Curated agent skills directory — 124 skills indexed

Top Agent Skills for Claude, Cursor & Codex

The curated SKILL.md directory. The most useful Skills for Claude, Cursor, Codex, Gemini & Antigravity — ranked by real utility, not install count.

124 skills indexed21 categoriesAgent-readable structureUpdated weekly

Browse by use case

Top Skills by category

Function-first taxonomy. 16 categories covering every common builder workflow.

5 skills

Documents

Create, parse, and edit DOCX, PDF, XLSX and PPTX files. OCR, format conversion, deterministic scripts.

19 skills

Frontend & UI

Distinctive UI generation, design systems, brand guidelines, and theming for production-grade frontends.

18 skills

Code Review

Structured reviews, refactoring, simplification passes, and TDD-driven workflows.

12 skills

Testing

Webapp testing, Playwright flows, browser test runners, and automated QA passes.

7 skills

Security

Static analysis, vulnerability scanning, k8s hardening, audit workflows, pentest tooling.

15 skills

DevOps

Terraform, AWS/GCP/Azure, CI/CD pipelines, container management, IaC patterns.

8 skills

Browser Automation

Headless browser control, scraping, navigation, and structured extraction at scale.

10 skills

Data & Analytics

Data sources, scientific datasets, BI dashboards, D3 viz patterns, structured extraction.

14 skills

Marketing

CRO, copywriting, growth engineering, campaign planning, analytics tracking.

5 skills

SEO & Content

Keyword research, technical SEO audits, GEO/AEO for AI search, content ops, schema.

6 skills

Video & Media

Programmatic video (Remotion), canvas design, algorithmic art, animated GIFs.

5 skills

Diagrams

Excalidraw, mermaid, D3 — generate architectural and conceptual diagrams from prose.

18 skills

AI Agent Tooling

Build skills, MCP servers, and agent infra. Claude API SDK, skill-creator, mcp-builder.

5 skills

Databases

PlanetScale workflows, schema design conventions, query patterns, migrations.

24 skills

Comms & Docs

Internal comms, doc co-authoring, technical writing, status reports, FAQs.

19 skills

Workflows

Brainstorm → spec → plan → build → review → merge. Multi-step multi-agent orchestration.

Domains & DNS

Domain registration, TLD selection, WHOIS lookups, DNS record management, nameserver and zone-file workflows.

2 skills

Hosting & Deploys

Ship to Vercel, Netlify, Cloudflare Pages, Railway, Fly.io. Edge config, SSL/TLS, environment management, rollback playbooks.

Payments

Stripe checkout, subscription billing, webhook handling, invoicing, refunds, tax and dunning workflows.

1 skill

Email & Messaging

Transactional email (Resend, Postmark, SendGrid), SMS, push notifications, Slack and Discord bot workflows.

3 skills

APIs & Integrations

REST and GraphQL clients, OAuth flows, webhook receivers, third-party SaaS connectors, rate-limit and retry patterns.

Browse by role

Skills by role and domain

Contextual picks for engineering, marketing, design, security and more.

Browse by agent

Skills for your agent of choice

Skills work cross-platform — but install commands and quirks differ. Pick your agent and see what's verified.

Why it matters

Why builders use Skills

Specialize the agent

Tailor Claude, Cursor or Codex for domain-specific work — branded docs, your DB schema, your team’s workflows.

Reduce repetition

Create once, used automatically when relevant. No more copy-pasting the same prompt across conversations.

Compose capabilities

Stack multiple Skills together. Claude orchestrates which one to load when — only what is needed enters context.

Cross-platform standard

SKILL.md is open and portable. Write once, run on Claude, Codex, Cursor, Gemini, Antigravity.

Quick comparison

Marquee Skills compared

A high-level look at the most widely used Skills across categories.

#SkillAgents
1frontend-design

Escape generic AI-generated UIs. Forces a bold, distinctive design direction before any code is written.

2pdf

Extract, fill, merge. PDF manipulation with bundled scripts so Claude does not waste tokens generating them.

3composio

1,000+ SaaS integrations as one skill. OAuth lifecycle, scoped credentials, structured actions.

4superpowers

Structured multi-step development. Brainstorm → spec → plan → build → review → merge — TDD baked in.

5trail-of-bits

Security audits run by a real security firm. CodeQL + Semgrep + audit workflows.

6remotion-best-practices

Programmatic video done right. Idiomatic Remotion patterns auto-activate when detected.

For detailed comparison within a category, see the category pages.

FAQ

Frequently asked questions about Skills

Direct answers — also indexed for agent recommendation queries.

What is an agent Skill?

A Skill is a folder containing a SKILL.md file (YAML frontmatter + markdown instructions, optional scripts and resources). It teaches an agent how to perform a specific task. The format is an open standard adopted by Anthropic, OpenAI Codex, Cursor, Gemini CLI, and Antigravity — write a Skill once, run it across all of them.

What's the difference between Skills, MCP servers, and plugins?

Skills change how an agent thinks (instructions, methodology, workflows). MCP servers change what an agent can do (connect to tools, APIs, databases). Plugins bundle Skills, MCP servers, and commands together. Most builders use Skills for methodology and MCP for connectivity.

Which Skills should I install first?

Start with: frontend-design (better UIs), pdf and docx (document I/O), simplify (code review), and superpowers (multi-step development). All are one-command installs. For domain-specific work, see the category pages.

How do I install a Skill?

For Claude Code: npx skills add <repo> --skill <name>. Cursor adds --target cursor. Codex CLI uses codex skills add. Each skill's detail page has the exact command for every supported agent.

Are Skills useful for autonomous agents?

Yes — Skills are how you give an autonomous agent reliable, repeatable expertise. Bundle deterministic scripts, document workflows, set best practices. Loaded on demand, not on startup, so they do not bloat the context window.

How are Skills ranked here?

A composite 0–100 score from 9 verifiable inputs: install count, provenance (Anthropic / verified org / community), GitHub stars, last-update recency, cross-agent compatibility, documentation depth, install ergonomics, license, and our verification freshness. The full rubric is published.

Ready to teach your agent something new?

Browse all 21 categories and 124 hand-curated Skills. Every page is structured for fast evaluation — by humans and agents.