Kipt

Your prompts and skills, one keystroke away.

Explore

google-agents-cli-deploy skill: ship agents to Google Cloud

The agents-cli deploy skill from google/agents-cli guides coding agents through deploying agents to Agent Runtime, Cloud Run or GKE, with CI/CD and secrets.

By the Kipt team · Updated 26 October 2026

Source: google/agents-cli · Maintainer: Google · License: Apache-2.0 · Checked: 26 October 2026

What is the google-agents-cli-deploy skill?

google-agents-cli-deploy is an Agent Skill that teaches a coding agent how to deploy an AI agent to Google Cloud with the agents-cli command. It covers choosing a target (Agent Runtime, Cloud Run or GKE), deploy flags, sizing, CI/CD pipelines, secrets, service accounts, rollback and common errors.

It is one of seven skills in google/agents-cli, Google's "CLI and skills for building agents on Gemini platform." The SKILL.md description says it "Applies to any framework agents-cli deploys (ADK, LangChain, ...)" and should not be used for agent code, evaluation or scaffolding, which have their own skills.

As of 26 October 2026, its skills.sh page listed about 512,600 installs (snapshot, 26 October 2026). The SKILL.md metadata lists Google as author and version 1.11.0.

What does it do?

The SKILL.md is a long deployment guide that tells the agent to prefer agents-cli commands because "they wrap Terraform, Docker, and deployment into a tested pipeline."

Pick a target

It gives a decision matrix and then tells the agent to ask you which target fits:

Agent Runtime Cloud Run GKE
Setup complexity Lower (managed, purpose-built for agents) Medium (Dockerfile, Terraform, networking) Higher (Kubernetes expertise required)
Cost model vCPU-hours + memory-hours (not billed when idle) Per-instance-second + min instance costs Node pool costs
Session state Sessions in Gemini platform In-memory (dev), Cloud SQL or a managed backend In-memory (dev), Cloud SQL or a managed backend
Best for Managed infrastructure, minimal ops Custom infra, full networking control Full Kubernetes control

All three are container-based, "so any language works." Agent Runtime is part of Google's Gemini Enterprise Agent Platform; the skill notes that "Agent Engine" and "Vertex AI Agent Engine" mean the same thing.

Deploy to dev, with a human gate

The dev workflow has four steps: enhance a prototype with agents-cli scaffold enhance . --deployment-target <target> if needed, paste eval scores and test results and ask "Ready to deploy to dev?", wait for explicit approval, then run agents-cli deploy. In bold: "Never run agents-cli deploy without explicit human approval."

Flags and sizing

A flag table covers --project, --region, --service-account, --secrets, --env-vars, --memory, --cpu, --min-instances, --max-instances, --concurrency, --timeout, --iap, --no-wait, --status, --dry-run and more. Defaults on Agent Runtime and Cloud Run are --cpu 1, --memory 4Gi, --concurrency 8, --min-instances 0 and --max-instances 10, so dev agents scale to zero. The skill's sizing advice is to scale out rather than up and to raise memory with concurrency, because "raising --concurrency without --memory is the main OOM cause."

Production and operations

For production it points to references/cicd-pipeline.md and agents-cli infra cicd. It also covers the two service accounts in scaffolded projects (app_sa and cicd_runner_sa), required APIs, Secret Manager, Cloud SQL permissions, IAP, rollback, custom Terraform, Agent Gateway binding, customer-managed encryption keys and a troubleshooting table. Eight reference files go deeper: cloud-run.md, agent-runtime.md, gke.md, terraform-patterns.md, batch-inference.md, cicd-pipeline.md, testing-deployed-agents.md and cmek.md.

How do you install it?

The Getting Started guide lists Python 3.11+, uv and Node.js as prerequisites, and the Google Cloud SDK and Terraform as optional extras for deployment. The recommended install adds the CLI and all seven skills:

uvx google-agents-cli setup

Or install skills only, either the whole set or just this one (the command shown on skills.sh):

npx skills add google/agents-cli
npx skills add https://github.com/google/agents-cli --skill google-agents-cli-deploy

The skill's metadata lists the agents-cli binary as a requirement, installed with uv tool install google-agents-cli. The guide supports macOS, Linux and Windows through WSL 2. In Claude Code, run /skills to confirm the agents-cli skills are listed. Codex and Cursor read the same SKILL.md format; see Codex skills and Cursor skills.

Before deploying, the SKILL.md lists three APIs to enable in your project: cloudbuild.googleapis.com, secretmanager.googleapis.com and run.googleapis.com.

How do you use it?

The skill is model-invoked when you ask to "deploy an agent", "set up CI/CD", "configure secrets" or "troubleshoot a deployment", per its description. You can also call /google-agents-cli-deploy directly in Claude Code.

