English | 中文

Claude Code Skill Manager

A local dashboard for managing, discovering, and analyzing Claude Code skills and plugins.

Features

• Dashboard — Usage statistics, call trends, heatmap, and insights for your installed skills
• My Skills — Browse all built-in commands and your custom skills with search, filter, and detail view
• Discover — Auto-discover skills from GitHub with multi-dimensional filtering (category, action, complexity), AI-generated descriptions, and one-click install
• Settings — Configure GitHub token for higher API limits
• Bilingual — Full Chinese/English UI switching

Quick Start

Prerequisites

• Node.js v18+
• Claude Code installed

Install & Run

# Clone
git clone https://github.com/kommxp/claude-code-skill-manager.git
cd claude-code-skill-manager

# Install dependencies & start
npm install
npm start

Or on Windows, just double-click start.bat.

The browser will open automatically. If port 3200 is busy, it auto-finds the next available port.

Optional: GitHub Token (Recommended)

Without a token, GitHub API is limited to 60 requests/hour. With a token, you get 5000/hour and discover far more skills.

1. Go to GitHub Settings → Developer settings → Personal access tokens → Tokens (classic)
2. Generate new token — no permissions needed (leave all unchecked)
3. Paste it in the Settings tab of the dashboard

How It Works

Auto-Discovery Engine

The discover engine runs in the background with zero configuration:

• 8 data sources: GitHub keyword search, topic search, Awesome lists, npm packages, aggregated repos, file signatures, network expansion, fork discovery
• Smart rate limiting: Tracks API quota, skips low-priority sources when limited
• Incremental merge: New discoveries merge with cache, never loses data
• SKILL.md detection: Uses GitHub Trees API to recursively find skills at any directory depth

AI Enrichment

• Descriptions: Auto-fetches README and translates to Chinese/English via Claude CLI
• Tags: AI-powered multi-dimensional tagging (category, action, target, complexity) using Claude Haiku
• Background processing: Gradually enriches skills by score priority, 10 per cycle

Service Stability

• Auto-finds available port if 3200 is busy
• Auto-opens browser on start
• Auto-shuts down after 30 minutes idle
• Auto-restarts on crash (up to 5 times via start.bat)

Tech Stack

• Frontend: React + Vite + Tailwind CSS (dark theme)
• Backend: Express.js (single process, serves both API and static files)
• Data: File-based caching in ~/.claude/, no database needed
• AI: Claude CLI for translations and tagging (optional, works without it)

Project Structure

├── server/
│   ├── index.js              # Express entry point
│   ├── routes/               # API routes (stats, skills, discover)
│   ├── services/
│   │   ├── discover.js       # Auto-discovery engine (8 data sources)
│   │   ├── enricher.js       # Description enrichment (3-layer)
│   │   ├── tagger.js         # AI tagging engine
│   │   ├── skill-scanner.js  # Local skill scanner
│   │   ├── history-parser.js # Usage history parser
│   │   └── stats-aggregator.js
│   └── utils/
├── client/src/
│   ├── tabs/                 # Dashboard, Skills, Discover, Settings
│   ├── lib/                  # API client, i18n
│   └── App.jsx
├── dist/                     # Pre-built frontend (ready to use)
├── start.bat                 # Windows launcher (auto-restart)
├── start.sh                  # Unix launcher
└── package.json

API Endpoints

Method	Path	Description
GET	/api/stats/overview	Dashboard overview (total skills, calls, top 5, insights)
GET	/api/stats/trend?range=30d	Call trend data (7d/30d/90d)
GET	/api/skills?lang=en	List all local skills with stats
POST	/api/skills/:id/toggle	Enable/disable a custom skill
DELETE	/api/skills/:id	Delete a custom skill
GET	/api/discover?category=&sort=hot&page=1	Browse online skill catalog
GET	/api/discover/detail/:name	Get enriched skill detail (description, use cases)
GET	/api/discover/categories	List available categories with counts
POST	/api/discover/refresh	Trigger re-index of online skills
POST	/api/refresh	Refresh all local data (history + skills)
GET	/api/health	Health check
GET/POST	/api/config	Read/write settings (GitHub token)

Contributing

Contributions are welcome! Please:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/my-feature)
3. Commit your changes with clear messages
4. Push to the branch and open a Pull Request

License

MIT