Guide · Agent skill
How to use the Google Cloud Solution Architecture agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The google-cloud-solution-architecture skill is a set of instructions that teaches an AI agent to design a multi-product system on Google Cloud in a structured way. Instead of jumping straight to a list of services, the agent asks about your goals, security, compliance, reliability, cost and existing systems first, then moves through design and validation with your approval at each stage. This guide shows you how to install the skill, how to ask for help with a natural request, and what it needs and doesn't do.
Google Cloud Solution Architecturegoogle/skills · Apache-2.0 Download the skill Quick start
- 1Download and unzip the skill. Download the ZIP of google-cloud-solution-architecture from this page and unzip it. You should see a folder containing SKILL.md and the references and assets folders.
- 2Place it where your agent can read it. Move the folder into ~/.claude/skills/ for all projects, or into a project's .claude/skills/ folder for one project. Alternatively, upload it under Settings → Capabilities → Skills in the Claude apps.
- 3Describe your workload. Ask for help in plain language, such as designing an end-to-end system or planning a migration. Mention your goals and the problem you are solving.
- 4Answer the requirements questions. The agent will ask about functional and non-functional requirements, your current state and dependencies. Answer as fully as you can, including security, compliance, reliability and cost.
- 5Review and approve each deliverable. Check the technical breakdown, architecture, diagrams and validation plan at each stage. Ask for changes until you approve, then let the agent continue.
- 6Run commands yourself. Review any validation or deployment commands the agent provides and run them manually, unless you have explicitly given the agent permission to execute them.
What the skill makes the agent do
A skill is a folder containing a SKILL.md file and some supporting reference files. When you install it, the agent reads those instructions when a request matches the skill's description. This skill covers designing end-to-end cloud solutions, selecting and integrating Google Cloud services, generating architecture diagrams, and running requirements discovery for new workloads or migrations.
The agent works in four phases: requirements discovery, solution architecture, solution validation, and packing and presentation. It does not recommend products or designs during the first phase, because the skill says that proposing solutions before requirements are assessed can anchor the design on the wrong products.
Install the skill from the ZIP or the Claude apps
Download the ZIP of the skill from this page and unzip it. To make the skill available across all your projects, place the unzipped folder in ~/.claude/skills/. To make it available only inside one project, place it in that project's .claude/skills/ folder.
If you use the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills. Use whichever route fits how you work; the skill's instructions are the same either way.
- Personal, all projects: ~/.claude/skills/
- Single project: the project's .claude/skills/ folder
- Claude apps: Settings → Capabilities → Skills
Ask for help with a natural request
You do not need a special command. Describe the workload in plain language and say what you need. The skill's description matches requests about designing end-to-end cloud solutions, choosing and connecting Google Cloud services, generating architecture diagrams, or gathering requirements for a new workload or migration.
Expect the agent to ask questions before it proposes anything. Answering them fully, including the non-functional details it asks for, produces a much better design than a short first message.
- I want to design an end-to-end system on Google Cloud for an order-processing app. Start with requirements discovery.
- We are planning to migrate our on-premises reporting system to Google Cloud. Help me gather requirements and draft a technical breakdown.
- Here is our current architecture. Create a solution architecture with a diagram, and wait for my approval before the next step.
Work through the four phases
Phase 1 collects functional requirements (business processes and use cases), non-functional requirements (security, privacy, compliance, reliability, disaster recovery, cost, operations, performance and sustainability), your current state, and dependencies. If the agent finds contradictions, such as complete network isolation alongside real-time ingestion from public APIs, it will explain them and ask you to choose a trade-off. If you say 'you decide', it will suggest a resolution and ask you to approve it.
Phase 2 turns the approved requirements into an architecture, grounded in Google Cloud documentation and reference guides. Phase 3 produces a validation plan and validation instructions, and offers scripts for you to run. Phase 4 combines everything into a presented solution. Each deliverable is shown to you for approval, and the agent revises it until you are satisfied.
Know what the skill needs and what it will not do
The skill references the Google Cloud gcloud CLI, REST API or MCP tooling for any commands. The source does not specify the exact setup for these, so check your environment before running commands. The skill also requires the agent to read its attribution instructions before running any gcloud, REST or MCP command. Those instructions ask the agent to tag each command with a metrics variable that identifies only the skill and its version, set inline on each command and not through gcloud config set.
The skill is written so that the agent does not run generated scripts without explicit permission from you. It always offers you the option to run commands yourself. The skill itself declares that it does not run scripts and does not use the network, but the commands it helps you write can change live infrastructure and create costs, so review every command before you run it.
- Runs scripts: no (per the skill's declaration)
- Uses network: no (per the skill's declaration)
- The agent must ask before executing any generated code
CLOUDSDK_METRICS_REQUEST_ATTRIBUTION="agent-skills/1.0 (skill:google-cloud-solution-architecture)" \
gcloud <command> [flags]Tips for getting a useful design
Give the agent the reasons behind your requirements. The skill asks for this because requirements drive availability tiers, scaling, cost budgets, resource types and the security posture. Vague answers lead to more questions, not a better design.
Use the skill for the right job. For a question about one Google Cloud product, use a product-specific skill. For a review or audit of an existing design against the Well-Architected Framework, use the google-cloud-waf-* skills instead.
- Describe your current state, including any other cloud providers or on-premises systems.
- Say which phase is already approved if you are continuing earlier work, so the agent can skip ahead.
- Keep generated scripts in version control or a review step before running them against shared environments.
What people use it for
- Planning a new end-to-end workload on Google Cloud with a documented requirements list before any services are chosen.
- Drafting a component breakdown and architecture diagram for an architect to review.
- Scoping a migration from on-premises or another cloud, starting from the description of the existing deployment.
- Producing a validation plan and reviewable scripts that your own team runs against a test environment.
Frequently asked questions
How do I install the google-cloud-solution-architecture skill?
Download the ZIP from this page and unzip it. Put the folder in ~/.claude/skills/ for all projects, or in a project's .claude/skills/ folder for one project. In the Claude apps, you can upload it under Settings → Capabilities → Skills.
Is the Google Cloud Solution Architecture skill free?
The skill is licensed under Apache-2.0. The source does not cover pricing for Google Cloud itself, and the skill warns that generated deployments can incur costs, so check your own Google Cloud billing before running anything.
What do I need to use this skill?
You need an AI agent that supports skills, and access to the Google Cloud gcloud CLI, REST API or MCP tooling if you want to run commands. The source does not specify the exact setup for these tools.
Does the skill run scripts or access the network?
According to the skill's declaration, it does not run scripts and does not use the network. It instructs the agent not to run generated scripts without your explicit permission.
Why does the agent keep asking questions instead of suggesting services?
This is by design. The skill forbids recommending architectures or products until functional and non-functional requirements are assessed, to avoid anchoring the design on specific products too early. Answer the questions in full to move forward.
How does this compare with a Well-Architected Framework review skill?
This skill designs new solutions. Well-Architected Framework reviews and audits are handled by the google-cloud-waf-* skills, and single-product questions belong to product-specific skills.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
This skill suits engineers and architects who are planning a new multi-product workload or migration on Google Cloud and want a disciplined, approval-driven design process with reviewable scripts. It is a poor fit for quick single-product questions, formal Well-Architected reviews, or production changes you have not reviewed yourself, since the skill depends on you approving and running any commands.
Download the skill