Antigravity CLI Training Course

A comprehensive training course for mastering Google's Antigravity CLI (agy) - the AI coding agent that brings Gemini, Claude, and open models directly into your terminal.

Course Overview

This 5-hour hands-on workshop covers everything from basic installation to advanced MCP configurations, plugins, and enterprise workflows.

What You'll Learn

• Installation & Setup: Authentication strategies, environment configuration
• Core Commands: Interactive mode, one-shot prompts, file references
• Context Management: AGENTS.md files, hierarchical context loading
• Safety & Control: Sandbox mode, permission rules, Artifact Review
• Advanced Features: MCP integration, plugins, skills, subagents
• Practical Skills: Real-world exercises in Python, JavaScript, and Java

Prerequisites

• Command-line experience
• Basic programming knowledge in at least one language
• Git familiarity
• Google account for sign-in (or an ANTIGRAVITY_API_KEY)
• Node.js 18+ (only needed to run the Slidev presentation locally)

Repository Structure

gemini-training/
├── slides.md                     # Slidev presentation
├── lab_handout.md               # Progressive hands-on labs
├── exercises/                    # Hands-on lab projects
│   ├── python/                  # Python projects
│   │   └── weather-app/         # Flask weather application
│   ├── javascript/              # JavaScript/TypeScript projects
│   │   └── task-manager/        # Node.js task manager
│   └── java/                    # Java projects
│       └── bookstore-api/       # Spring Boot REST API
├── config-examples/              # Sample configurations
│   ├── settings-basic.json      # Starter settings
│   ├── settings-advanced.json   # Full-featured settings
│   ├── settings-safe.json       # Safety-focused settings
│   └── settings-mcp.json        # MCP-focused settings
├── gemini-md-examples/          # GEMINI.md templates
│   ├── python-flask.md          # Python Flask template
│   ├── java-spring.md           # Spring Boot template
│   └── javascript-react.md      # React project template
└── commands/                     # Custom command examples
    ├── review.toml              # Code review command
    └── test-gen.toml            # Test generation command

Quick Start

1. Install Antigravity CLI

# macOS / Linux (installs the `agy` binary to ~/.local/bin)
curl -fsSL https://antigravity.google/cli/install.sh | bash

# Verify installation
agy --version

2. Authenticate

# First run launches Google Sign-In automatically
agy

# Or use an API key for scripts / headless use
export ANTIGRAVITY_API_KEY="your-api-key"

3. Run Your First Command

agy -p "Hello! What can you help me with?"

4. Start the Training

# Clone this repository
git clone https://github.com/kousen/gemini-training
cd gemini-training

# Install dependencies for slides
npm install

# Start the presentation
npm run dev

# Open browser to http://localhost:3030

Course Schedule

Section	Duration
Introduction & Setup	10 min
Installation & Prerequisites	15 min
Authentication & Account Setup	10 min
First Steps & Basic Interface	15 min
Core Commands & Functionality	65 min
Configuration & Customization	65 min
Advanced Features & Extensions	65 min
Practical Applications & Workflows	45 min
Wrap-up & Q&A	15 min

Key Antigravity CLI Features Covered

Core Features

• Interactive REPL and one-shot (-p) modes
• File references with @ syntax
• Shell integration with ! prefix
• Sandbox mode for safe execution
• Multi-model: Gemini, Claude, and GPT-OSS via /model

Configuration

• AGENTS.md context files (hierarchical)
• settings.json customization
• Permission rules (/permissions)
• Plugins and skills

Advanced Features

• Model Context Protocol (MCP) servers (mcp_config.json)
• Plugin system (agy plugin)
• Skills and subagents (/agent)
• Conversation management (SQLite) and Artifact Review
• Antigravity 2.0 desktop integration (/export)

New in Antigravity CLI 1.0.x

• Initial 1.0.0 release of the Go-based Antigravity CLI
• Multi-model sessions: Gemini, Claude, GPT-OSS via /model (1.0.5)
• models subcommand + --model flag to choose a model (1.0.5)
• /permissions to manage tool permission rules in-CLI (1.0.5)
• G1 credits with /credits, /usage, /quota (1.0.3)
• MCP url support in mcp_config.json; parallel init (1.0.5/1.0.6)
• SQLite conversation format + /resume, --continue (1.0.4)
• Plugin discovery for skills and subagents (1.0.1)

Tips for Success

1. Start with default approval: Get comfortable before --dangerously-skip-permissions
2. Use AGENTS.md: Provide context for better, more consistent results
3. Use Artifact Review: Inspect diffs (Ctrl+R, /diff) before accepting
4. Review Generated Code: Always verify AI-generated changes
5. Leverage MCP: Extend capabilities with custom tools

Useful Commands Reference

# Basic usage
agy                                 # Interactive mode
agy -p "prompt"                     # One-shot (print) mode
agy -i "context"                    # Interactive with initial prompt

# Configuration
agy --sandbox                       # Run in sandbox mode
agy --dangerously-skip-permissions  # Auto-approve all actions (use with care)
agy --model "Gemini 3.1 Pro (High)" # Specify a model (see `agy models`)

# Session management
agy -c                              # Continue most recent conversation
agy --conversation <id>             # Resume a specific conversation

Slash Commands

Command	Description
/help	Show available commands and shortcuts
/context	View loaded context and token usage
/model	Switch the active model mid-session
/settings	Open settings and preferences
/permissions	Add/edit/remove tool permission rules
/resume	Open the conversation browser
/agent <task>	Dispatch an asynchronous subagent
/export	Send the session to the desktop app
/diff	Review changes (supports commit selection)
/usage · /credits · /quota	Quota and credit status

Resources

• Official Antigravity CLI Documentation
• Antigravity CLI Source Repository
• MCP Server Registry
• Course Slides
• Lab Exercises

Instructor

Kenneth Kousen

• President, Kousen IT, Inc.
• Author & Technical Trainer
• ken.kousen@kousenit.com
• https://www.kousenit.com

License

This training material is licensed under the MIT License. See LICENSE file for details.

---

Ready to start? Run npm run dev and navigate to http://localhost:3030!