awesome agent skills

Abstract

A curated collection of 1000+ agent skills from official dev teams and the community, compatible with Claude Code, Codex, Gemini CLI, Cursor, and more.

The Problem

Engineering teams that build AI agents need a reliable source of vetted, production‑grade “skills” (pre‑packaged tool wrappers). Public lists are often noisy, auto‑generated, or out‑of‑date, forcing teams to spend time vetting each entry before they can be reused.

What This Does

awesome-agent-skills is a documentation‑only repository that aggregates >1 500 curated skill definitions contributed by official dev teams (Claude, Google Gemini, Stripe, etc.) and the community. The repo consists of four top‑level files:

README.md – overview, sponsor info, and a markdown table linking to individual skill files (hosted elsewhere).
CONTRIBUTING.md – guidelines for adding new skill entries.
LICENSE – MIT license text.
.gitignore – generic ignore rules.

All skill descriptors live in external repositories or sub‑folders referenced by the README; the repo itself does not contain executable code.

How It Is Wired

Because the repository contains only markdown and static assets, there is no runtime wiring, entry point, or call graph to map. The only “effect” is the distribution of markdown links that point to skill definition files (e.g., skills/claude/…md). Consequently:

File	Responsibility	External Effect
`README.md`	Central index; links to skill docs	Consumers clone or view the list to discover skill URLs
`CONTRIBUTING.md`	Process for adding new entries	Guides external contributors; no code execution
`LICENSE`	Legal terms	None
`.gitignore`	VCS ignore rules	None

No code path touches a database, network API, or filesystem beyond the standard Git operations.

How To Use It

Clone the repository

   git clone https://github.com/moses-y/awesome-agent-skills
   cd awesome-agent-skills

Browse README.md to locate the markdown tables that categorize skills. Follow the links to the actual skill definitions (hosted in external repos or sub‑folders).
Integrate the referenced skill files into your own agent framework (e.g., copy the JSON/YAML payload into your skill registry). The repo does not provide a CLI or build step.

If you plan to contribute new skills, follow the steps in CONTRIBUTING.md and submit a pull request.

Real‑World Use

A CI pipeline that generates an internal skill catalog can script:

git clone https://github.com/moses-y/awesome-agent-skills tmp
grep -hoE 'https://.*\.json' tmp/README.md | xargs -n1 curl -O
# now the local JSON files are imported into the company’s skill registry

The script treats the README as a manifest, pulls the referenced JSON definitions, and loads them into the team’s agent runtime.

Code Health & Issues

Med – No test suite – repository contains only markdown; no automated verification of skill definitions.
Med – No CI/CD pipeline – no .github/, Makefile, or other CI config detected, so changes are not automatically linted or validated.
Low – No executable code – the repo does not ship binaries or scripts; usefulness depends on external skill files being reachable and correctly formatted.

The Bottom Line

awesome-agent-skills is a well‑organized, community‑curated catalog that can serve as a discovery layer for teams building AI agents. Its value is purely informational; there is no runnable code, test coverage, or CI automation. Teams that need a vetted skill inventory will find it useful, but they must handle validation, integration, and automation themselves.