SkillNet: Create, Evaluate, and Connect AI Skills

• 📖 Overview
• 🚀 Features
• 🌐 API Access
• 🐍 Python Toolkit (skillnet-ai)
  • 📥 Installation
  • 🛠 Python SDK Usage
  • 💻 CLI Usage
• 📂 Skill Structure
• 🗺 Roadmap & Contributing

SkillNet is an open infrastructure for creating, evaluating, and organizing AI skills at scale.

• 🔍 Search: Find skills using keywords match or semantic search.
• 📦 One-Line Installation: Download skill packages directly from GitHub repositories.
• ✨ Skill Creation: Automatically convert various sources into structured, reusable skills using LLMs:
  • Execution trajectories / conversation logs
  • GitHub repositories
  • Office documents (PDF, PPT, Word)
  • Direct text prompts
• 📊 Evaluation: Evaluate and score skills for quality assurance (Safety, Completeness, Excutability, Modifiability, Cost-Aware).
• 🕸️ Relationship Analysis: Automatically map the connections between skills in your local library, identifying structural relationships between skills (similar_to, belong_to, compose_with, depend_on).

SkillNet provides a public API to search skills. Support both keywords match and semantic search.

Base Endpoint: http://api-skillnet.openkg.cn/v1/search

1. Keywords Match

Find "development" tools sorted by stars.

curl -X GET "http://api-skillnet.openkg.cn/v1/search?q=pdf&sort_by=stars&limit=5" \
     -H "accept: application/json"

2. Vector Semantic Search

Find skills related to "reading charts" using AI similarity.

curl -X GET "http://api-skillnet.openkg.cn/v1/search?q=reading%20charts&mode=vector&threshold=0.8" \
     -H "accept: application/json"

Mode Specific Parameters:

• Keyword Mode: page (int), min_stars (int), sort_by (string: stars or recent)
• Vector Mode: threshold (float: 0.0 to 1.0)

{
  "data": [
    {
      "skill_name": "pdf-extractor-v1",
      "skill_description": "Extracts text and tables from PDF documents.",
      "author": "openkg-team",
      "stars": 128,
      "skill_url": "http://...",
      "category": "Productivity"
    }
  ],
  "meta": {
    "query": "pdf",
    "mode": "keyword",
    "total": 1,
    "limit": 10,
    ...
  },
  "success": true
}

skillnet-ai is the official Python Toolkit. It functions seamlessly as both a library and a CLI to Create, Evaluate, and Organize skills.

pip install skillnet-ai

The SkillNetClient is your main entry point.

from skillnet_ai import SkillNetClient

client = SkillNetClient(
    api_key="sk-...",       # Required for Creation, and Evaluation
    # base_url="...",       # Optional: Custom LLM base URL
    # github_token="ghp-..." # Optional: For private repos or higher rate limits
)

Perform keywords match or semantic searches to find skills. (See Parameter Reference for configuration details.)

# 1. Standard Keywords Match
results = client.search(q="pdf")

# 2. Semantic Search
results = client.search(q="Help me analyze financial PDF reports", mode="vector")

if results:
    top_skill = results[0]
    print(f"Found: {top_skill.skill_name} (Stars: {top_skill.stars})")
    print(f"URL: {top_skill.skill_url}")

Download and install a skill directly from a URL (e.g., from above search results) into your local workspace.

skill_url = "https://github.com/anthropics/skills/tree/main/skills/skill-creator"

try:
    # Downloads to ./my_agent_skills
    local_path = client.download(url=skill_url, target_dir="./my_agent_skills")
    print(f"Skill successfully installed at: {local_path}")
except Exception as e:
    print(f"Download failed: {e}")

Turn local trajectory, gitHub repository, office documents or text description into a polished Skill Package (SKILL.md, scripts, etc.).

