agent-platform-endpoint-management
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- Standard Command-Line Tooling: The skill utilizes the
gcloudCLI, which is the standard tool for managing Google Cloud resources. The commands provided (create,list,update,delete) are consistent with the skill's stated purpose of managing serving endpoints. - Confirmation Tiers and Safety Protocols: A significant portion of the instructions is dedicated to safety tiers (Tier R, M, D). These protocols ensure that the AI agent seeks explicit user confirmation before performing actions that are mutating or destructive, which is a security best practice to prevent unintended infrastructure changes.
- Interactive Setup and Authentication: The skill guides the user through standard authentication processes (
gcloud auth login) and requires project-specific configuration. These are necessary steps for the intended functionality and follow standard operating procedures for cloud management. - No External Dependencies or Remote Execution: The skill does not perform any external downloads, execute scripts from remote sources, or install third-party packages, limiting the risk of supply chain attacks or unauthorized code execution.
Audit Metadata