Antigravity Awesome Skills: The Complete Guide to 1,200+ Agentic Skills for AI Coding Assistants
AI coding assistants like Claude Code, Cursor, and Gemini CLI are remarkably capable — but they're generalists. They don't know your company's deployment protocol, your preferred testing framework conventions, or the specific syntax for AWS CloudFormation templates. Antigravity Awesome Skills solves this by providing a massive, curated library of 1,200+ agentic skills that teach your AI assistant how to perform specific tasks perfectly, every time.
Think of skills as micro-brains — small markdown files that give your AI assistant deep expertise in specific domains. Install them once, and your AI transforms from a generalist into a specialized expert across architecture, security, DevOps, testing, content creation, and dozens more categories.
With 20,700+ GitHub stars, 3,600+ forks, 68 releases, and official skills from Anthropic, OpenAI, Google, Microsoft, Supabase, Vercel, and Apify, this is the most comprehensive agentic skills collection available.
Antigravity Awesome Skills on GitHub
What Are Agentic Skills?
Agentic skills are modular capabilities that extend an AI agent's functionality. Each skill is a directory containing a SKILL.md file — a structured markdown document with YAML frontmatter that defines:
- Name and description — What the skill does
- Instructions — Step-by-step guidance for the AI
- Risk level — Safety classification
- Source attribution — Where the skill came from
- Optional resources — Scripts, templates, examples
When you install skills and reference them in your AI assistant, the agent reads the SKILL.md and follows its instructions. This is fundamentally different from a simple prompt — it's a persistent, reusable capability that the AI applies automatically when relevant.
skills/
├── aws-cost-optimizer/
│ └── SKILL.md
├── react-patterns/
│ └── SKILL.md
├── security-audit/
│ ├── SKILL.md
│ └── scripts/
│ └── scan.sh
└── docker-expert/
└── SKILL.md
Universal Compatibility — 10 AI Tools Supported
What makes this collection unique is its universal compatibility. The same skills work across all major AI coding assistants:
| Tool | Developer | Skill Path | Invocation |
|---|---|---|---|
| Claude Code | Anthropic | .claude/skills/ | >> /skill-name help me... |
| Gemini CLI | Google DeepMind | .gemini/skills/ | Use skill-name... |
| Codex CLI | OpenAI | .codex/skills/ | Auto-loads on demand |
| Kiro CLI/IDE | AWS | ~/.kiro/skills/ | /skill-name or auto |
| Antigravity IDE | Google DeepMind | ~/.gemini/antigravity/skills/ | @skill-name (in Chat) |
| GitHub Copilot | Microsoft | .cursor/skills/ | Agent Mode |
| Cursor | Cursor | .cursor/skills/ | Paste content |
| OpenCode | Open-source | .agents/skills/ | opencode run @skill-name |
| AdaL CLI | Open-source | .adal/skills/ | Auto-loads on demand |
This "write once, use everywhere" approach means skill authors contribute once and the entire ecosystem benefits.
Quick Start — 1 Minute Setup
Option A: npx (Recommended)
# Default: Antigravity global
npx antigravity-awesome-skills
# Claude Code
npx antigravity-awesome-skills --claude
# Gemini CLI
npx antigravity-awesome-skills --gemini
# Cursor
npx antigravity-awesome-skills --cursor
# Codex CLI
npx antigravity-awesome-skills --codex
# Kiro CLI
npx antigravity-awesome-skills --kiro
# Custom path
npx antigravity-awesome-skills --path ./my-skills
Run npx antigravity-awesome-skills --help for all options. If the directory already exists, the installer runs git pull to update.
Option B: git clone
# Antigravity global (matches npx default)
git clone https://github.com/sickn33/antigravity-awesome-skills.git ~/.gemini/antigravity/skills
# Claude Code
git clone https://github.com/sickn33/antigravity-awesome-skills.git .claude/skills
# Gemini CLI
git clone https://github.com/sickn33/antigravity-awesome-skills.git .gemini/skills
# Workspace-specific
git clone https://github.com/sickn33/antigravity-awesome-skills.git .agent/skills
Option C: Kiro IDE Import (GUI)
For Kiro IDE users, import individual skills through the Agent Steering & Skills panel → + → Import → GitHub.
Using Skills
Once installed, just ask your AI assistant naturally:
"Use the @brainstorming skill to help me plan a SaaS."
"Run @lint-and-validate on this file."
"Apply @react-patterns to refactor this component."
Features & Skill Categories
The 1,200+ skills are organized into specialized domains covering the entire software development lifecycle:
🏗️ Architecture & Design
senior-architect— System design with C4 model diagramsc4-context— C4 architecture context diagrams- Domain-driven design, microservices patterns
🤖 AI & Machine Learning
rag-engineer— RAG pipeline design and optimizationprompt-engineer— Prompt engineering best practiceslanggraph— LangGraph workflow patterns
💻 Development
typescript-expert— Advanced TypeScript patternspython-patterns— Pythonic design patternsreact-patterns— Modern React architecture (official from Vercel Labs)
✍️ Content & Marketing
copywriting— Conversion-focused copypricing-strategy— SaaS pricing optimizationseo-audit— Technical SEO analysis
🔒 Security
api-security-best-practices— API hardeningsql-injection-testing— SQL injection detectionvulnerability-scanner— Automated vulnerability scanning
🧪 Testing
test-driven-development— TDD workflowstesting-patterns— Testing strategy patternstest-fixing— Intelligent test failure analysis
☁️ DevOps & Infrastructure
docker-expert— Container optimizationaws-serverless— AWS Lambda/SAM patternsvercel-deployment— Vercel deployment best practices
⚡ Automation
workflow-automation— Process automationinngest— Inngest event-driven workflowstrigger-dev— Trigger.dev job orchestration
📝 Documentation & Writing
brainstorming— Structured ideationdoc-coauthoring— Collaborative documentationwriting-plans— Content planning
Curated Bundles — Starter Packs by Role
With 1,200+ skills, picking the right ones can be overwhelming. Curated bundles group skills by role or goal:
| Bundle | Focus |
|---|---|
| Web Wizard | Full-stack web development |
| Security Engineer | Security auditing and hardening |
| OSS Maintainer | Open-source project management |
| DevOps Pro | Infrastructure and deployment |
| AI Builder | AI/ML pipeline development |
Important: Bundles are NOT separate installations. They're curated lists that help you identify which installed skills to focus on for your specific role.
Workflows — Step-by-Step Execution
While bundles help you choose skills, workflows help you execute them in order. They're step-by-step playbooks for concrete goals:
Ship a SaaS MVP
A structured workflow that chains skills like brainstorming → architecture → development → testing → deployment.
Security Audit for a Web App
Chains vulnerability scanning → SQL injection testing → API security → compliance checking skills in a logical sequence.
Build an AI Agent System
Guides you through designing, implementing, and testing an AI agent system using RAG, prompt engineering, and automation skills.
QA and Browser Automation
Combines testing patterns with browser automation skills, including optional Playwright integration for Go stacks.
Workflows are available in both human-readable (docs/WORKFLOWS.md) and machine-readable (data/workflows.json) formats.
Official Skills from Industry Leaders
What sets Antigravity Awesome Skills apart from other collections is the inclusion of official skills from major technology companies:
| Source | Skills Included |
|---|---|
| Anthropic (anthropics/skills) | Document manipulation (DOCX, PDF, PPTX, XLSX), Brand Guidelines, Internal Communications |
| OpenAI (openai/skills) | Codex agent skills, Skill Creator, Concise Planning |
| Google Gemini (google-gemini/gemini-skills) | Gemini API, SDK and model interactions |
| Microsoft (microsoft/skills) | Azure cloud, Bot Framework, Cognitive Services, enterprise patterns (.NET, Python, TS, Go, Rust, Java) |
| Vercel Labs (vercel-labs/agent-skills) | React Best Practices, Web Design Guidelines |
| Supabase (supabase/agent-skills) | Postgres Best Practices |
| Apify (apify/agent-skills) | Web scraping, data extraction, automation |
| Remotion (remotion-dev/skills) | Video creation in React (28 modular rules) |
This means you're not installing random community prompts — you're leveraging skill definitions vetted and maintained by the same teams that build the underlying products.
Deep Dive: The SKILL.md Format
Every skill follows a standardized structure defined in the project's SKILL_ANATOMY.md:
Frontmatter (YAML):
---
name: react-patterns
description: Modern React architecture patterns and best practices
risk: low
source: vercel-labs/agent-skills
date_added: "2026-01-15"
---
Body (Markdown): After the frontmatter, the skill body contains instructions, examples, and quality checks:
- Instructions — Step-by-step guidance (e.g., "Use functional components with hooks," "Prefer React Context for shared state," "Use Server Components where possible")
- Examples — Real code snippets the AI uses as reference patterns
- Quality Checks — Checklists the AI validates against (e.g., single responsibility, no prop drilling, error boundaries)
Here's an example of a code snippet from the react-patterns skill:
// Good: Server Component with async data fetching
export default async function ProductList() {
const products = await getProducts();
return <ProductGrid items={products} />;
}
The frontmatter provides metadata for indexing and the interactive web app, while the body contains the actual skill instructions.
Contributing New Skills
The project has a well-defined contribution process:
- Fork the repository
- Create a new directory inside
skills/for your skill - Add a
SKILL.mdwith required frontmatter (name, description, risk, source) - Follow the quality bar defined in
docs/QUALITY_BAR.md - Validate with
npm run validate(ornpm run validate:strictfor CI) - Submit a Pull Request
The project enforces quality through automated validation scripts, an audit process documented in docs/AUDIT.md, and a full maintenance chain described in .github/MAINTENANCE.md.
Real-World Impact
For Individual Developers
Skills eliminate the cognitive overhead of remembering best practices. Instead of googling "React Server Components patterns" or "Docker multi-stage build optimization," you invoke a skill and get expert-level guidance in context.
For Teams
Skills standardize how your team works with AI. Everyone uses the same deployment-protocol skill, the same testing-patterns skill, and the same security-audit skill — ensuring consistent quality across the codebase.
For Organizations
The bundle and workflow system means you can define organization-specific skill sets for different roles. A new frontend developer gets the "Web Wizard" bundle; a security engineer gets the "Security Engineer" bundle. Onboarding becomes instant.
Antigravity Awesome Skills vs Alternatives
| Feature | Antigravity Awesome Skills | Prompt Libraries | Custom CLAUDE.md | Agency Agents |
|---|---|---|---|---|
| Skills Count | 1,200+ | ~100-500 | ~10-50 rules | ~55 agents |
| Official Sources | ✅ (9 vendors) | ❌ | ❌ | ❌ |
| Multi-Tool Support | 10 tools | 1-2 tools | 1 tool | 1 tool |
| Curated Bundles | ✅ | ❌ | ❌ | ❌ |
| Workflows | ✅ | ❌ | ❌ | ❌ |
| npx Installer | ✅ | ❌ | ❌ | ❌ |
| Auto-Update | ✅ (git pull) | Manual | Manual | Manual |
| Validation CI | ✅ | ❌ | ❌ | ❌ |
When to choose Antigravity Awesome Skills: You want the most comprehensive, battle-tested skill library with official vendor backing, cross-tool compatibility, and structured workflows.
When to choose custom CLAUDE.md: You need simple, project-specific rules without the overhead of a full skill library.
When to choose Agency Agents: You want rich, personality-driven agent personas rather than modular technical skills.
FAQ
How many skills are included?
1,200+ skills across 35+ categories, sourced from 35+ community repositories and 9 official vendor sources.
Which AI tools are supported?
Claude Code, Gemini CLI, Codex CLI, Kiro CLI/IDE, Antigravity IDE, GitHub Copilot, Cursor, OpenCode, and AdaL CLI.
How do I install them?
The fastest way is npx antigravity-awesome-skills with a flag for your tool (e.g., --claude, --cursor, --gemini). You can also git clone directly.
How do I update?
Run npx antigravity-awesome-skills again — if the directory exists, it automatically runs git pull. Or navigate to the skills directory and run git pull manually.
Are these safe to use?
Each skill has a risk level in its frontmatter. The project includes security guardrails documented in docs/SECURITY_GUARDRAILS.md. Official vendor skills are maintained by the respective companies.
Can I contribute my own skills?
Yes. Fork the repo, add a SKILL.md in a new skills/ subdirectory, validate with npm run validate, and submit a PR.
What are bundles?
Curated groups of skills organized by role (Web Wizard, Security Engineer, etc.). They help you focus on the skills most relevant to your work without browsing 1,200+ individual skills.
Is this free?
Yes. MIT licensed — free for commercial and personal use. Optional donations via Buy Me a Coffee support ongoing maintenance.
Conclusion
Antigravity Awesome Skills represents the most ambitious effort to standardize and curate agentic skills for AI coding assistants. With 1,200+ skills, official backing from 9 major vendors, compatibility with 10 AI tools, and a structured approach including bundles, workflows, and validation — it transforms any AI assistant from a capable generalist into a domain-specific expert.
The universal SKILL.md format is the key innovation: it creates an interoperable skills ecosystem where a skill written for Claude Code works identically in Gemini CLI, Cursor, or Codex. This is the emerging standard for how developers will customize their AI assistants.
With 20,700+ GitHub stars, 68 releases, and a rapidly growing community, the project has strong momentum. Whether you're a solo developer looking to supercharge your AI assistant or a team lead standardizing AI-assisted workflows, Antigravity Awesome Skills is the definitive starting point.
Explore Antigravity Awesome Skills on GitHub