How to Set Up a Local Development Environment for google/skills: 9-Step Guide

Clone the repository, install Python 3.9+, Node 18+, Terraform ≥1.5, and the Google Cloud SDK, then use uvx google-agents-cli setup to register skill modules and verify with a sample skill run.

The google/skills repository is a collection of Markdown-based "skill" definitions that power interactive, code-generation agents. Setting up a local development environment for google/skills requires specific tooling and a repeatable workflow, even though the repository contains no compiled source code. This guide walks through the exact steps derived from the repository's own onboarding instructions and skill files.

Prerequisites for the google/skills Development Environment

Before cloning the repository, install these core tools. They are referenced across multiple skill files and are essential for running skills that provision GCP resources and execute code samples.

Tool	Minimum Version	Purpose
Git	Latest	Clone the repository
Python	3.9+	Run the agents-CLI (via uvx)
Node.js	18+	Execute JavaScript/TypeScript reference implementations
Terraform	≥ 1.5	Provision GCP resources via IaC
Google Cloud SDK	Latest	Authenticate and run GCP commands

The google-cloud-recipe-onboarding SKILL.md outlines billing setup and project creation, and points to the official Cloud SDK installer at SKILL.md#L60-L62.

Step 1: Clone the google/skills Repository

Retrieve the full set of skill definitions, reference code, and documentation:

git clone https://github.com/google/skills.git
cd skills

Step 2: Create an Isolated Python Environment

Use a virtual environment to isolate dependencies required by the agents-CLI and any utility scripts:

python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

Step 3: Install the Agents CLI

The google-agents-cli is the entry point for all skills. Install uvx for lightweight package execution, then run the setup command:

pip install uvx
uvx google-agents-cli setup

This registers available skill modules on your machine. The setup process is documented in skills/cloud/google-agents-cli-onboarding/SKILL.md#L35-L40.

Step 4: Authenticate to Google Cloud

Most skills interact with GCP resources. Authenticate before running any cloud-related skill:

gcloud auth login --no-browser  # Or use the standard interactive flow

This step is required for skills like Workload Manager and GKE basics to provision resources successfully.

Step 5: Select Your Development Environment

The repository distinguishes production and development contexts. As noted in skills/cloud/agent-platform-deploy/references/usage.md#L5, you should select the dev environment for local work. This uses separate Terraform state files and avoids billing-impacting resources.

Step 6: Verify with a Sample Skill Run

Test that the CLI can load a skill, resolve dependencies, and execute defined steps:

uvx google-agents-cli run skills/cloud/gke-basics/SKILL.md

This validates your complete google/skills local development environment setup.

Step 7: Configure Linting and Testing Tools

Before contributing, run static checks as encouraged in [CONTRIBUTING.md](https://github.com/google/skills/blob/main/CONTRIBUTING.md):

Check	Command
Markdown lint	markdownlint-cli2 "**/*.md"
Terraform lint	tflint on any .tf examples
Python tests	pytest in skill tests/ folders

Step 8: Use the DevContainer (Optional)

For a reproducible environment without local tool installation, use the pre-configured DevContainer. The repository includes .devcontainer/devcontainer.json with Python, Node, Terraform, and Cloud SDK pre-installed.

Open in VS Code with the Dev Containers extension, or run:

code .  # Then select "Reopen in Container"

Key Configuration Files in google/skills

File	Location	Purpose
README.md	/README.md	Top-level overview and entry points
CONTRIBUTING.md	/CONTRIBUTING.md	Workflow, linting tools, submission guidelines
google-cloud-recipe-onboarding	/skills/cloud/google-cloud-recipe-onboarding/SKILL.md	Definitive GCP project setup wizard
google-agents-cli-onboarding	/skills/cloud/google-agents-cli-onboarding/SKILL.md	CLI installation and usage
agent-platform-deploy	/skills/cloud/agent-platform-deploy/references/usage.md	Environment selection (dev vs prod)
gke-basics	/skills/cloud/gke-basics/SKILL.md	End-to-end CLI verification skill

Summary

• Core stack: Python 3.9+, Node 18+, Terraform ≥1.5, Google Cloud SDK, and Git
• CLI setup: Use uvx to install and configure google-agents-cli
• Authentication: Run gcloud auth login before any GCP-interacting skill
• Environment: Select dev context to isolate from production resources
• Verification: Execute uvx google-agents-cli run skills/cloud/gke-basics/SKILL.md
• Quality: Apply markdownlint, tflint, and pytest before commits per CONTRIBUTING.md
• Alternative: Use the provided DevContainer for zero-local-install setup

Frequently Asked Questions

Does google/skills require Docker or containers?

No. Docker is optional. The repository works with directly installed tools. However, the DevContainer in .devcontainer/devcontainer.json provides a pre-configured environment with all dependencies if you prefer containerized development.

What Python version does google/skills require?

Python 3.9 or higher. This requirement comes from the agents-CLI and related tooling referenced in skill onboarding files.

How do I know if my local development environment for google/skills is working correctly?

Run the verification command: uvx google-agents-cli run skills/cloud/gke-basics/SKILL.md. If the skill loads, resolves dependencies, and executes without errors, your environment is properly configured.

Can I develop google/skills skills without a GCP project?

Partially. You can edit Markdown skill definitions and run local tests. However, any skill that provisions cloud resources requires GCP authentication and an active project, as documented in google-cloud-recipe-onboarding/SKILL.md.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}