agent-platform-skill-registry
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFE
Full Analysis
- Standard Cloud Authentication: The skill utilizes official Google Cloud authentication mechanisms. Users are prompted to authenticate using
gcloud auth application-default login, and the internal scripts usegoogle-authto retrieve access tokens, ensuring operations are performed within the user's own project context. - Official Service Interaction: Network operations are directed towards
aiplatform.googleapis.comendpoints. These are well-known, official service domains used to manage agent platform resources. The skill constructs these URLs using user-provided project and location identifiers. - Controlled File Operations: The lifecycle management scripts include functionality to package local files or folders into a base64-encoded zip payload for upload. This behavior is transparent and aligned with the stated purpose of uploading skills to a remote registry.
- Instruction Scaffolding and Human Review: The
Generate Skillcapability involves the agent creating new instruction sets based on user requirements. The skill incorporates best practices for this process, specifically directing the agent to validate generated markdown and explicitly requiring a human review of the output before it is finalized.
Audit Metadata