Updated November 2026

Agentic Coding with Google Antigravity: the Antigravity App, CLI, and Gemini Models

Class Duration

14 hours of live training delivered over 2-3 days to accommodate your scheduling needs.

Student Prerequisites

Professional software development experience in at least one language
Working knowledge of Git and GitHub
Comfort with the command line
A Google account; a Google AI Pro or Ultra plan or a Gemini Enterprise license is recommended
Prior use of an AI coding assistant is helpful

Target Audience

Software engineers, tech leads, and platform teams who want to use Google's agentic coding tools on real projects. Teams moving from Gemini CLI or Gemini Code Assist, and organizations that run agents under Google Cloud identity and billing, will also benefit. Teams still comparing tools can start with Choosing Your AI IDE.

Description

Google Antigravity is Google's agentic development platform, powered by Gemini models. It includes the Antigravity App for running many agents, the agy CLI, IDE extensions, and a Python SDK. This course covers planning with implementation plans, reviewing artifacts, and checking UI changes with browser agents. Students run parallel agents in worktrees, add rules, skills, and MCP servers, and script the CLI for CI. The course also covers permissions, enterprise controls, and moving from Gemini CLI.

Learning Outcomes

Choose between the Antigravity App, CLI, IDE extensions, and SDK.
Sign in with a personal, enterprise, or Google Cloud identity.
Choose Gemini models and reasoning effort for a task.
Run plan-then-execute work with implementation plans and artifacts.
Verify front-end changes with browser agents.
Run parallel agents and subagents in isolated worktrees.
Ground agents with AGENTS.md, GEMINI.md, rules, and skills.
Extend agents with MCP servers, plugins, and hooks.
Automate agent tasks in CI with CLI headless mode.
Apply permission presets, sandboxing, and quota controls.
Migrate Gemini CLI configuration to Antigravity.

Training Materials

Comprehensive courseware is distributed online at the start of class. All students receive a downloadable MP4 recording of the training.

Software Requirements

The Antigravity App and Antigravity CLI installed (macOS, Windows, or Linux), a Google account (Google AI Pro or Ultra, or a Gemini Enterprise license, recommended), Google Chrome, Git, a GitHub account with a repository, and optionally VS Code or a JetBrains IDE with the Antigravity extension.

Training Topics

Google's Agentic Coding Tools

The Antigravity App
The agy CLI
IDE extensions and the Antigravity IDE
The Antigravity SDK
Choosing a surface per task
Where Gemini CLI fits now
Gemini Code Assist license changes
The Jules wind-down
Personal plans vs. Gemini Enterprise
Keeping a human in the loop

Installing the App and CLI
Installing IDE extensions
Personal Google account sign-in
Gemini Enterprise accounts and projects
Gemini API keys for the CLI
Application Default Credentials
Configuration in ~/.gemini/config/

Gemini Models and Model Choice

Current Gemini model tiers
Choosing a model per conversation
/model and --model in the CLI
Reasoning effort with /effort
Claude models in Antigravity
Matching models to task and quota

Working with Agents in the Antigravity App

Projects and scoped settings
Local mode vs. worktree mode
Planning mode and /plan
Implementation plans
Walkthroughs, screenshots, and recordings
Inline feedback on artifacts
The plan review policy
Reviewing diffs in the Review pane
Staging and committing changes

Browser Use and Verification

The browser subagent and /browser
Checking responsive layouts
Console errors and failed requests
Screenshots and recordings as evidence
URL allow, ask, and deny lists
Protections for sign-ins and payments

Parallel Agents and Subagents

Parallel conversations across projects
Built-in research and browser subagents
Custom subagents in .agents/agents/
Shared workspace vs. isolated worktree
Agent teams with /swarm
Dynamic Workflows with /run-workflow
Automations for scheduled work
Remote Control and Remote Agents

The Antigravity CLI

Interactive sessions with agy
Approvals and /diff
Execution modes
Resuming and forking conversations
Working over SSH
Antigravity in editors with agy acp

Project Context: Rules and Agent Skills

AGENTS.md and GEMINI.md
Rules in .agents/rules/
Rule activation triggers
Global rules under ~/.gemini/
Agent Skills in .agents/skills/
Invoking skills as slash commands
Migrating workflows to skills

MCP, Plugins, and Hooks

MCP servers in mcp_config.json
Workspace and global MCP scopes
stdio and Streamable HTTP
OAuth and Google credentials
Google Cloud servers from the MCP Store
Plugins and the Antigravity Marketplace
Managing plugins with agy plugin
Lifecycle hooks in .agents/hooks.json
Guardrails with PreToolUse hooks

CI and Automation

Headless runs with agy -p
Text, JSON, and streaming output
Structured output with --json-schema
Scoped allow rules for unattended runs
Risks of --dangerously-skip-permissions
Authenticating automation
The run-gemini-cli GitHub Action
Custom pipelines with the Antigravity SDK

Security, Governance, and Cost

Permission presets
Allow, ask, and deny rules
The terminal sandbox
Filesystem and network allowlists
Antigravity in Gemini Enterprise
Data governance and third-party models
Quotas and AI credits
AI Credit overage settings

Migrating from Gemini CLI and Gemini Code Assist

Who moved to Antigravity
Who keeps Gemini CLI
Carrying over GEMINI.md and AGENTS.md
Importing extensions with agy plugin import
Moving MCP servers and skills
Known feature gaps
Running both tools side by side