A self-governed Python project that automatically searches GitHub for Claude/Anthropic agent skills, analyzes them using AI, categorizes them, and keeps the SkillFlow repository updated with the latest skills.

X-Skills is a npm-style plugin system that allows you to install and manage AI sub-agents as executable commands. Skills are packaged as plugins that can be searched, installed, and run directly from your CLI.

One-Line Install (Recommended):

curl -fsSL https://raw.githubusercontent.com/tools-only/X-Skills/main/xsk-install.sh | bash

Via Claude Code CLI:

# Use the xskills command directly
xskills search "research"
xskills install research-agent

Full Installation:

git clone https://github.com/tools-only/SkillFlow.git ~/.skillflow
cd ~/.skillflow
bash scripts/setup_xskills.sh
echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.bashrc
source ~/.bashrc

# Search plugins
xskills search "research"              # Search by keyword
xskills search "代码审查" --tier expert  # Filter by quality tier

# Install plugins
xskills install research-agent         # Single plugin
xskills install @academic_research     # Scenario bundle (multiple plugins)

# Manage plugins
xskills list-plugins                   # List installed
xskills capabilities research-agent    # View plugin capabilities
xskills validate research-agent        # Validate plugin
xskills info                           # System info

# Run plugins
xskills run research-agent "搜索最新的AI论文"
research-agent "搜索最新的AI论文"      # Direct command after install

# Update/Uninstall
xskills update research-agent
xskills uninstall research-agent

# Scenario bundles
xskills scenarios                      # List all 22 scenarios
xskills scenarios --category education # Filter by category

Plugins are rated across four dimensions (0-25 each):

• Depth: Technical depth and domain expertise
• Practicality: Real-world applicability
• Reliability: Consistency and robustness
• Collaboration: Integration and compatibility

Tiers:

• Expert (★): Top 5% - Highest quality skills
• Good: Reliable, production-ready
• Basic: Functional skills with room for improvement

• GitHub Search Integration: Automatically finds repositories containing skill files
• AI-Powered Analysis: Uses Claude API to analyze and categorize skills
• Smart Organization: Organizes skills into nested category/subcategory folders
• Duplicate Detection: Tracks processed skills to avoid duplicates
• Automated Updates: Cron job support for periodic updates
• Git Integration: Automatic commit and push of new skills

SkillFlow/
├── skills/                    # Downloaded skill files (git tracked)
│   └── {category}/           # Nested categories (e.g., daily-assistant/writing/)
│       └── {subcategory}/    # Sub-categories for organization
│           └── {skill-name}.md
├── src/
│   ├── __init__.py
│   ├── main.py              # Entry point for cron jobs
│   ├── config.py            # Configuration loader
│   ├── github_searcher.py   # GitHub Search API integration
│   ├── skill_analyzer.py    # AI-based skill analysis
│   ├── skill_fetcher.py     # Clone and extract skill files
│   ├── organizer.py         # Category-based file organization
│   ├── tracker.py           # Track processed skills (SQLite + JSON)
│   └── updater.py           # Git operations for repo updates
├── config/
│   ├── config.yaml          # Main configuration file
│   └── search_terms.yaml    # Custom search terms configuration
├── data/
│   └── skills_tracker.db    # SQLite database of processed skills
├── logs/
│   └── skillflow.log        # Application logs
├── requirements.txt
├── .env.example             # Environment variables template
├── .gitignore
├── README.md
└── scripts/
    └── setup_cron.sh        # Cron job setup script

• Python 3.11+
• GitHub Token (optional, for higher API rate limits)
• Anthropic API Key (required)

1. Clone the repository

   git clone <your-repo-url>
   cd SkillFlow

2. Create a virtual environment

   python3 -m venv .venv
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate

3. Install dependencies

   pip install -r requirements.txt

4. Configure environment variables

   cp .env.example .env
   # Edit .env and add your API keys

   Required variables:

   • ANTHROPIC_API_KEY: Your Anthropic API key (get from https://console.anthropic.com/)
   • GITHUB_TOKEN: Your GitHub token (optional, get from https://github.com/settings/tokens)

# Run a single update cycle
python -m src.main

# Show statistics only
python -m src.main --stats

# Dry run (no git commits)
python -m src.main --dry-run

Run the setup script:

bash scripts/setup_cron.sh

This will:

• Create a virtual environment
• Install dependencies
• Verify configuration
• Set up a cron job (default: every hour)
• Configure log rotation

To customize the schedule, edit the crontab:

crontab -e

Default schedule (every hour):

0 * * * * cd /path/to/SkillFlow && .venv/bin/python -m src.main >> logs/skillflow.log 2>&1

Main configuration file with GitHub token, API keys, and paths.

github:
  token: "${GITHUB_TOKEN}"
  max_results: 20
  min_stars: 5

anthropic:
  api_key: "${ANTHROPIC_API_KEY}"
  model: "claude-3-5-sonnet-20241022"
  max_tokens: 2000

paths:
  skills_dir: "skills"
  data_dir: "data"
  log_dir: "logs"

search:
  languages: ["python", "javascript", "typescript"]
  sort_by: "updated"
  order: "desc"

Custom search terms for finding skill repositories.

terms:
  - "claude skill"
  - "anthropic agent"
  - "claude function"
  - "ai assistant skill"

excluded_repos:
  - "tools-only/SkillFlow"

required_file_patterns:
  - "**/*.md"

Skills are automatically categorized into:

• daily-assistant: Personal organization, scheduling, reminders
• commercial: E-commerce, business tools, customer service
• investment: Financial analysis, trading, portfolio management
• development: Coding assistance, DevOps, software tools
• research: Academic research, data gathering
• content-creation: Writing, media creation, editing
• data-analysis: Statistics, visualization
• automation: Workflow automation, scripting
• communication: Email, messaging, collaboration
• productivity: Efficiency tools, optimization
• other: Anything that doesn't fit above

Each skill file includes a YAML header with metadata:

---
name: Skill Name
description: What this skill does
source: https://github.com/user/repo
original_path: path/to/skill.md
source_repo: user/repo
updated_at: 2024-01-31T10:00:00Z
category: daily-assistant
subcategory: writing
tags: ['writing', 'content']
primary_purpose: Main function of this skill
file_hash: abc123...
---

# Original skill content...

tail -f logs/skillflow.log

python -m src.main --stats

python -c "from src.config import Config; c = Config(); print('GitHub token:', bool(c.github_token)); print('Anthropic key:', bool(c.anthropic_api_key))"

rm data/skills_tracker.db

MIT License - See LICENSE file for details