UseSkill — AI Agent Skill Meta-Registry

Global leading AI Agent skill aggregation protocol. Break skill silos and achieve discover & integrate for Agent capabilities.

English | 中文

---

Overview

UseSkill is a locally-deployable AI Agent skill meta-registry that:

• 🔍 Aggregates multi-source data: Automatically crawls and indexes 100,000+ Agent skills from ClawHub, Skills.sh, Smithery, MCP Market, GitHub, and more.
• ⚡ Full-text search: Millisecond-level full-text search via MySQL FULLTEXT, with fuzzy fallback.
• 📊 Popularity ranking: Real-time Top 20 most popular skills by downloads.
• 🌐 i18n: Web UI supports Chinese / English switching (with language preference persistence).
• 🔄 Auto sync: Background scheduled tasks (hourly) incrementally sync latest skill data from upstream sources.

---

Screenshots

---

Architecture

UseSkill
├── run.py                   # Unified entry (API + scheduled sync)
├── skill-cli.py             # CLI tool
├── static/index.html       # Web frontend (single file, i18n)
├── config.yaml             # Config (DB DSN, etc.)
├── src/
│   ├── api/main.py         # FastAPI routes
│   ├── core/schema.py      # Core models (SkillMetadata, etc.)
│   ├── registry/
│   │   ├── db.py           # DB ops (search / ranking / categories)
│   │   ├── engine.py       # Retrieval engine
│   │   ├── sync_manager.py # Sync logic (incremental/full)
│   │   └── sync_service.py # Scheduler
│   └── adapters/           # Platform adapters
│       ├── clawhub_adapter.py
│       ├── skills_sh_adapter.py
│       ├── smithery_adapter.py
│       ├── mcp_market_adapter.py
│       ├── github_adapter.py
│       └── awesome_claude_adapter.py
└── requirements.txt

---

Quick Start

1. Install dependencies

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

2. Configure database

Copy config.example.yaml to config.yaml and set database.mysql_dsn, or set env var MYSQL_DSN.

3. Start service

python3 run.py

Then:

• 🌐 Web UI: http://localhost:8000
• ⚙️ API docs: http://localhost:8000/docs
• 🔄 Background sync: Hourly incremental sync

---

API Reference

Method	Endpoint	Description
GET	/api/v1/skills/search	Search skills (pagination, category filter)
GET	/api/v1/skills/top100	Top 100 by popularity
GET	/api/v1/skills/categories	Categories and skill counts
POST	/api/v1/skills/mount	Mount/install skill to Agent context

Example:

curl "http://localhost:8000/api/v1/skills/search?query=finance&limit=10&category=clawhub"

---

CLI Usage

# Search skills
python3 skill-cli.py search "data"

# Search by source
python3 skill-cli.py search "finance" --limit 3

# Install skill
python3 skill-cli.py install "skill://github/langchain-ai/finance-tools@main"

---

Data Sources

Source	Description	Skills
🦞 ClawHub	Claude community curated skills	~24,000+
🚀 Skills.sh	Community automation skill aggregation	~65,000+
🛠️ Smithery	MCP Server ecosystem	~14,000+
🛒 MCP Market	Multi-source MCP tools	~200+
GitHub	Awesome Agent skill repos	~2,700+

---

Tech Stack

• Backend: Python 3.12 + FastAPI + Uvicorn
• Database: MySQL (FULLTEXT)
• Frontend: Vanilla HTML / CSS / JavaScript
• Scheduler: APScheduler
• Data Models: Pydantic v2

---

Deployment

See DEPLOY.md for Railway deployment guide.

---

Development

Mark src as Sources Root in your IDE, or:

export PYTHONPATH=$PYTHONPATH:$(pwd)/src

Manual full sync:

from src.registry.sync_manager import SyncManager
import asyncio
asyncio.run(SyncManager().full_sync())

---

License

MIT © 2026 UseSkill Community