agent-platform-prompt-management
Pass
Audited by Gen Agent Trust Hub on Jun 13, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides Python code snippets that utilize official Google Cloud SDKs (
google-cloud-aiplatform,google-genai) for interacting with Vertex AI services. - [SAFE]: The instructions implement a tiered security model (Tier M for Mutating, Tier D for Destructive), explicitly requiring the agent to obtain human confirmation before providing code for creating or deleting resources.
- [SAFE]: Authentication is handled through standard Google Cloud Application Default Credentials (ADC) using official
gcloudCLI commands, avoiding hardcoded secrets or unsafe credential handling. - [SAFE]: The skill correctly uses placeholders for sensitive identifiers (Project ID, Location, Prompt ID), instructing the agent to populate these based on user context rather than using hardcoded values.
Audit Metadata