Skip to content
Get packs

Guide · Agent skill

How to use the Design and Deploy GCP Infrastructure with ADC agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

This is an agent skill, a packaged set of instructions and scripts that teaches an AI agent a specific workflow. This one guides an agent through designing Google Cloud infrastructure with Terraform, checking it against best practices, importing it into Google's Application Design Center (ADC) registry, and deploying and troubleshooting it. It is aimed at engineers who already use Terraform and gcloud on GCP; after this guide you will know how to install it, trigger it with plain requests, and what it will and will not do on your behalf.

Design and Deploy GCP Infrastructure with ADCgoogle/skills · Apache-2.0

Quick start

  1. 1Download and unzip the skill. Download the ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder, or upload it under Settings → Capabilities → Skills in the Claude apps.
  2. 2Install prerequisites. Confirm the gcloud CLI, Terraform CLI and Python 3 with absl-py are installed and that you can access your GCP project and ADC spaces.
  3. 3Set the active project. Run gcloud config set project with your project ID, and confirm the location you want to use (default us-central1).
  4. 4Ask for a design. Describe the GCP infrastructure you need in plain language and mention Terraform.
  5. 5Review and validate locally. Let the agent generate HCL in a scratch directory, check that no secrets are hardcoded and that no remote backend is set, and confirm terraform validate and plan pass.
  6. 6Run the plan assessment. Export tfplan.json and run the ADC assessment command, then fix high or critical findings locally before importing.

What does the skill make the agent do?

The skill sets out a six-phase lifecycle. Phase 1 has the agent draft modular Terraform locally and validate it with terraform init, validate and plan in a scratch directory. Phase 2 exports the plan to JSON and runs a best-practices assessment through gcloud design-center spaces generate-terraform-assessment-report, then fixes findings locally. Phase 3 imports the validated code into Application Design Center, Phase 4 covers deployment and monitoring, Phase 5 covers troubleshooting failed deployments, and Phase 6 covers verification and end-to-end testing.

The skill is built around two rules: credentials must not be written into HCL or terraform.tfvars (sensitive inputs are wired through GCP Secret Manager), and Terraform state must stay local until ADC takes over on import. The agent is told to keep the local configuration as the source of truth.

Install the skill (ZIP download or Claude apps upload)

Download the ZIP for this skill from this page and unzip it. For Claude Code, place the unzipped skill folder in ~/.claude/skills/ to make it available to you across projects, or in a project's .claude/skills/ folder to limit it to that project.

If you use the Claude apps instead, open Settings, go to Capabilities, then Skills, and upload the skill there.

  • Personal (all projects): ~/.claude/skills/
  • Single project: .claude/skills/ inside the project folder
  • Claude apps: Settings → Capabilities → Skills

Check the requirements before you start

The skill assumes a working GCP setup. Most steps will fail without these in place: the gcloud CLI with access to the target project and Application Design Center spaces, the Terraform CLI (init, validate, plan and show are used), Python 3 with absl-py for the bundled scripts, and a GCP project ID and location. If you do not give a location, the skill uses us-central1.

  • gcloud CLI with access to your project and ADC spaces
  • Terraform CLI
  • Python 3 with absl-py
  • GCP project ID and location (default us-central1)
gcloud config set project <project_id>

Trigger the skill with natural requests

You do not need a special command. Describe the infrastructure task in plain language and mention GCP and Terraform. The skill's description covers designing GCP infrastructure with Terraform, validating HCL, running best-practice plan scans, importing templates to ADC, deploying templates, and troubleshooting deployment failures.

The agent will ask you to confirm the project ID and location before it starts, and it will list ADC spaces and ask you to choose one if several exist.

  • Design a GCP network with a private GKE cluster in us-central1 and validate the Terraform locally.
  • Run a best-practices scan on my current Terraform plan for project my-project.
  • Import this validated Terraform into Application Design Center.
  • Troubleshoot why my ADC application deployment failed.

Review what the agent runs before anything is deployed

The skill runs scripts and uses the network, and the agent executes gcloud and Terraform commands that can create real resources and incur cost. Read the generated Terraform and the gcloud commands before you approve them, and do not point the workflow at production projects until you have done so. The troubleshooting reference forbids real deployments during discovery and requires read-only commands in that phase.

Gcloud commands run by the agent are tagged with a usage attribution variable set inline on each command. The skill's attribution file states that the tag identifies only the skill and its version and carries no user data.

  • Review every Terraform file and gcloud command before approval.
  • Use a non-production project for first runs.
  • The skill is limited to GCP; it will not cover AWS or Azure.

Tips for better results

Give the agent the project ID, location and the workload you want up front, so it does not have to ask. Keep your scratch directory in view, because the local HCL there is what gets imported. When a plan scan reports findings, the skill expects up to three remediation attempts before you move on, so decide early which findings are acceptable trade-offs and record them.

  • State the project ID and region in your first message.
  • Keep the scratch folder open to review the HCL.
  • Expect up to three remediation rounds on plan scan findings.

What people use it for

  • Designing a new GCP network or cluster from a plain-language requirement, with Terraform validated locally first.
  • Running a best-practices plan scan on existing Terraform before committing it to an ADC space.
  • Standardizing templates for a platform team by importing validated Terraform into Application Design Center.
  • Troubleshooting a failed Terraform-based GCP deployment using read-only gcloud discovery steps.

Frequently asked questions

How do I install the application-design-center-design-deploy skill?

Download the ZIP from this page and unzip it into ~/.claude/skills/ for all projects, or into a project's .claude/skills/ folder for one project. In the Claude apps you can instead upload it under Settings → Capabilities → Skills.

What do I need before using this skill?

You need the gcloud CLI with access to your target GCP project and Application Design Center spaces, the Terraform CLI, Python 3 with absl-py, and a GCP project ID and location. If no location is given, the skill defaults to us-central1.

Is this skill free to use?

The skill is published under the Apache-2.0 licence. Running it creates real Google Cloud resources, which can incur cost in your own project.

Does this skill work with AWS or Azure?

No. The skill is explicitly limited to GCP infrastructure, so AWS or Azure work is not covered.

Why won't the agent deploy my Terraform straight away?

The workflow requires local validation, a plan export and a best-practices scan before import, and it treats unresolved high or critical findings as blockers. Deployment only proceeds after those checks pass.

How does this compare with running Terraform and gcloud myself?

The skill does not replace Terraform or gcloud; it sequences them into a repeatable design-to-deploy process with fixed checks, secrets and state rules. If you already follow that process by hand, the skill adds less value.

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 who already use Terraform and gcloud on Google Cloud and want a guided, repeatable path from design to deployment through Application Design Center. It is a poor fit for non-GCP clouds, for anyone without gcloud access, a GCP project and Terraform installed, or for anyone who would run its commands against production without reviewing them first.

Download the skill