Example prompts (suggestions, not quotes from the repo):

  • Deploy this agent to Cloud Run in us-central1. Show me the dry run first.
  • Compare Agent Runtime and Cloud Run for this agent. It needs OAuth access to users' Google Drive.
  • Move the OpenAI key into Secret Manager and pass it at deploy time instead of as an env var.
  • Set up a CI/CD pipeline with staging and prod projects using GitHub Actions.
  • The deployed agent returns 403 when I call it. Walk me through the likely causes.

To test a deployed agent, the skill suggests agents-cli run --url <service-url> --mode a2a "your prompt", which "handles auth, sessions, and streaming automatically" for Agent Runtime and Cloud Run.

Tips and limits

From the SKILL.md, the repo docs and skills.sh, not from testing:

  • Project confirmation. If the project is resolved automatically, agents-cli deploy prompts for confirmation. Agents usually run non-interactively, so the skill says to pass --no-confirm-project or --project.
  • Timeouts are not failures. Agent Runtime deploys can take 5 to 10 minutes. Check with agents-cli deploy --status.
  • No gcloud for Agent Runtime. The skill says to deploy with agents-cli deploy and query with the Python SDK.
  • Rollback differs by target. Cloud Run can shift traffic to an earlier revision, GKE uses kubectl rollout undo, and Agent Runtime has no revision-based rollback.
  • Infrastructure belongs in Terraform. The skill says custom resources such as Cloud SQL or Pub/Sub "MUST" go in Terraform for production, with quick experiments in gcloud as the exception.
  • Costs are yours. The README notes that deploying creates resources in your own Google Cloud project, and you are responsible for them.
  • Security audits. On skills.sh, Gen Agent Trust Hub rated it Pass with risk level Safe (22 October 2026); its report notes an indirect prompt injection surface in the BigQuery and Eventarc examples and credits the human-approval gate before deploys. Socket (20 October 2026) and Snyk (risk level Low, 20 October 2026) rated it Pass.

Under Apache 2.0 you can copy and modify the skill as long as you keep the license and notices and mark files you change.

Save it in your Kipt library

In Kipt, open Skills and paste the skill's GitHub folder link to import it with its eight reference files. Paste the repo link instead and Kipt shows a list of all seven agents-cli skills to pick from. You can also drop the folder or a .zip. Your skills live in your own private GitHub repo and install into Claude Code or Codex with one command:

npx skills add https://github.com/<you>/<repo>.git --all -g

Kipt is not affiliated with Google.

Similar skills

Questions

What is the google-agents-cli-deploy skill?

It is an Agent Skill in Google's google/agents-cli repo that guides a coding agent through deploying an agent to Google Cloud. It covers the three targets agents-cli supports (Agent Runtime, Cloud Run and GKE), deploy flags, sizing, CI/CD, secrets, service accounts, rollback and troubleshooting. It applies to any framework agents-cli deploys, including ADK and LangChain.

Will the deploy skill let my coding agent deploy without asking?

No. The SKILL.md says to paste eval scores and test results, ask the person whether they are ready to deploy to dev, and wait for explicit approval. It states: Never run agents-cli deploy without explicit human approval.

Which deployment target does the skill recommend?

None by default. It gives a decision matrix comparing Agent Runtime, Cloud Run and GKE on scaling, networking, session state, cost model and setup complexity, then tells the agent to ask the user which target fits. It does say Agent Runtime is the target for agents that need managed OAuth user consent.

How do I install the agents-cli deploy skill?

Run uvx google-agents-cli setup to install the agents-cli command and all seven skills, or add only this one with npx skills add https://github.com/google/agents-cli --skill google-agents-cli-deploy. The skill needs the agents-cli binary, and deployment needs a Google Cloud project.

How do I roll back a deployment made with agents-cli?

The skill says the main rollback is git-based: fix, commit and push, and CI/CD redeploys. For Cloud Run you can shift traffic to an earlier revision with gcloud run services update-traffic, and for GKE use kubectl rollout undo. Agent Runtime has no revision-based rollback, so you fix and redeploy.

Why does an Agent Runtime deploy seem to hang?

The skill says Agent Runtime deploys can take 5 to 10 minutes and may exceed command timeouts, but the deployment continues server-side. Run agents-cli deploy --status to check progress, or start with --no-wait and check later.

More in Skill directory

All guides

Keep your prompts and skills in one place, in your own GitHub. Install every skill with one command.

Get started

Kipt is an independent product, not affiliated with or endorsed by Anthropic, OpenAI, Cursor or GitHub. Claude, ChatGPT, Codex, Cursor and GitHub are trademarks of their owners.