This repository provides example skills for Google Antigravity and the Gemini CLI, demonstrating the "Agentic Command Line" concept. It offers modular units of expertise, workflows, and tools for AI agents, targeting developers and power users seeking to leverage AI for command-line tasks. The benefit lies in practical, progressive examples that illustrate how to build and integrate AI-powered capabilities into developer workflows.

How It Works

Antigravity Skills define agent behavior, tool usage, and context. This project showcases skill development across five complexity levels. These range from Level 1's pure prompt engineering (git-commit-formatter) to Level 2's asset utilization (license-header-adder), Level 3's few-shot learning (json-to-pydantic), Level 4's deterministic script delegation (database-schema-validator), and Level 5's composition of scripts, templates, and examples (adk-tool-scaffold). The Gemini CLI skills include always-verify-gcp, which validates GCP commands against documentation and user input. This structured approach facilitates learning and implementation of increasingly sophisticated agentic functionalities.

Quick Start & Requirements

Clone this repository. Copy desired skill folders from skills_tutorial/ or gemini-cli-skills/ into your workspace's .agent/skills/ directory or your global ~/.gemini/antigravity/skills/ directory. Restart your agent session. Prerequisites include a functional Antigravity or Gemini CLI environment.

• Blog Post: https://medium.com/google-cloud/tutorial-getting-started-with-antigravity-skills-864041811e0d
• Codelab: https://codelabs.developers.google.com/getting-started-with-antigravity-skills?hl=en#0

Highlighted Details

• Demonstrates a progressive 5-level complexity framework for building AI agent skills.
• Features practical skills like automatic Git commit message formatting and license header injection.
• Includes advanced examples such as JSON-to-Pydantic model generation via few-shot learning and SQL schema validation using Python scripts.
• The always-verify-gcp skill exemplifies agent interaction with official documentation and user prompts for command accuracy.

Maintenance & Community

The project mentions a Google Antigravity Community Hub for additional resources, articles, and information. No specific details on active contributors, sponsorships, or a public roadmap are provided in the README snippet.

Licensing & Compatibility

The project is licensed under the Apache 2.0 license. This license is permissive and generally compatible with commercial use and linking within closed-source projects.

Limitations & Caveats

This repository serves as a collection of sample skills; users must integrate them into their specific Antigravity or Gemini CLI environments. The effectiveness and functionality of these skills are dependent on the underlying AI agent's capabilities and the Antigravity framework's implementation.