What it does
This skill helps design enterprise Gemini platform architectures and organize implementation plans and post-deployment checks. It covers project layouts with Host, Governance Hub, Runtime Spoke, and Tool/MCP Spoke; communication patterns using Agent Gateway and Agent Connectivity Template; and outbound traffic through Secure Web Proxy or Policy-Based Routing. It also addresses Private CA connections, token exchange with external identity providers, Unified Access Policy, Semantic Governance Policy, Model Armor, ADK agent identity on Agent Runtime, registry synchronization, and day-to-day operations. The result can include an architecture proposal, CLI commands, configuration files, and checks using audit logs.
When it helps
Use it when planning a Gemini-based agent platform across multiple projects or networks. It is useful for teams that need to consider network paths, identity boundaries, and safety policies together, and organize audit and operations steps after design.
Example requests
- “Draft a multi-project layout and network plan for our agents.”
- “Design communication patterns and safety policies using Agent Gateway.”
- “List the audit logs and checks we should review after deployment.”
Notes
Google Cloud SDK, authentication, required APIs, billing setup, and appropriate IAM permissions are prerequisites. This skill does not cover general IAM troubleshooting or read-only log analysis and incident diagnosis. Explicit confirmation is required before granting permissions, changing organization policies or access controls, or deleting resources.
This overview was written by AI (OpenAI Codex) from the public SKILL.md. Check the original below for the exact behavior.