Video tutorial · The Metaverse Guy
Google Anti-Gravity Agent Skills: Add Custom AI Skills to Your Code Editor
I walk through setting up custom agent skills in Google Anti-Gravity, using NanoBanana Pro for image generation, and integrating TestSprite for automated testing.
This video was made in partnership with TestSprite. It carries YouTube's paid-promotion label.
- Published
- 17 January 2026
- Length
- 8 min
- Views
- 8,712 views
Google Anti-Gravity now supports Agent Skills, a feature that lets you define reusable AI behaviours directly in your project. In this video, I show how to structure the .agent folder, write a skills.md file, and chain skills together. I also demonstrate generating images with NanoBanana Pro and using TestSprite to catch bugs automatically.
What Are Agent Skills?
Agent Skills are a new feature in Google Anti-Gravity that let you define custom behaviours for the AI coding assistant. Unlike other editors like Cursor or Windsurf, Anti-Gravity allows you to package reusable instructions that the agent can follow across different parts of your project.
Setting Up Your First Skill
To add a skill, you create an .agent folder in your project root and place a skills.md file inside it. The file needs to follow a specific format that Anti-Gravity understands. I show the exact structure in the video, so you can copy it for your own projects.
Generating Images with NanoBanana Pro
One practical use of Agent Skills is generating real images for landing pages. I use NanoBanana Pro, available through Google AI Studio, to create high-quality visuals directly inside the frontend code. The skill tells the agent when and how to call the model, so you get consistent results every time.
Chaining Skills for Better Results
You can combine multiple skills to boost the AI's performance. For example, one skill might handle layout generation while another manages image placement. By chaining them, the agent produces more polished, production-ready code without extra prompting.
Automated Testing with TestSprite
I also integrate TestSprite, an AI testing tool, to automatically validate the code and fix bugs. Once your skills are set up, TestSprite can run checks and suggest fixes, helping you ship more reliable applications. This step is completely optional but saves a lot of manual QA time.
Tools used in this video
- Google Anti-Gravity
- NanoBanana Pro (via Google AI Studio)
- TestSprite
Questions
- Do I need to pay for Google Anti-Gravity Agent Skills?
- No, the Agent Skills feature is free to use within Google Anti-Gravity. You only need a Google account and access to the editor.
- What is the .agent folder and skills.md file?
- The .agent folder is a directory you create in your project root. Inside it, a skills.md file defines the custom instructions that the AI agent will follow. The file must be formatted correctly, which I demonstrate in the video.
- Can I use other image generation models besides NanoBanana Pro?
- The video focuses on NanoBanana Pro because it integrates easily with Google AI Studio, but the same skill structure could be adapted for other models if they have an API.
- Is TestSprite free to use?
- TestSprite offers a free tier that you can try using the link in the description. It provides AI-powered testing and bug fixing that works alongside Anti-Gravity.
Topics
Chapters
- 00:00 Add Agent Skills to Google Anti-Gravity
- 00:19 Official Skills Documentation Overview
- 00:31 Setting Up the .agent Folder Structure
- 01:07 Creating Your First skills.md File
- 01:18 Using AI to Generate Custom Agent Skills
- 01:45 Building an Image Generation Skill
- 02:11 Using NanoBanana Pro for High-Quality Images
- 02:29 Getting API Keys from Google AI Studio
- 02:50 Generating Skills Content Automatically
- 03:12 Applying Skills to a Real Landing Page
- 03:41 AI Generates Images for the Website
- 04:08 Creating Advanced Design & Color Skills
- 04:50 Boosting AI Code Quality with Agent Skills
- 05:00 Introducing TestSprite for AI Testing
- 05:19 Integrating TestSprite MCP Server
- 05:41 Running Frontend Tests Automatically
- 06:00 Creating a Product Specification Document
- 06:39 Uploading Specs & Running Tests
- 07:00 Viewing Failed Tests & Screen Recordings
- 07:15 Fixing Bugs Automatically with AI
- 07:36 Final Results & Wrap-Up
Video description and links
This article was written from the video's description with the help of AI. Watch the video for the full walkthrough.



