Google Agent Skills: 11 Broad Categories for Cloud and AI Automation

Google Agent Skills organizes cloud automation capabilities into 11 broad functional categories—from AI/ML and Infrastructure to Security and Advertising—defined in the repository's README.md and installable via npx skills add or the skills.sh installer.

The google/skills repository provides a structured taxonomy of agentic capabilities for Google Cloud Platform. These categories appear as second-level headings under "## Available Skills" in the canonical documentation, grouping related automation workflows, APIs, and CLI operations into logical domains for easy discovery.

The 11 Broad Categories of Google Agent Skills

The repository defines 11 primary functional categories that classify all available skills. Each category corresponds to a specific directory structure under the skills/ folder and represents a distinct domain of Google Cloud operations.

1. Getting Started with Google Cloud

This category covers foundational onboarding skills, including authentication workflows, foundation builder patterns, and introductory guides for new Google Cloud users. Skills in this section help agents establish initial access and configure basic project resources.

2. Multi-Product Solution Skills

End-to-end solution architectures that span multiple Google Cloud products live here. This includes skills for solution architecture design, multi-agent security implementations, agentic analytics, data lakehouse construction, RAG (Retrieval-Augmented Generation) pipelines, hybrid search systems, and comprehensive AI agent workflows.

3. AI/ML

The AI/ML category contains skills specific to the Agent Platform, including alert configuration, endpoint management, inference optimization, Model Garden integrations, model registry operations, tuning workflows, and prompt management. It also covers Gemini API utilization and Genkit SDK implementations for generative AI applications.

4. Infrastructure

GKE-centric infrastructure and lifecycle management skills populate this category. Coverage includes cluster creation, autoscaling policies, upgrade orchestration, networking configuration, storage management, and TPU monitoring for high-performance machine learning workloads.

5. Databases and Analytics

Database-focused automation resides here, spanning AlloyDB, BigQuery, Bigtable, Cloud SQL, and Spanner operations. This category also includes Managed Airflow workflows and data analytics pipeline management tools.

6. Developer Tools

Developer experience utilities live in this section, including device platform integrations, gcloud CLI operations for agents, Google Agents CLI onboarding procedures, Cloud Build basics, and developer knowledge retrieval systems.

7. Management Tools

Cloud monitoring and observability skills are organized here. Capabilities include chart generation, time-series data requests, PromQL generation for Prometheus monitoring, logging configuration, cost analysis automation, and Workload Manager basics.

8. Well-Architected Framework

Skills aligned with the Google Cloud Well-Architected Framework (WAF) pillars reside here. These cover the six core pillars: Cost Optimization, Operational Excellence, Performance Efficiency, Reliability, Security, and Sustainability.

9. Security and Identity

IAM-related skills dominate this category, including policy simulation, Privileged Access Manager operations, Security Command Center (SCC) queries, detection coverage analysis, and comprehensive GKE platform and workload security implementations.

10. Web and App Hosting

This category focuses on serverless and application hosting platforms, specifically Cloud Run deployment automation and Firebase configuration basics for mobile and web applications.

11. Advertising

Mobile advertising and marketing automation skills are grouped here, including Mobile Ads SDK guides, IMA SDK and DAI SDK implementations, Google Ads API quickstarts, and data manager API operations for advertising platforms.

Note: The repository also includes an "Others" category that serves as an index linking to external Google skill repositories for advanced Cloud Storage, ADK, Android, Dart, Firestore, Flutter, and Maps Platform capabilities.

Repository Structure and Category Mapping

The taxonomy is enforced through the repository's physical directory structure. In README.md, each second-level heading (##) defines a broad category, while individual skills map to subdirectories under skills/.

For example, the AI/ML category heading in README.md corresponds to skills located in paths like skills/cloud/agent-platform-inference and skills/cloud/model-garden. Each skill directory contains a SKILL.md file defining the specific implementation details, parameters, and automation logic.

According to CONTRIBUTING.md, new skill submissions must be placed under one of these existing broad categories, ensuring the taxonomy remains consistent as the repository grows.

Installing Skills by Category

Users interact with categories during the installation process. The npx skills add command reads the README's category sections and can filter skills based on these classifications.

To install a specific skill from the AI/ML category:

npx skills add google/skills/cloud/agent-platform-inference

To browse all available skills within a specific category using the interactive installer:

npx skills add google/skills

# When prompted, type "AI/ML" to filter the list to that category

The installer parses the README.md structure dynamically, presenting only skills listed under the selected category heading.

Programmatically Accessing Category Definitions

You can extract the current category list directly from the repository's canonical documentation using Node.js:

import fetch from 'node-fetch';

async function getCategories() {
  const url = 'https://raw.githubusercontent.com/google/skills/main/README.md';
  const text = await (await fetch(url)).text();
  const categories = [...text.matchAll(/^##\s+(.+)$/gm)]
                     .map(m => m[1].trim())
                     .filter(t => t !== 'Available Skills');
  console.log(categories);
}

getCategories();

This script fetches the raw README.md content and extracts all second-level headings, which correspond exactly to the 11 broad categories listed under "Available Skills."

Summary

• Google Agent Skills organizes capabilities into 11 broad functional categories: Getting Started, Multi-Product Solutions, AI/ML, Infrastructure, Databases and Analytics, Developer Tools, Management Tools, Well-Architected Framework, Security and Identity, Web and App Hosting, and Advertising.

• The canonical category definitions live in README.md under the "## Available Skills" section, with each category mapping to specific subdirectories under skills/.

• Individual skill implementations are defined in SKILL.md files within category-specific folders (e.g., skills/cloud/ for cloud-related categories).

• Users install skills by category using npx skills add google/skills and filtering by category name, or by specifying the full path to a specific skill repository.

Frequently Asked Questions

What are the 11 broad categories in Google Agent Skills?

The 11 broad categories are: Getting Started with Google Cloud, Multi-Product Solution Skills, AI/ML, Infrastructure, Databases and Analytics, Developer Tools, Management Tools, Well-Architected Framework, Security and Identity, Web and App Hosting, and Advertising. These group together related Google Cloud automation capabilities ranging from infrastructure management to advertising SDKs.

How do I install skills from a specific category?

Use the command npx skills add google/skills to launch the interactive installer, then type the category name (e.g., "AI/ML") when prompted to filter the list. Alternatively, install a specific skill directly using npx skills add google/skills/[category-path]/[skill-name], such as npx skills add google/skills/cloud/agent-platform-inference for an AI/ML skill.

Where are skill categories defined in the repository source code?

Categories are defined as second-level headings (##) in the README.md file under the "## Available Skills" section. Each category heading corresponds to a folder structure under the skills/ directory, where individual skills reside in subdirectories containing SKILL.md implementation files.

Can I contribute new skills to existing categories?

Yes. According to CONTRIBUTING.md, new skill submissions must be placed within the existing folder structure under one of the 11 broad categories. The skill directory should contain a SKILL.md file defining the automation logic, and the contribution should be referenced under the appropriate category heading in README.md.

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