AgentSkills-Hub

A universal agent skill development and management repository: providing specifications, templates, and script tools for quickly creating, validating, and packaging distributable skills.

中文 | 日本語

Key Features

• Multi-language Support (ZH/EN/JA): All public skills support Chinese, English, and Japanese output
• Zero-dependency Runtime: Core functionality works with Python 3.10+ only, no external packages required
• Graceful Degradation: Works without LLM/network; optional AI-powered translation
• Production-ready Templates: Battle-tested skill templates with i18n support built-in

Installation

One-line Install (Recommended)

Install skills directly using skills.sh:

# Install all skills from this repository
npx skills add deletexiumu/AgentSkills-Hub

# Or install a specific skill
npx skills add deletexiumu/AgentSkills-Hub/ai-news-digest
npx skills add deletexiumu/AgentSkills-Hub/smart-data-query
npx skills add deletexiumu/AgentSkills-Hub/x-ai-digest

Manual Install

# For Claude Code
mkdir -p ~/.claude/skills
git clone https://github.com/deletexiumu/AgentSkills-Hub.git
cp -R AgentSkills-Hub/skills/public/* ~/.claude/skills/

# For Codex
mkdir -p ~/.codex/skills
cp -R AgentSkills-Hub/skills/public/* ~/.codex/skills/

For more details, see: docs/skill-installation.md.

---

Quick Start (Create Your Own Skill)

Create a new skill (example: place it in skills/public):

python3 scripts/init_skill.py my-skill --path skills/public --resources scripts,references,assets

Validate and package:

python3 scripts/validate_skill.py skills/public/my-skill
python3 scripts/validate_i18n.py skills/public/my-skill   # i18n check
python3 scripts/package_skill.py skills/public/my-skill dist

For more workflows and specifications, see: docs/skill-workflow.md.

---

Available Skills

| Skill | Description | Highlights | |-------|-------------|------------| | ai-news-digest | Multi-source AI news aggregation and digest generation | 20+ sources, auto-dedup, 5 categories, image export | | smart-data-query | Intelligent data warehouse Q&A with executable SQL output | Auto-iteration, business questionnaire, Q&A logging | | x-ai-digest | X platform AI news scraper with reply suggestions | Browser automation, share card generation |

---

ai-news-digest

Multi-source AI news aggregation and digest generation with deduplication, classification, and source tracing.

Automatically fetches news from 20+ AI sources (OpenAI, Anthropic, DeepMind, Google AI, TechCrunch, Hugging Face, etc.), intelligently deduplicates and categorizes, and generates daily briefings in Markdown/JSON/Image format.

Why ai-news-digest?

| Advantage | Description | |-----------|-------------| | 20+ Curated Sources | OpenAI, Anthropic, DeepMind, Google AI, Meta AI, TechCrunch, Hugging Face, arXiv, and more | | Smart Deduplication | Same news from multiple sources automatically merged with cross-references | | 5 Theme Categories | Research / Products / Open Source / Funding / Policy - auto-classified | | Multi-language Output | Chinese, English, Japanese - use --lang zh/en/ja | | Natural Language Time | "today", "yesterday", "2026-01-20" - all supported | | Zero Dependencies | Only Python 3.10+ required for core functionality | | Multiple Output Formats | Markdown, JSON, shareable PNG images | | Graceful Degradation | Works without LLM; optional AI translation (Anthropic/OpenAI) |

Usage Examples

In Claude Code / Codex conversation:

# Chinese
"Use ai-news-digest to generate today's AI news digest"

# English
"Use ai-news-digest to generate today's AI news in English"

# Japanese
"ai-news-digest で今日のAIニュース要約を日本語で作成して"

CLI:

cd skills/public/ai-news-digest/scripts

# Default (Chinese)
python run.py --day today

# English output
python run.py --day yesterday --lang en

# Japanese output
python run.py --day today --lang ja

# Export shareable image
python run.py --day today --format image --image-preset landscape

Installation

# One-line install (recommended)
npx skills add deletexiumu/AgentSkills-Hub/ai-news-digest

# Or manual install for Claude Code
mkdir -p ~/.claude/skills
cp -R skills/public/ai-news-digest ~/.claude/skills/ai-news-digest

See: ai-news-digest/SKILL.md

---

smart-data-query

Intelligent data warehouse Q&A: input business requirements + DWH catalog, output executable SQL queries.

Transform natural language business requirements into production-ready SQL queries by intelligently searching your data warehouse catalog.

Why smart-data-query?

| Advantage | Description | |-----------|-------------| | Catalog-aware | Builds index from ADS/DWS/DWT tables, DDL, and ETL scripts | | Progressive Loading | Only loads relevant tables, keeps context minimal | | Multi-dialect Support | Hive, SparkSQL, GaussDB - handles syntax differences | | Business Questionnaire | Structured template ensures no requirements missed | | Q&A Logging | Every session logged for iteration and improvement | | Auto-iteration | Bad cases trigger automatic questionnaire optimization | | Static Validation | Checks field existence, join keys, partition pruning |

Usage Examples

# Chinese
"Use smart-data-query: query new users by channel for last 7 days"

# English
"Use smart-data-query: SQL for daily active users by channel"

# Japanese
"smart-data-query：チャネル別DAUを取得するSQLを作成して"

See: smart-data-query/SKILL.md

---

x-ai-digest

Scrape AI-related posts from X platform's "For You" feed, generate daily digest with reply suggestions.

Connects to your logged-in browser, scrapes AI-related posts from X's recommendation feed, and generates structured daily briefings with intelligent reply suggestions.

Why x-ai-digest?

| Advantage | Description | |-----------|-------------| | Real-time X Content | Scrapes live "For You" feed, not cached data | | AI Topic Filtering | Smart keyword matching for AI-related content | | Reply Suggestions | AI-generated reply ideas in original language | | Share Card Generation | Beautiful PNG cards for WeChat/social sharing | | Browser Integration | Uses your existing login, no API keys needed | | Multi-language Output | Digest and suggestions in ZH/EN/JA |

Usage Examples

# Chinese
"Use x-ai-digest to scrape today's AI hot topics from X"

# English
"Use x-ai-digest to summarize AI posts from yesterday in English"

# Japanese
"x-ai-digest で今日のAI関連投稿を日本語で要約して"

See: x-ai-digest/SKILL.md

---

i18n Specification

All public skills follow the i18n specification:

• Frontmatter: description: [ZH] 中文；[EN] English；[JA] 日本語
• Examples Block: Invoke examples in all three languages
• CLI Parameter: --lang auto|zh|en|ja
• Natural Language: Time expressions in all languages (today/今天/今日)

Validate i18n compliance:

python scripts/validate_i18n.py skills/public/ai-news-digest

---

Contributing

1. Fork this repository
2. Create your skill using scripts/init_skill.py
3. Follow the i18n specification in docs/skill-i18n.md
4. Validate with validate_skill.py and validate_i18n.py
5. Submit a pull request

License

MIT