Guide · Agent skill

How to use the Agent Platform GenAI Inference agent skill

Updated 2026-10-29 · written with AI from the project's own documentation and checked against our catalogue data

This guide explains the agent-platform-inference skill, a set of instructions and Python scripts that lets an AI agent send prompts to Gemini and open models on Google Cloud's Agent Platform. It is written for people who have not used agent skills before. By the end you will know how to install the skill, set up Google Cloud credentials, trigger it with a plain request and approve a run safely.

Agent Platform GenAI Inferencegoogle/skills · Apache-2.0 Download the skill

Quick start

1Download and unzip. Download the ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder.
2Log in to Google Cloud. Run gcloud auth login and gcloud auth application-default login with your Google account.
3Enable the API. Run gcloud services enable aiplatform.googleapis.com on your project if it is not already enabled.
4Check Python packages. Run the import probe and install scripts/requirements.txt only if the packages are missing.
5Ask with a natural request. Describe the model, SDK, and prompt you want to use in plain language.
6Approve the confirmation card. Review the project, region, model, SDK and prompt, then reply Yes to run the inference.

What does the skill make the agent do?

Once installed, the agent knows how to connect to Google Cloud's Agent Platform, authenticate with your Google credentials, and send prompts to Gemini models or third-party open models such as Llama, DeepSeek and Qwen. It then returns the answers. The skill also includes ready-made Python examples for three SDKs: the Google GenAI SDK, the OpenAI SDK and the Vertex AI SDK.

It also gives guidance for troubleshooting common errors, such as 429 quota errors (Resource Exhausted) and 404 Not Found errors.

Install the skill

You can install the skill in one of two ways. Download the ZIP from the skill's page on this site and unzip it into ~/.claude/skills/ for your user account, or into .claude/skills/ inside a project folder if you want it limited to that project.

Alternatively, in the Claude apps, upload the ZIP under Settings → Capabilities → Skills.

Personal install: unzip into ~/.claude/skills/
Project install: unzip into the project's .claude/skills/ folder
Claude apps: Settings → Capabilities → Skills, then upload the ZIP

Set up Google Cloud before the first run

You need a Google Cloud account with a project, and billing set up on that project, because every call is billed to it. The Agent Platform API (aiplatform.googleapis.com) must be enabled. You also need Python 3 and the gcloud CLI. Run the two login commands so the scripts can find your credentials.

The skill's Phase 0 setup asks you to probe for the Python packages first and install only what is missing. The scripts need google-genai, google-cloud-aiplatform (which provides vertexai) and openai, listed in scripts/requirements.txt. The source advises against creating a virtual environment for this step, and against installing requirements.txt on top of a working environment.

Python 3 installed
Google Cloud project with billing enabled
Agent Platform API enabled
gcloud CLI installed and logged in

gcloud auth login
gcloud auth application-default login
gcloud services enable aiplatform.googleapis.com
python3 -c "import vertexai, google.genai, openai" || pip install -r scripts/requirements.txt

Trigger the skill with a natural request

You do not need special syntax. Ask in plain language for an inference task, such as asking a model a question, running a test prompt, running a chat completion, or generating code that calls a Gemini or open model. Also name the authentication or endpoint problem if you are debugging one.

The agent will not guess the model. If you do not name a model or model family, and an SDK preference, it will ask. It will also ask for your project ID and region if they are missing.

"Ask Gemini 2.5 Pro to summarize the plot of Hamlet in 3 sentences using the Google GenAI SDK."
"Run a test prompt on DeepSeek through the OpenAI SDK."
"I get a 429 Resource Exhausted error on Agent Platform, help me debug it."

Approve the confirmation card before any run

Before any inference that can cost money, the agent must show a confirmation card listing the project ID, region or location, exact model ID, SDK, and input prompt, plus any extra generation settings. It then waits for your explicit Yes. A natural-language summary without these fields does not count. Check the card carefully, especially the project and region, before you answer.

After a run, the agent's reply should state the model ID, SDK, project, region and the generated output, so you can confirm what was actually used.

Safety notes: scripts and network access

This skill runs scripts and uses the network. The scripts send your prompts to Google Cloud models, so treat any prompt you approve as data leaving your machine. The skill is described as read-only, and it excludes deploying models to endpoints and running model evaluations. Use a separate deployment skill for deployments.

Review the scripts in the scripts/ folder before running them if you are in a sensitive environment. Use a project where you are comfortable with the charges.

Every inference call is billed to your Google Cloud project
Deployment and model evaluation are out of scope for this skill
Scripts use your Application Default Credentials

What people use it for

Compare how Gemini and a Llama or DeepSeek model answer the same prompt
Check that your Google Cloud credentials and project work before building an app
Test the same prompt across the GenAI SDK, OpenAI SDK and Vertex AI SDK
Get help fixing 429 quota, 400 user validation or 404 not found errors
Generate starter code for calling a Gemini or open model from Python

Frequently asked questions

Is the agent platform GenAI inference skill free?

The source does not state a price or licence cost for the service. The skill is published under the Apache-2.0 licence. Every model call is billed to your Google Cloud project, so check your project's billing before running prompts.

What do I need to run the agent-platform-inference skill?

You need Python 3, a Google Cloud account with a project and billing, the Agent Platform API enabled, and the gcloud CLI. You also need the Python packages google-genai, google-cloud-aiplatform and openai, listed in scripts/requirements.txt.

How do I install an agent skill from a ZIP file?

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

Why does the skill say 429 Resource Exhausted?

A 429 Resource Exhausted error means a quota has been reached. The skill includes troubleshooting guidance for this error, and it also covers 404 Not Found and 400 User Validation errors. Check your project's quota and the region you are calling.

Can this skill deploy models or run evaluations?

No. The skill explicitly excludes deploying models to endpoints and running model evaluations. Use a deployment skill for deployments.

What is the difference between the GenAI SDK, OpenAI SDK and Vertex AI SDK here?

The skill provides example scripts for all three, so you can run the same kind of prompt through each one. This is useful for comparing outputs across SDKs. The source does not rank them or say which gives better answers.

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 developers and teams who want an agent to run test prompts on Gemini or open models on Google Cloud, compare SDKs, or debug authentication and quota errors. It is not a fit if you need to deploy models or run evaluations, or if you have no Google Cloud project with billing set up.

Download the skill