Java World Knowledge Sharing Portal: Antigravity Kit powerful open-source enhancement for Google Antigravity

The Antigravity Kit vudovn is a powerful open-source enhancement for Google Antigravity, which is Google's "vibe coding" (agent-first) development environment.

Essentially, while Google Antigravity provides the basic kitchen (the editor, terminal, and browser control), the Antigravity Kit provides the "Michelin-star chefs"—a pre-configured team of specialized AI agents, expert skills, and automated workflows.

1. The Core Components

The kit organizes its intelligence into three distinct layers located in the .agent/ directory:

Specialist Agents (20+): Instead of one generic AI, you get a roster of experts. Examples include:
- SEO Specialist: Optimizes metadata and Core Web Vitals.
- Debugger Agent: Specifically tuned to hunt and fix runtime errors.
- Game Designer: Handles physics, logic, and visual effects for games.
- Security Auditor: Scans for secrets and vulnerabilities.
Skills (35+): These are "knowledge packs" that the agents can load as needed. They contain specific instructions for technologies like Docker, React, Supabase, or MCP (Model Context Protocol).
Workflows (11+): These are automated sequences triggered by "slash commands" (e.g., /create, /review, /brainstorm). A single command can trigger a chain: Brainstorm → Plan → Code → Test.

2. Key Features & "Vibe Coding"

The kit is designed for a new style of development where the human acts more like a Director than a writer.

Verifiable Execution: The agents don't just write code; they provide "Artifacts." They generate a plan first, then use a built-in browser to take screenshots or videos of the running app to prove it works.
Self-Healing Loops: If the agent runs a command in the terminal and it fails, the kit uses its "Debugger" and "Linter" skills to automatically read the error, reason through a fix, and retry.
Model Agnostic: While it shines with Google’s Gemini 3 Pro, it is fully compatible with Claude 3.5 Sonnet and Claude 4.5, allowing you to switch "brains" depending on the complexity of the task.

3. How the Architecture Works

The repository follows a strict "Trust-Grade" governance model. All agent operations are scoped to the .agent/ folder to ensure they don't overwrite your sensitive project files without explicit permission.

Feature	Description
Slash Commands	`/plan`, `/review`, `/deploy` to trigger complex automation.
Intelligent Routing	Automatically loads the right agent based on keywords (e.g., "login" triggers the Security Agent).
Context Management	Keeps track of "Context as Artifact"—persistent markdown files that record decisions and architectural rules.
Browser Subagent	Allows the AI to interact with the web, click buttons, and perform E2E testing.

4. Getting Started

The kit is typically installed via npx to instantly configure an existing project:

npx @vudovn/ag-kit init

This command sets up the directory structure and populates the .agent/ folder with the templates needed to turn Google Antigravity into a production-ready IDE.

Generally we developer get following type of BAU work

Work Category:-

New Feature Development
Change Request (CR)
Bugs & Defects
Security Enhancement
Security Issues (Hotfixes)
Performance Enhancement
Performance Issues
UI/UX Refactoring
Middleware/API Updates
Technical Debt/Refactoring
DevOps & Infrastructure.

As Antigravity Kit is a comprehensive framework of AI agent templates designed to streamline software development through specialized personas, domain-specific skills, and automated workflows.It functions by automatically detecting the domain of a user's request and applying the most relevant "specialist" agents and "skills" to ensure expert-level implementation and validation.

Below is a detailed explanation of how this project assists across various work categories.

Project Overview

The kit consists of four primary pillars:

Agents (20): Specialist AI personas like @backend-specialist, @security-auditor, and @debugger that handle domain-specific tasks.
Skills (37): Knowledge modules containing best practices, patterns (e.g., api-patterns), and validation scripts.
Workflows (11): Slash commands (e.g., /create, /debug, /plan) that execute multi-step procedures for common development tasks.
Validation Layer: Automated scripts like checklist.py and verify_all.py that perform security scans, UX audits, and test execution before delivery.

Application by Work Category

1. New Feature Development & Change Requests (CR)

Workflow: Use /plan to break down requirements into milestones . and /create to scaffold new features based on established templates.
Agent Utility: The @project-planner handles initial discovery ., while @frontend-specialist and @backend-specialist implement the logic using domain-specific skills like app-builder.
Socratic Gate: For new features, the system triggers 3 strategic questions to ensure scope and purpose are clear before coding begins.

2. Bugs, Defects & Performance Issues

Workflow: Invoke /debug for a systematic root-cause analysis.
Agent Utility: The @debugger uses the systematic-debugging skill to analyze stack traces and logs, while the @performance-optimizer applies performance-profiling to resolve bottlenecks.
Validation: After a fix, the /test workflow generates and runs tests to ensure no regressions occurred.

3. Security Enhancements & Hotfixes

Agent Utility: @security-auditor and @penetration-tester are automatically engaged for security-related tasks.
Skills: They utilize vulnerability-scanner and red-team-tactics to identify and mitigate risks.
Automated Checks: The validation layer runs security scans as part of every execution to catch vulnerabilities early.

4. UI/UX Refactoring

Workflow: /ui-ux-pro-max provides access to 50 design styles and 21 color palettes to overhaul interfaces professionally.
Skills: Uses frontend-design, web-design-guidelines, and ux-psychology to ensure accessibility and aesthetic quality.
Validation: Includes a ux_audit.py script to check for common design flaws.

5. Middleware/API Updates

Agent Utility: @backend-specialist manages API architecture.
Skills: The api-patterns skill provides standardized versions for REST, GraphQL, and tRPC, ensuring updates follow industry best practices
Validation: Uses api_validator.py to ensure schema compliance and performance.

6. Technical Debt & Refactoring

Workflow: /enhance is designed specifically to improve existing code quality.
Agent Utility: The @code-archaeologist analyzes legacy codebases to understand context before refactoring.
Skills: Knowledge modules like clean-code and architecture guide the restructuring of code to reduce complexity.

7. DevOps & Infrastructure

Workflow: /deploy automates deployment procedures across platforms like AWS, Vercel, or Docker.
Agent Utility: @devops-engineer handles infrastructure as code and deployment scripts.
Skills: Utilizes deployment-procedures and docker-expert to manage builds and environment configurations.

But we will not use this predefined workflow but we will inform Antigravity agent itself to find and utlize the best of avaliable workflow for the given use case.

Lets first install the required package using npx command.

npx @vudovn/ag-kit init

Now lets use our PRD.txt as a product requirement document and place it in folder document inside our project.

Now lets ask the Antigravity Management window to provide the workflow for the given scenario