Discover, publish, and share skills for Claude, GPT, Copilot, and other AI agents.
"The only skill ranking based on real agent usage, not vanity metrics."
| Problem | Solution |
|---|---|
| Finding quality skills is hard | Curated directory with 90+ verified skills, ranked by quality score and auto-indexed hourly |
| GitHub stars don't reflect real usage | Agent Feedback Loop — real usage data from AI agents |
| No incentive for skill authors | Points system rewards authors for every successful call |
| Skills scattered across GitHub | One-stop marketplace with search, filters, and categories |
| Skill | Stars | Description | Use Case |
|---|---|---|---|
| Browser Use | 52K | Make AI agents interact with websites using natural language | Automate form filling, web testing, data entry |
| Crawl4AI | 42K | LLM-friendly web crawler and scraper | Extract structured data from any website |
| Firecrawl | 28K | Turn websites into LLM-ready markdown | Build RAG pipelines, content indexing |
| MediaCrawler | 43K | Multi-platform social media crawler | Scrape Xiaohongshu, Douyin, Bilibili, Weibo, Zhihu |
| Skill | Stars | Description | Use Case |
|---|---|---|---|
| Composio | 15K | 250+ app connectors for AI agents | Connect agents to Slack, GitHub, Notion, etc. |
| Google Workspace CLI | 11K | CLI for Google Workspace APIs | Automate Gmail, Docs, Calendar with agents |
| Skill | Stars | Description | Use Case |
|---|---|---|---|
| Cursor Rules | 7.8K | Curated cursor rules for frameworks | Improve AI coding with framework-specific rules |
| Awesome Claude Code | 25K | Tools for Claude Code workflows | Enhance Claude coding capabilities |
| Skill | Stars | Description | Use Case |
|---|---|---|---|
| BlockBeats Skill | - | Crypto news, market data, on-chain analytics | Track ETF flows, derivatives, macro trends via 1,500+ sources |
| Finance Skills | 94 | Options, stocks, sentiment analysis | Build trading bots, financial research agents |
| Claude Scientific Skills | 10K | Research and scientific skills | Academic research, data analysis |
专为中文 AI 社区构建的 skills。在网站上点击「中文 Chinese」分类筛选查看。
| Skill | Stars | Description | Use Case |
|---|---|---|---|
| 同事.Skill | 6.8K | 把同事的工作经验和性格永久保存为 AI Skill | 用同事的语气写代码、回答问题,支持飞书/钉钉/微信数据源 |
| WeChat Article Exporter | 7.9K | 微信公众号文章批量下载工具 | 导出文章含阅读量、评论,支持 Docker/Cloudflare 部署 |
| DBSkill | - | 从 12,307 条推文提炼的商业诊断工具箱 | 商业模式诊断、对标分析、内容创作、执行力诊断 |
| Skill | Stars | Description | Use Case |
|---|---|---|---|
| ClawFeed | 1.6K | AI-powered news digest from multiple sources | Curate Twitter, RSS, HackerNews, Reddit into summaries |
| Planning With Files | 14K | Persistent markdown planning workflows | Project management with AI agents |
| Eigent | 12K | Build custom AI workforce | Multi-agent team collaboration |
The unique feature that sets us apart. Track real agent usage, not just GitHub stars.
curl -X POST https://www.openagentskill.com/api/agent/feedback \
-H "Content-Type: application/json" \
-d '{
"skill_slug": "browser-use",
"agent_id": "claude-3.5",
"success": true,
"latency_ms": 1200
}'# Get stats for a specific skill
curl "https://www.openagentskill.com/api/agent/feedback?skill_slug=browser-use"
# Get leaderboard (sorted by agent calls)
curl "https://www.openagentskill.com/api/agent/feedback"{
"skill_slug": "browser-use",
"total_calls": 12847,
"success_rate": 94.2,
"avg_latency_ms": 850,
"unique_agents": 156
}Visit openagentskill.com/skills to explore all skills with filters by category, popularity, and trending.
