Antigravity skills: 740 worth installing

Gives an agent UI and UX design guidance for web, mobile and desktop: styles, palettes, typography, accessibility and layout.

Interviews you relentlessly about a plan or design until every open decision has been resolved.

A full software development workflow for coding agents, built from composable skills for brainstorming, planning, TDD, debugging and review.

Switches the agent to an ultra-compressed way of talking that cuts output tokens while keeping technical accuracy, in several levels.

Creates new skills, improves existing ones and measures how well they perform with evals and benchmarks.

Steers an agent away from templated-looking frontends: it reads the brief, picks a design direction and builds landing pages and redesigns.

Gives the agent a clear aesthetic direction when building interfaces, with guidance on typography and avoiding templated-looking defaults.

Designs, critiques, audits and polishes frontend interfaces through a set of commands, from shaping a brief to final refinement.

Rewrites AI-sounding text so it reads like a person wrote it, removing common AI tells without changing the meaning.

Helps you discover and install agent skills when you ask how to do something or whether a skill exists for it.

Routes an agent to the right Remotion skill for the task and applies Remotion's rules when writing video code.

Researches what people have said about a topic in the last 30 days across Reddit, X, YouTube, TikTok, Hacker News and the web.

A browser automation CLI for AI agents: navigate pages, fill forms, click buttons, take screenshots and extract data.

Operates Google's NotebookLM (the skill calls it Gemini Notebook) through the notebooklm-py CLI: manage notebooks and sources, ask grounded questions and generate artifacts.

Explores intent, requirements and design with you before any feature or component is implemented.

Works out why a superpowers session went wrong: repeated work, ignored plans or a skill that did not fire.

Splits two or more independent tasks across parallel agents when they share no state.

Executes an implementation plan in the current session, step by step.

Guides the choice of how to integrate finished work once tests pass: merge, pull request or cleanup.

Handles code review feedback with technical rigour, verifying suggestions before implementing them.

Requests a code review to verify that completed work meets its requirements before merging.

Makes the agent find the root cause of a bug, test failure or unexpected behaviour before it proposes a fix.

Makes the agent write a failing test before any implementation code, for features and bug fixes alike.

Sets up an isolated workspace with git worktrees before feature work or plan execution begins.