Skip to content

Repository files navigation

skills.ws

Agent skills for AI coding assistants. 86 skills across 8 categories — built for OpenClaw, Claude Code, Cursor, Codex, and any agent that supports the SKILL.md format.

Website: skills.ws | npm: skills-ws | Docs: llms-full.txt


Quick Start

npx skills-ws install          # interactive picker
npx skills-ws install seo-geo  # install a specific skill
npx skills-ws install all      # install everything
npx skills-ws list             # list available skills

Skills are SKILL.md files that give AI coding assistants specialized knowledge — workflows, checklists, code patterns, and domain expertise. Install a skill and your agent gains expert-level capability in that domain.


Skills (86 across 8 categories)

Marketing (15)

SEO/GEO, content strategy, copywriting, paid ads, email sequences, PR/media, influencer marketing, brand strategy, webinars, blog engine, and more.

Dev (22)

Git workflow, CI/CD, API design, database design, testing, web performance, security hardening, prompt engineering, AI agent building, MCP client/server, Stripe billing, Telegram Mini Apps, MVP launcher, Next.js stack.

Growth (11)

Social media, community building, customer feedback, business development, cold outreach, competitor intelligence, affiliate marketing, product-led growth.

Operations (11)

EU legal compliance (GDPR, AI Act, DSA), hiring/team building, project management, CRM, accounting, revenue ops, Docker, AWS, monitoring.

Conversion (8)

Signup flows, popups, A/B testing, pricing optimization, lead scoring, page CRO, sales funnels, CRM builder.

Analytics (7)

Google Analytics, Search Console, Bing/Yandex Webmaster, data analytics, data management, retention analytics.

Web3 (8)

Solidity, smart contract auditing, DeFi integration, wallet integration, onchain analytics, Polymarket, Ophis swaps, Aleph Cloud deployment.

Design (4)

UI/UX Pro Max, design systems, landing page builder, ASCII banner.


How Skills Work

┌─────────────────────────────────────────┐
│            AI Coding Assistant           │
│  (OpenClaw, Claude Code, Cursor, Codex) │
│                                         │
│  ┌───────────────────────────────────┐  │
│  │          SKILL.md files           │  │
│  │  Loaded into agent context on     │  │
│  │  startup — gives the agent        │  │
│  │  domain expertise, workflows,     │  │
│  │  checklists, and code patterns    │  │
│  └───────────────────────────────────┘  │
└─────────────────────────────────────────┘
  1. You install a skill via npx skills-ws install <name>
  2. A SKILL.md file is placed in your agent's skills directory
  3. Your AI assistant reads it on startup and gains that expertise
  4. Ask your assistant to do anything related to that domain — it now knows the best practices, patterns, and workflows

Skills use progressive disclosure: a compact SKILL.md entrypoint routes the agent to topic-specific references/, deterministic scripts/, or reusable assets/ only when needed. Every skill also includes agents/openai.yaml interface metadata. The optional polymarket-trading scanner runs only with your approval and reads credentials from your environment or keychain, never from hardcoded values.


CLI Reference

# Interactive mode — browse and select skills
npx skills-ws install

# Install a specific skill
npx skills-ws install seo-geo

# Install multiple skills
npx skills-ws install seo-geo copywriting ab-testing

# Install all skills
npx skills-ws install all

# List available skills with categories
npx skills-ws list

# Install to a custom directory
npx skills-ws install seo-geo --dir ./my-agent/skills

The CLI auto-detects your agent by probing the skills directories each tool actually reads (.claude/skills, .cursor/skills, .agents/skills, the OpenClaw workspace skills/, and their user-level equivalents) and installs to the first one found. If none exists it defaults to ~/.agents/skills, the cross-tool path read by Cursor, Codex, Copilot, and OpenClaw.


