs, ensuring recommendations are based on verified capability data rather than fabricated claims.
Audit Metadata