agy-customizations

Explains the Antigravity customization system for skills, rules, plugins, hooks, and MCP servers.

Updated Jan 8, 2026
One-click install
npx skills add https://github.com/arslan9024/White-Caves --skill agy-customizations-arslan9024
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agy-customizations
Source: https://github.com/arslan9024/White-Caves/tree/main/.agents/skills/agy-customizations
Command: npx skills add https://github.com/arslan9024/White-Caves --skill agy-customizations-arslan9024

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Understanding how to tailor an AI agent's behavior across skills, rules, plugins, hooks, and MCP servers is confusing without a single reference covering discovery locations, loading priority, and configuration formats. This Skill consolidates the entire Antigravity Customization System into one guide so you can correctly create, register, and debug agent customizations. ## Core Features & Use Cases - Customization Type Selection: Compares rules, skills, plugins, hooks, and MCP servers side-by-side so you pick the right mechanism for your goal. - Discovery & Priority Reference: Documents workspace, hierarchical, and global discovery paths plus the five-level loading precedence that resolves naming conflicts. - Configuration Guides: Details hooks.json event contracts, mcp_config.json stdio/SSE transports, plugin.json manifests, and skills.json inheritance schemas. - Use Case: You want to enforce a lint check after every shell command the agent runs. Use this Skill to learn the PostToolUse hook contract, then write a hooks.json entry with a matcher and command script. ## Quick Start Ask the agent to explain how to create a new skill with a SKILL.md file and register it via skills.json for team sharing.

Frequently Asked Questions about agy-customizations

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create a custom skill for the Antigravity agent?▼

Create a directory under .agents/skills/ containing a SKILL.md file with YAML frontmatter defining a name and description. The body holds step-by-step instructions, and you can add scripts/, references/, or resources/ subdirectories for supporting files.

What is the difference between rules, skills, and plugins in Antigravity?▼

Rules (GEMINI.md/AGENTS.md) enforce guidelines hierarchically, skills are on-demand runbooks loaded only when activated, and plugins bundle skills, rules, hooks, and MCP configs into one shareable unit. Choose based on scope and distribution needs.

How do lifecycle hooks work in hooks.json?▼

Hooks run shell commands at agent lifecycle events like PreToolUse, PostToolUse, PreInvocation, and Stop. Each handler receives JSON on stdin and returns a JSON decision on stdout, such as allow, deny, or ask for tool gating.

Can I share agent customizations with my team via git?▼

Yes. Commit customizations to a shared repo directory and add a skills.json or plugins.json in .agents/ pointing to it with a workspace-relative path. Teammates cloning the repo get the customizations loaded automatically.

How do I connect an MCP server to the agent?▼

Define the server in mcp_config.json under mcpServers, using command/args/env for local stdio servers or serverUrl for remote SSE endpoints. Discovered tools are injected into the agent's toolset automatically.

What are the limitations of Antigravity hooks?▼

Hooks currently support only command-type shell execution, with no HTTP or prompt hooks. They also run synchronously and block the agent loop, so long-running commands delay execution.