Google Cloud

**Judgement:** A strong, opinionated reference skill for GCP practitioners and agents that need to reason about real-world production setups. It focuses on *rules and anti-patterns* rather than automation, making it best as a knowledge enhancer and checklist rather than a standalone deployment tool. **What it’s good for** - Teams already running or planning serious workloads on GCP who want embedded best practices in their agents. - Architecture and code reviews where the agent can flag likely cost traps (BigQuery scans, Cloud NAT pricing, idle disks/IPs), insecure defaults (default Editor service account, primitive IAM roles), or design flaws (lack of partitioning, poor VPC layout). - Building prompts or workflows where the agent must explain trade-offs between GCP options (e.g., gen1 vs gen2 Cloud Functions, global vs regional load balancers, GKE Autopilot limits). **Key risks / limitations** - **Static guidance:** This is a curated ruleset, not a live integration with GCP APIs; it cannot inspect your actual projects or verify current resource configurations. - **Potential staleness:** GCP evolves quickly; while the rules read as accurate and modern, they may drift over time (e.g., service limits, pricing nuances, new features) and should be periodically validated against current docs. - **Not for beginners:** The content assumes familiarity with GCP concepts (VPC, IAM, Cloud NAT, Shared VPC, etc.); as a result it’s more useful to intermediate/advanced users and agents than to novices. **Recommended scenarios** - Use as a **production sanity-check layer** in agents that design or critique GCP architectures, Terraform modules, or deployment plans. - Use as a **cost and security advisory skill** in internal tooling where agents suggest improvements to existing GCP projects. - Combine with other skills that can actually read or modify infrastructure (Terraform, CI/CD, or GCP API integrations) so this skill acts as a best-practices brain, not an executor.