Awesome Agent Skills

Awesome Agent Skills is a curated list of Agent Skills, reusable capability packages that extend what AI coding assistants like Claude Code, Codex, Cursor, and Gemini CLI can do. Instead of aggregating auto-generated skill dumps, the maintainers hand-pick skills published by real engineering teams (Anthropic, Stripe, Vercel, Cloudflare, Netlify, Sentry, Figma, Hugging Face, and others) alongside vetted community contributions. It's aimed at developers who use AI coding agents and want proven, team-authored skills rather than sifting through low-quality, mass-generated ones.

The repo is essentially a big README organized into collapsible sections by publisher, each linking to the skill's source or a documentation page on officialskills.sh. It works as a browsable catalog: you find a skill by vendor or task (testing, documents, design, DevOps, marketing, security), then copy it into your agent's skills directory.

Key features

• Curated, not generated: skills are hand-picked from official teams and reviewed community submissions, filtering out low-effort AI-slop skill repos.
• Broad vendor coverage: includes skills published by Anthropic, Stripe, Vercel, Cloudflare, Netlify, Sentry, Figma, Hugging Face, Trail of Bits, Microsoft, OpenAI, and dozens more.
• Cross-tool compatibility: skills work with Claude Code, Codex, Antigravity, Gemini CLI, Cursor, GitHub Copilot, OpenCode, Windsurf, and other agent runtimes.
• Task-oriented organization: skills are grouped by publisher and cover document editing, test automation, frontend design, security auditing, marketing, and product management.
• Quality standards section: the repo defines criteria for what counts as a legitimate, well-built skill before it's added.
• Companion site: officialskills.sh provides a browsable, searchable version of the same catalog with per-skill documentation pages.
• Community-driven growth: contribution is open, and the list is described as the most actively contributed Agent Skills repository.

Ideal use cases

This repo fits developers already using an AI coding agent (Claude Code, Cursor, Codex, etc.) who want to extend it with domain-specific abilities, such as generating Playwright tests, editing Office documents, building MCP servers, or applying brand guidelines to generated artifacts. It's useful for teams standardizing on skill sets across engineering, QA, and design workflows, and for anyone evaluating what official vendor-published skills exist before writing a custom one.

It's not the right fit if you're looking for a skill-authoring framework or runtime itself, this repo is a directory, not a tool that installs or manages skills for you. It also won't help if your agent doesn't support the Agent Skills format, or if you need guaranteed long-term maintenance guarantees on every listed skill, since many entries point to third-party repos maintained by their original publishers, not by this project.

Installation

There's no package to install; the repository is a curated reference list. To use a skill, browse the README or officialskills.sh, find the skill you need, and follow the link to its source repository or documentation page. From there, the typical flow is:

1. Locate the skill entry, e.g. anthropics/pdf or testmu-ai/playwright-skill.
2. Open the linked source repo or officialskills.sh page for setup instructions specific to that skill.
3. Copy or clone the skill files into your agent's skills directory (the exact path depends on your tool, Claude Code, Cursor, Codex, etc.).
4. Restart or reload your agent so it picks up the new skill.

If you want to contribute a skill, check the quality standards section of the README, then open a pull request adding your entry to the relevant category table. The project is MIT licensed, so both the list itself and most linked skills are free to use and adapt, though individual skill licenses may vary by publisher.