| name | google-cloud |
| description | Plan, build, secure, and operate Google Cloud workloads. Triggers: `gcloud`, AlloyDB, BigQuery / BQML, Cloud SQL, Cloud Run (services, jobs, worker pools), Firebase, GKE, Vertex AI / Agent Platform / Gemini API, Google AIP API conventions, Well-Architected Framework (Cost, Reliability, Security), VPC Flow Logs, IAM / ADC / service identities, `*.googleapis.com`, `gke-gcloud-auth-plugin`. Skip when: working on AWS, Azure, or cloud-agnostic design with no GCP-specific surface, or pure prompt-engineering for a Gemini model with no GCP infra. |
| license | AGPL-3.0-only |
| metadata | {"author":"marcodsl","tags":"gcp, google-cloud, alloydb, bigquery, cloud-run, cloud-sql, firebase, gke, gemini, vertex-ai, agent-platform, iam, networking, waf, well-architected, aip, api-design","version":3} |
Google Cloud
Umbrella skill covering Google Cloud services, the Gemini API on Agent Platform, Google AIP API conventions, and the Well-Architected Framework. Pick the branch that matches the task; load multiple branches when work spans services.
- Designing, configuring, or troubleshooting Google Cloud workloads.
- Building on the Gemini API or Agent Platform (Vertex AI).
- Adopting Google AIP API conventions or the Well-Architected Framework.
<do_not_use_when>
- Working with other clouds (AWS, Azure) or generic vendor-neutral concepts.
- The task is pure prompt-engineering for a Gemini model unrelated to GCP infrastructure.
</do_not_use_when>
<governing_rule>
Route to the branch that matches the task, load multiple branches when work spans services, and let each branch's intra-document links carry the depth.
</governing_rule>
Pick the branch(es) matching the task and read the referenced file:
- AlloyDB for PostgreSQL (clusters, instances, backups, MCP server) →
references/alloydb-basics.md
- BigQuery (datasets, tables, jobs, SQL, BQML, Gemini-in-BigQuery) →
references/bigquery-basics.md
- Cloud SQL (MySQL, PostgreSQL, SQL Server) →
references/cloud-sql-basics.md
- Cloud Run (services, jobs, worker pools) →
references/cloud-run-basics.md
- Firebase (mobile and web app development) →
references/firebase-basics.md
- GKE (clusters, networking, security, scaling, observability, AI/ML inference, cost, upgrades) →
references/gke-basics.md
- Gemini API on Agent Platform / Vertex AI (SDKs, Live API, tools, multimedia, caching, batch, tuning) →
references/gemini-api.md
- Google AIP API design (resource models, standard vs custom methods, field behavior, versioning) →
references/google-aip-adoption.md
- IAM and authentication (principals, ADC, service identities, secure access) →
references/google-cloud-recipe-auth.md
- Networking observability (VPC Flow Logs, NAT, firewall logs, Connectivity Tests, latency metrics) →
references/google-cloud-recipe-networking-observability.md
- First steps on Google Cloud (account, billing, projects, first deploy) →
references/google-cloud-recipe-onboarding.md
- Well-Architected Framework: Cost Optimization →
references/google-cloud-waf-cost-optimization.md
- Well-Architected Framework: Reliability →
references/google-cloud-waf-reliability.md
- Well-Architected Framework: Security →
references/google-cloud-waf-security.md
Each branch carries deeper references under references/<branch>/ (core concepts, CLI, client libraries, MCP usage, IaC, IAM/security, service-specific deep dives). Follow the branch's intra-document links when more depth is needed.
Most workflows assume:
- Authenticated `gcloud` CLI with an active project. If unset, read `references/google-cloud-recipe-auth.md` and `references/google-cloud-recipe-onboarding.md`.
- The relevant API enabled via `gcloud services enable `.
- Permissions appropriate for the action; per-service IAM detail lives in each branch's `references//iam-security.md`.
