wtf is agent-skills?

This project is a public collection of ready-made skills, review agents, and rule files for AI coding assistants like Cursor. It is mirrored from a private repository, sharing only the parts the author has approved for public use. The idea is that a cloud machine or laptop can install this package instead of needing access to the private setup, since it copies secrets and environment configuration only, not the actual skill content. You install it with a single command through a skills installer, either grabbing the whole set or picking a single skill by name. Once installed, the files land in specific folders on your machine so your AI assistant can read them. The repo bundles three kinds of files. Skills are things like brainstorming techniques, drawing diagrams, writing session handoff notes, doing research, refactoring code safely, removing old features, finding overly complex edge case handling, running a review and ship process, checking technical SEO, cleaning up issues and pull requests across repos, and explaining code differences in an interactive way. Review agents are specialized checkers meant to run during a pull request review. They look for things like accessibility issues, bugs, code quality problems, database migration risks, dependency and security scanning, error handling gaps, and test coverage gaps. There is also a confidence scorer that helps decide how serious a single finding really is. The cited rules are shorter reference documents about coding style, using git worktrees for new work, strict TypeScript expectations, and rules for when an agent is allowed to touch production systems. This is meant for teams or individuals who already use AI coding agents and want a consistent, reusable library of prompts and checks rather than writing them from scratch each time. The README notes this repo is not where new content should be written, it is only published from elsewhere.