A collection of Agentic AI Skills—curated instructions, static analysis modules, and executable testing logic—designed to be injected into Autonomous AI Agents (like Roo, Claude Desktop, Cursor, or Windsurf).
Instead of relying solely on raw LLM prompts, these skills use a Tool-Augmented Orchestration pattern: providing the agents with the scripts and structured context they need to perform high-level software engineering tasks accurately and autonomously.
A state-of-the-art accessibility skill that achieves near 85% automated WCAG 2.1 Level AA coverage.
- What it does: Orchestrates static code LLM reasoning alongside dynamic headless browser testing (Axe-core + viewport reflow + focus traps).
- Auto-Fix: Capable of instantly acting on its own report to write semantic, accessible code replacements (Next.js, Nuxt/Vue, React).
This repository is a Claude Code plugin marketplace. Add the marketplace once, then install any skill on demand:
/plugin marketplace add Brunopaun/awesome-agent-skills
/plugin install ada-review@awesome-agent-skills
Installed plugins live in ~/.claude/plugins/cache/ and are available across every project on the machine. Update with /plugin update.
Each skill is self-contained in ./<skill>/skills/<skill>/. Copy that folder into your agent's context directory (e.g., .cursor/rules/).
For detailed instructions, refer to the README inside each specific skill folder.