Website Features

The skills.ws website provides:

  • Searchable skill grid — filter by category, search by name
  • Individual skill pages — full SKILL.md content rendered as Markdown
  • One-click install commands — copy npx command to clipboard
  • Live npm download counter — monthly download stats
  • Security scan status — skill files are scanned before each release

Keyboard Shortcuts

Key Action
/ Focus search
Esc Clear search

Development

git clone https://github.com/san-npm/skills-ws.git
cd skills-ws
npm install
npm run dev       # http://localhost:3000
npm run build     # static export to out/
npm run lint      # ESLint + Next.js linter

Stack

  • Next.js 16 — static export (SSG), App Router
  • React 19 + TypeScript 5
  • Tailwind CSS 3.4 — dark theme
  • Three.js — WebGL ASCII art background (homepage)
  • react-markdown + remark-gfm — SKILL.md rendering

Project Structure

skills-ws/
├── app/                    # Next.js App Router
│   ├── page.tsx            # Homepage — ASCII hero, skill grid, stats
│   ├── skills/[name]/      # Dynamic skill detail pages (SSG)
│   ├── docs/               # How agent skills work
│   ├── cli/                # CLI reference
│   ├── faq/                # FAQ with accordion
│   └── sitemap.ts          # Dynamic XML sitemap
├── components/             # React components
│   ├── AsciiBackground.tsx # Three.js WebGL → ASCII canvas
│   ├── SkillsGrid.tsx      # Searchable/filterable skill grid
│   ├── SkillContent.tsx    # Markdown renderer + safe resource links
│   ├── InstallBox.tsx      # Copy-to-clipboard install command
│   └── NpmDownloads.tsx    # Live npm download counter
├── lib/
│   └── skills.ts           # Skill data access + TypeScript interfaces
├── skills/                 # 86 complete skill bundles
├── public/
│   ├── skills.json         # Skills database (86 skills, all metadata + content)
│   ├── llms.txt            # LLM-readable skill index
│   ├── llms-full.txt       # Full content dump for LLMs
│   ├── robots.txt          # Crawl directives
│   └── install.sh          # Bash installer script
└── skills.json             # Master skills database

Build Output

Static export generates ~90 pages:

  • Homepage + docs + CLI + FAQ + 404
  • 86 individual skill detail pages
  • XML sitemap

No server needed — deploy to any static host (Vercel, Netlify, GitHub Pages, S3).


SEO & Structured Data

  • Dynamic XML sitemap with all skill pages
  • OpenGraph + Twitter Card meta tags
  • JSON-LD schemas: Organization, WebSite, SoftwareApplication, BreadcrumbList, FAQPage
  • Google Analytics 4 integration
  • llms.txt + llms-full.txt for AI crawlers
  • robots.txt with sitemap reference
  • IndexNow instant-indexing for Bing, Yandex, Seznam, Naver & Yep (see below)

IndexNow

IndexNow pushes URL changes to participating search engines immediately instead of waiting to be re-crawled.

  • Key file: public/81658695aad54ee3a7ce2951b5cb4a4e.txt, served at https://skills.ws/81658695aad54ee3a7ce2951b5cb4a4e.txt
  • Submitter: scripts/indexnow-submit.mjs builds the URL list from skills.json (kept in lockstep with app/sitemap.ts) and POSTs once to the shared api.indexnow.org endpoint, which fans out to every engine.
  • Automation: .github/workflows/indexnow.yml runs on every push to main that touches content, plus a weekly cron and manual dispatch. The script polls the live key file first, so it only fires after the matching Vercel deploy is serving.
npm run indexnow                      # submit the whole sitemap
npm run indexnow -- --dry-run         # print payload, send nothing
npm run indexnow -- /faq /skills/seo  # submit specific paths only

Security

Skills are instruction and resource bundles; the CLI only copies them and never executes skill content.

  • Zero runtime dependencies in the npm package (no supply chain risk)
  • All skills built in-house, no third-party content
  • Pre-release validation: metadata, context budgets, resource links, hidden-risk patterns, remote installer safety, and client-secret exposure
  • Environment-only credentials (nothing hardcoded)
  • One documented exception to markdown-only: polymarket-trading ships an optional scripts/scan.mjs helper, run only with your approval
  • Releases are published by the maintainer with 2FA; npm trusted publishing (OIDC) with provenance attestation is planned

See SECURITY.md for vulnerability reporting.


License

MIT — Commit Media SARL

About

Agent Skills Directory by Dr Clawdberg

Resources

Security policy

Stars

2 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages