A GitHub-ready list of notable Claude / OpenClaw / Agent Skills packs, including author/source info, what each is useful for, compatibility notes, install hints, and live GitHub star badges.
Last reviewed: 2026-04-26
Tip: The star badges below use Shields.io, so if this file is published to GitHub they will auto-refresh periodically instead of becoming stale.
- Live stars are dynamic badges, not hardcoded counts.
- Official means published by the company/project named in the row.
- Community means a third-party repo, even if it is inspired by a famous person.
- Audit first means inspect
SKILL.md, scripts, install commands, and required permissions before installing.
[](https://github.com/OWNER/REPO)| Priority | Skill / pack | Why install first | Repo / source | Live stars |
|---|---|---|---|---|
| 1 | Skill vetter / scanner | Check untrusted skills before using them | Feahter/openclaw-tools |
 |
| 2 | Agent Browser | Browser automation, screenshots, extraction, form-filling | vercel-labs/agent-browser / openclaw/skills |
 |
| 3 | Tavily Search | Live web search and research for agents | tavily-ai/skills / Tavily docs |
 |
| 4 | Anthropic webapp-testing | Playwright-style local app testing, screenshots, browser logs | anthropics/skills |
 |
| 5 | Addy Osmani agent-skills | Full coding lifecycle: spec, plan, build, test, review, ship | addyosmani/agent-skills |
 |
| 6 | Garry Tan gstack | Opinionated startup/product-engineering workflow pack | garrytan/gstack |
 |
| 7 | Karpathy read-arxiv-paper | Reads arXiv TeX source and writes technical paper summaries | karpathy/nanochat |
 |
| 8 | Product Manager Skills | PRDs, roadmaps, discovery, story mapping, prioritization | deanpeters/Product-Manager-Skills |
 |
| 9 | Marketing Skills | Copywriting, SEO, CRO, analytics, growth workflows | coreyhaines31/marketingskills |
 |
| 10 | Hugging Face Skills | Dataset creation, model training, evaluation, HF Hub workflows | huggingface/skills |
 |
| # | Skill / pack | Author / source | Repo / source | Live stars | Useful for | Notable skills / components | Works with / likely portable to OpenClaw? | Install hint | Notes |
|---|---|---|---|---|---|---|---|---|---|
| 1 | gstack |
Garry Tan | garrytan/gstack |
|
Startup/product engineering, review loops, deployment discipline | CEO review, engineering review, design review, QA, security, deploy, browse, retro, guard/freeze safety tools | High; skill folders can be inspected and adapted | npx skills add garrytan/gstack |
Very broad pack; audit commands before using on production repos. |
| 2 | agent-skills |
Addy Osmani | addyosmani/agent-skills |
|
Full software-development lifecycle | Spec, plan, build, test, review, ship, TDD, code review, browser testing, simplification | Very high; designed around Agent Skills conventions | npx skills add addyosmani/agent-skills |
Strong first install for coding agents. |
| 3 | web-quality-skills |
Addy Osmani | addyosmani/web-quality-skills |
 |
Frontend quality, performance, SEO, accessibility | Lighthouse, Core Web Vitals, framework-agnostic audits | Very high for web projects | npx skills add addyosmani/web-quality-skills |
Best for frontend audit/fix workflows. |
| 4 | impeccable |
Paul Bakaus | pbakaus/impeccable |
 |
UI taste, frontend polish, avoiding generic AI-generated UI | Typography, spacing, color, motion, UX writing, audits, anti-patterns | High; supports multiple coding agents | npx impeccable or follow repo instructions |
Good companion to frontend-design/web-quality skills. |
| 5 | read-arxiv-paper |
Andrej Karpathy, inside nanochat |
karpathy/nanochat/.claude/skills/read-arxiv-paper |
|
Reading ML/AI research papers from source TeX | Normalize arXiv URL, download source, inspect LaTeX, write project-focused summary | High; copy the skill folder if your agent supports SKILL.md |
Manual copy from .claude/skills/read-arxiv-paper |
This is the most directly Karpathy-owned skill found. |
| 6 | karpathy-guidelines |
Forrest Chang; Karpathy-inspired, not authored by Karpathy | forrestchang/andrej-karpathy-skills |
 |
Better coding behavior and safer changes | Think before coding, simplicity first, surgical changes, goal-driven verification | Medium-high; more of a behavior/rules skill | Follow repo plugin instructions or copy skills/karpathy-guidelines |
Useful guardrail; clearly label as “inspired by,” not official Karpathy. |
| 7 | Datasette skill | Simon Willison / Datasette project | datasette/skill |
 |
Datasette plugin/project work | Project-specific guidance for Datasette | Medium; niche but useful if you use Datasette | Manual install/copy | Best for Datasette-specific development. |
| 8 | marketingskills |
Corey Haines | coreyhaines31/marketingskills |
|
Marketing, growth, product marketing, CRO | Copywriting, content strategy, social content, SEO, analytics, growth engineering | High; follows Agent Skills spec | npx skills add coreyhaines31/marketingskills |
Very useful for technical founders and marketers. |
| 9 | Product-Manager-Skills |
Dean Peters | deanpeters/Product-Manager-Skills |
|
Product management and strategy | PRD development, roadmap, discovery, story mapping, pricing, workshops | High; docs mention Claude, Cowork, Codex, AI agents | npx skills add deanpeters/Product-Manager-Skills --list |
Good for product specs and planning. |
| 10 | advertising-skills |
Kim Barrett | realkimbarrett/advertising-skills |
 |
Direct-response advertising and paid acquisition | Ads, funnels, positioning, copy systems, campaign workflows | High; repo describes Open Claw and Claude Code support | npx skills add realkimbarrett/advertising-skills |
Good for ads/funnel work; validate claims before running campaigns. |
| 11 | BMAD Method skills | BMAD Method / bmad-code-org | bmad-code-org/BMAD-METHOD |
 |
Agentic agile/software workflows | PR review, file-reference audit, workflow agents, project methods | Medium-high; copy/adapt individual skill folders | Manual copy of relevant .claude/skills/* |
Large framework; avoid installing everything blindly. |
| # | Skill / pack | Author / source | Repo / source | Live stars | Useful for | Notable skills / components | Works with / likely portable to OpenClaw? | Install hint | Notes |
|---|---|---|---|---|---|---|---|---|---|
| 1 | Official Claude / Agent Skills | Anthropic | anthropics/skills |
|
General Claude/agent productivity | docx, pdf, pptx, xlsx, frontend-design, webapp-testing, skill-creator, mcp-builder, internal-comms |
High for Agent Skills-compatible tools | Follow Anthropic repo/plugin instructions | Best reference implementation for skill structure. |
| 2 | webapp-testing |
Anthropic | anthropics/skills |
|
Local web app testing | Playwright-style tests, browser logs, screenshots, frontend behavior checks | High; useful in OpenClaw if path/tooling is adapted | Copy webapp-testing folder to your skills directory |
Good companion to Agent Browser. |
| 3 | Vercel Agent Skills | Vercel Labs | vercel-labs/agent-skills |
 |
Vercel deployments and web app workflows | Deployment, project setup, web design guidelines, platform guidance | High if you use Vercel | npx skills add vercel-labs/agent-skills |
Official Vercel collection. |
| 4 | Vercel skills CLI/tool |
Vercel Labs | vercel-labs/skills |
 |
Discovering/installing skills | npx skills add, find-skills, marketplace-style discovery |
High; useful as installer/discovery utility | npx skills add OWNER/REPO |
Useful meta-tool, not just a skill pack. |
| 5 | Agent Browser | Vercel Labs / OpenClaw archive | vercel-labs/agent-browser, openclaw/skills |
|
Browser automation | Open pages, click, fill, scrape, screenshot, PDFs, network, tabs, state | Very high for OpenClaw/browser work | clawhub install thesethrose/agent-browser or npx skills add vercel-labs/agent-browser |
Prefer official/upstream repo where possible. |
| 6 | Tavily Agent Skills | Tavily | tavily-ai/skills / docs |
|
Live web search, extraction, crawling, research | Search API, extract, crawl, research workflows | High, but needs Tavily API key | Follow Tavily docs or CLI instructions | Free tier may have limits; good research skill. |
| 7 | Hugging Face Skills | Hugging Face | huggingface/skills |
|
ML/AI development | Dataset creation, model training, evaluation, experiment workflows, Hub tasks | High for ML projects | npx add-skill huggingface/skills or repo instructions |
Strong for AI builders. |
| 8 | Hugging Face kernel skills | Hugging Face Kernels | huggingface/kernels |
 |
GPU kernel optimization | CUDA kernels, ROCm kernels, benchmarking | Medium-high for ML infra | kernel-builder skills add |
Specialized but valuable for performance work. |
| 9 | MongoDB Agent Skills | MongoDB | mongodb/agent-skills |
 |
MongoDB / Atlas workflows | MCP setup, schema/query guidance, connection workflows, Atlas usage | High for MongoDB users | /plugin install mongodb or repo docs |
Official MongoDB collection. |
| 10 | Testcontainers Claude Skills | Testcontainers | testcontainers/claude-skills |
 |
Container-based integration testing | Docker/Testcontainers testing patterns | Medium-high for backend projects | Follow repo instructions | Good for integration test generation/debugging. |
| 11 | Figma / design-to-code skills | Figma ecosystem / community | Figma MCP skills docs / ihlamury/design-skills |
 |
Design systems, Figma-to-code, UI consistency | Design-system rules, Figma MCP workflows, design constraints | Medium-high; depends on MCP setup | npx skills add ihlamury/design-skills |
Use with Figma MCP for best results. |
| # | Skill / pack | Author / source | Repo / source | Live stars | Useful for | Notable skills / components | Works with / likely portable to Claude? | Install hint | Notes |
|---|---|---|---|---|---|---|---|---|---|
| 1 | OpenClaw skills archive | OpenClaw | openclaw/skills |
 |
Finding ClawHub-hosted skills | Archived skill versions, SKILL.md files |
Medium-high; inspect individual folders | Browse and copy only trusted skills | Use as source-of-truth archive, not automatic install list. |
| 2 | Awesome OpenClaw Skills | VoltAgent | VoltAgent/awesome-openclaw-skills |
 |
Discovery | Categorized OpenClaw skills registry, safety warnings | N/A; directory/list | Browse categories | Good for discovery; still audit each skill. |
| 3 | Capability Evolver | EvoMap / OpenClaw ecosystem | EvoMap/evolver |
 |
Agent self-evolution / improvement | Runtime-history analysis, protocol-constrained evolution | Medium; high-risk category | Manual review first | Audit carefully: self-modifying/network/shell skills are sensitive. |
| 4 | Self-improving agent | Peterskoett / OpenClaw community | peterskoett/self-improving-agent |
 |
Capturing learnings/errors/preferences | Persistent learning notes, corrections, feature-request memory | Medium-high, privacy-sensitive | Manual install from repo | Check where it stores memory and what it reads. |
| 5 | OpenClaw master skills | LeoYeAI | LeoYeAI/openclaw-master-skills |
 |
Big OpenClaw skills bundle | Tavily search, multi-search, local search, scraping, research tools | Medium; pick individual skills | Manual pick/copy | Avoid bulk-installing all skills. |
| 6 | OpenClaw tools | Feahter | Feahter/openclaw-tools |
|
OpenClaw utilities and safety | skill-vetter, skill-web-search, auto-updater, retrieval templates |
Medium-high | Manual pick/copy | Good source for vetting and utility patterns. |
| 7 | Web Search Pro | Zjianru | Zjianru/web-search-pro |
 |
Multi-engine web search | Tavily, Exa, Serper, SerpAPI, filtering, date ranges, news mode | Medium-high; requires API keys | Follow repo instructions | Use when one search provider is not enough. |
| 8 | OpenClaw skill-agent-browser wrapper | clawdbrunner | clawdbrunner/skill-agent-browser |
 |
OpenClaw wrapper for Agent Browser | Browser automation CLI instructions for OpenClaw | High for OpenClaw | Follow repo instructions | Wrapper points to vercel-labs/agent-browser; prefer upstream for code review. |
| 9 | Skill of Skills | the911fund | the911fund/skill-of-skills |
 |
Discovery and meta-skill workflows | Skill discovery, evaluation, registry-like metadata | Medium | Follow repo instructions | Useful as a catalog, but verify downstream sources. |
| 10 | Acontext | memodb-io | memodb-io/Acontext |
 |
Agent memory layer | Captures learnings from agent runs and stores them as skill files | Medium; privacy-sensitive | Follow repo instructions | Inspect storage location and data captured. |
| Skill | Source repo | Live stars | Useful for | Why it matters | Portability note |
|---|---|---|---|---|---|
docx |
anthropics/skills |
|
Word document creation/editing | One of the official document skills | Usually portable if runtime tools exist |
pdf |
anthropics/skills |
|
PDF generation/analysis | Official document skill | Check dependencies and sandbox rules |
pptx |
anthropics/skills |
|
Slide generation/editing | Official presentation skill | Portable if slide tooling exists |
xlsx |
anthropics/skills |
|
Spreadsheet work | Official spreadsheet skill | Portable if spreadsheet runtime exists |
skill-creator |
anthropics/skills |
|
Creating new skills | Helps write well-structured skills | Useful in any SKILL.md ecosystem |
mcp-builder |
anthropics/skills |
|
MCP server creation | Helps agent connect to tools | Useful if your workflow uses MCP |
frontend-design |
anthropics/skills |
|
Better frontend UI | Base design skill; complements impeccable |
Portable as guidance |
webapp-testing |
anthropics/skills |
|
Browser-based app tests | Great for local UI testing | Needs Playwright/browser tools |
find-skills |
vercel-labs/skills |
|
Skill discovery | Helps identify/install relevant skills | Good meta-skill |
core-web-vitals |
addyosmani/web-quality-skills |
|
Web performance | Focused on LCP, INP, CLS style fixes | Needs project/build access |
copywriting |
coreyhaines31/marketingskills |
|
Marketing copy | Practical for landing pages and campaigns | Guidance skill, low tooling risk |
prd-development |
deanpeters/Product-Manager-Skills |
|
PRDs and product specs | Helps structure requirements | Guidance skill, low tooling risk |
# List available skills in a repo, if supported
npx skills add OWNER/REPO --list
# Install a specific skill globally for Claude/Codex-style agents, if supported
npx skills add OWNER/REPO --skill SKILL_NAME -g# Example: install OpenClaw Agent Browser from ClawHub
clawhub install thesethrose/agent-browser
# Or with npx
npx clawhub@latest install thesethrose/agent-browser# Claude-style global skills
mkdir -p ~/.agents/skills
cp -R ./some-skill-folder ~/.agents/skills/some-skill
# OpenClaw-style skill folder; adjust path to your OpenClaw setup
mkdir -p ~/.openclaw/skills
cp -R ./some-skill-folder ~/.openclaw/skills/some-skill- Read the full
SKILL.md. - Search for hidden shell commands, curl/wget pipes, obfuscated code, and external network calls.
- Check scripts under
scripts/,bin/,hooks/,.claude/,.agents/, and install files. - Avoid skills that ask for broad access to
.env, SSH keys, browser profiles, crypto wallets, or session logs. - Run high-risk skills in a sandbox, container, or throwaway VM.
- Prefer official repos or well-known maintainers.
- Pin a commit SHA for production use instead of installing from a moving
mainbranch. - Do not bulk-install a large skill pack into a sensitive machine.
- For OpenClaw, be extra cautious because skills may guide an agent with local file, shell, network, browser, and account access.
- Star badges auto-update via Shields.io, but GitHub and Shields may cache them.
- If a repo moves or is renamed, update both the link and badge URL.
- For source credibility, prefer rows with a direct GitHub repo containing
SKILL.md. - For non-GitHub or marketplace-only skills, use a plain link and mark live stars as
N/A. - Re-check security issues before recommending any OpenClaw skill pack publicly.