Back to directory

Antigravity Skill Vault

A curated collection of more than three hundred Agent Skills for Google Antigravity, ported from the Claude Code Agents ecosystem, covering development, infrastructure, security and workflow tasks.

Antigravity Skill Vault

Antigravity Skill Vault ports the Claude Code Agents ecosystem into the Agent Skills format for Google Antigravity, giving its agent repeatable workflows, domain expertise and specialized tools across more than three hundred skills. Skills fall into three types: domain skills that package specialized knowledge, such as a Kubernetes manifest generator or async Python patterns; specialist agents, persona-based instruction sets for complex reasoning such as a backend architect or security auditor; and commands and workflows, structured multi-step procedures such as full-stack feature orchestration.

The skills are flattened into a single directory but organized by broad category, covering languages and frameworks such as Python, TypeScript, Rust, Go and React Native, infrastructure and operations topics such as Kubernetes, cloud architecture, Terraform and CI/CD pipelines, security and quality topics such as auditing, refactoring and test generation, and broader workflow and architecture topics. It is written in JavaScript and licensed under MIT.

More in Agent skills

See all 581
  • Provenance gate

    github.com

    A deterministic authorization gateway that sits between an AI agent and its tools, tracing where each parameter of a side-effect call came from, and blocking the call or asking a human when the source does not check out.

  • A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.

  • Vitruvius

    github.com

    An engineering research agent that runs a discover, read, synthesize, verify and review loop over technical questions, reading sources directly rather than inferring from titles, and publishing its own scored verifier benchmark, weaknesses included.