- Go to openagentskill.com/submit
- Enter your GitHub repository URL
- AI reviews your skill automatically
- Approved skills appear in the directory within minutes
We automatically discover new skills from GitHub hourly using a skill-only search matrix. MCP servers and Model Context Protocol integrations are intentionally excluded from automated imports.
topic:agent-skills "agent skill" topic:ai-agent
topic:claude-tool topic:openai-plugin topic:langchain-tool
topic:browser-use topic:rag topic:workflow-automation
Skill authors earn points for contributions:
| Event | Points |
|---|---|
| Skill called successfully by an agent | +1 |
| Skill submitted and approved | +50 |
| Invite a new user | +100 |
Points unlock badges and future rewards.
| Layer | Technology |
|---|---|
| Framework | Next.js 16 (App Router) |
| Database | Supabase (PostgreSQL + Auth) |
| Styling | Tailwind CSS v4 |
| Components | shadcn/ui |
| Deployment | Vercel |
| AI Review | Vercel AI Gateway |
# Clone
git clone https://github.com/Leon-Drq/openagentskill.git
cd openagentskill
# Install
pnpm install
# Setup environment
cp .env.example .env.local
# Fill in your Supabase credentials
# Run
pnpm dev| Variable | Description |
|---|---|
NEXT_PUBLIC_SUPABASE_URL |
Supabase project URL |
NEXT_PUBLIC_SUPABASE_ANON_KEY |
Supabase anonymous key |
SUPABASE_SECRET_KEY / SUPABASE_SERVICE_ROLE_KEY |
Server-only Supabase key for privileged automation/admin routes |
GITHUB_TOKEN |
GitHub token for API access |
INDEXER_SECRET |
Bearer secret required for indexer/blog routes and reviewed skill-submission RPC writes in production |
INDEXER_RUN_TARGET |
Optional. Hourly bulk-import target per run, defaults to 10, max 500 |
INDEXER_MIN_STARS |
Optional. Minimum GitHub stars for bulk import, defaults to 500 |
INDEXER_MAX_SEARCH_REQUESTS |
Optional. GitHub search requests per bulk run, defaults to 10 without GITHUB_TOKEN |
CRON_SECRET |
Bearer secret required for cron-triggered maintenance routes in production |
X_CLIENT_ID |
X OAuth 2.0 client ID for authorizing the posting account |
X_CLIENT_SECRET |
X OAuth 2.0 client secret for token exchange and refresh |
X_ALLOWED_USERNAME |
Optional. X username allowed for OAuth storage, defaults to openagentskill |
Apply SQL files in scripts/ in order. They create the skills catalog, profiles and points ledger, activity feed, agent feedback loop, aggregate stats view, and RLS policies.
Production writes should go through the Next.js API routes. Public feedback, reviewed skill submissions, and indexer writes use narrow Supabase RPCs guarded by server secrets, while public clients can only read approved skills and aggregate stats directly.
The hourly indexer defaults to high-star, skill-only bulk discovery. It scans GitHub repositories matching agent skill, browser automation, RAG, workflow, and developer-tool signals, excludes MCP repositories, then imports up to INDEXER_RUN_TARGET new approved skills per run. Use GET /api/indexer/logs with automation auth to inspect recent run summaries. Use POST /api/indexer/run with mode: "reviewed" for the slower README/AI-review path.
X posting supports two paths. The free, manual path is /api/x/intent, which generates a scenario-led OpenAgentSkill Update draft and redirects to the official X Web Intent composer; the user must click Post, and no X API credits are used. The main draft includes the skill URL so it can be moved into a reply before posting; /api/x/intent/reply?tweet_id=<posted-tweet-id> can also prefill a manual reply with the same skill URL. The paid API path uses OAuth 2.0 with PKCE: visit /api/x/auth while signed in as the allowed X account to store an encrypted refresh token, then /api/x/post-daily can publish one high-quality unposted skill when called with automation auth. It is intentionally not scheduled in vercel.json.
app/
├── api/
│ ├── agent/feedback/ # Agent Feedback API
│ ├── indexer/ # Auto-indexer endpoints
│ └── skills/ # Skills CRUD
├── skills/ # Skills listing and detail pages
├── submit/ # Skill submission flow
└── profile/ # User profile and points
lib/
├── indexer/ # GitHub auto-discovery
├── ai-review/ # AI-powered skill review
└── db/ # Database queries
- Skill marketplace with search and filters
- Auto-indexer for GitHub skill discovery
- User authentication and profiles
- Points and rewards system
- Agent Feedback Loop API
- Skill composition engine
- Bounty system for skill requests
- Revenue sharing for popular skills
Contributions are welcome! See CONTRIBUTING.md for guidelines.
Ways to contribute:
- Submit your skills to the directory
- Report bugs or suggest features via Issues
- Improve documentation
- Add new integrations
MIT License — see LICENSE