# 1. Create skill from Local Trajectory
# Prepare your trajectory (e.g., a conversation log string)
trajectory_log = """
User: I need to rename all .jpg files in this folder to .png.
Agent: I will write a python script to iterate through the folder...
Agent: Script executed. Renamed 5 files.
"""
# Generate Skill, Returns a list of paths to the generated skill folders
created_paths = client.create(
    trajectory_content=trajectory_log,
    output_dir="./created_skills",
    model="gpt-4o"
)

# 2. Create skill from GitHub Repository
created_paths = client.create(
    github_url="https://github.com/zjunlp/DeepKE",
    output_dir="./created_skills",
    model="gpt-4o"
)

# 3. Create skill from a office documents (PDF, Word, PPT)
created_paths = client.create(
    office_file="./docs/user_guide.pdf",
    output_dir="./created_skills"
)

# 4. Create skill from a prompt description
created_paths = client.create(
    prompt="Create a skill for web scraping that extracts article titles and content",
    output_dir="./created_skills"
)

print(f"Created {len(created_paths)} new skills.")
for path in created_paths:
    print(f"- {path}")

Assess the Safety, Completeness, Executability, Modifiability and Cost-Aware of a skill. Supports both remote GitHub URLs and local directories.

# Evaluate from local directory
# target_skill = "./my_skills/web_search"

# Evaluate from GitHub URL (uses github_token if provided during initialization)
target_skill = "https://github.com/anthropics/skills/tree/main/skills/algorithmic-art"

result = client.evaluate(target=target_skill)
print(f"Evaluation Result: {result}")

Analyze a local directory containing multiple skills to infer a relationship graph. It identifies relationships like dependencies (depend_on), collaboration (compose_with), hierarchy (belong_to), and alternatives (similar_to).

# Directory containing multiple skill folders
skills_directory = "./my_agent_skills"

# Analyze relationships between skills
# This will also save a 'relationships.json' in the directory by default
relationships = client.analyze(skills_dir=skills_directory)

# Display the relationships
for rel in relationships:
    print(f"{rel['source']} --[{rel['type']}]--> {rel['target']}")
    # Output: PDF_Parser --[compose_with]--> Text_Summarizer

The CLI is powered by Typer and Rich for a beautiful terminal experience.

Tip: Use skillnet [command] --help to see all available options (e.g., thresholds, sorting).

Search the registry using keywords match or semantic search.

# Basic keywords match
skillnet search "pdf"

# Semantic/Vector search (finds skills by meaning)
skillnet search "Help me analyze financial PDF reports" --mode vector --threshold 0.85

# Filter by category and sort results
skillnet search "visualization" --category "Development" --sort-by stars --limit 10

Download and install a skill directly from a GitHub repository subdirectory.

# Download to the current directory
skillnet download https://github.com/anthropics/skills/tree/main/skills/algorithmic-art

# Download to a specific target directory
skillnet download https://github.com/anthropics/skills/tree/main/skills/algorithmic-art -d ./my_agent/skills

# Download from a private repository
skillnet download <private_url> --token <your_github_token>

Create structured Skill from various sources using LLMs.

Linux/macOS:

# Requirement: Ensure API_KEY is set in your environment variables.
export API_KEY=sk-xxxxx
export BASE_URL=xxxxxx  # Optional custom LLM base URL

Windows PowerShell:

# Requirement: Ensure API_KEY is set in your environment variables.
$env:API_KEY = "sk-xxxxx"
$env:BASE_URL = "xxxxxx"  # Optional custom LLM base URL

Usage Examples:

# From a trajectory file
skillnet create ./logs/trajectory.txt -d ./generated_skills

# From a GitHub repository
skillnet create --github https://github.com/owner/repo

# From an office document (PDF, PPT, Word)
skillnet create --office ./docs/guide.pdf

# From a direct prompt
skillnet create --prompt "Create a skill for extracting tables from images"

# Specify a custom model
skillnet create --office report.pdf --model gpt-4o

Generate a comprehensive quality report (Safety, Completeness, Executability, Modifiability, Cost Awareness) for a skill.

Linux/macOS:

# Requirement: Ensure API_KEY is set in your environment variables.
export API_KEY=sk-xxxxx
export BASE_URL=xxxxxx  # Optional custom LLM base URL

Windows PowerShell:

# Requirement: Ensure API_KEY is set in your environment variables.
$env:API_KEY = "sk-xxxxx"
$env:BASE_URL = "xxxxxx"  # Optional custom LLM base URL

Usage Examples:

# Evaluate a remote skill via GitHub URL
skillnet evaluate https://github.com/anthropics/skills/tree/main/skills/algorithmic-art

# Evaluate a local skill directory
skillnet evaluate ./my_skills/web_search

# Custom evaluation config
skillnet evaluate ./my_skills/tool --category "Development" --model gpt-4o

Scan a local directory of skills to analyze their connections using AI.

Linux/macOS:

# Requirement: Ensure API_KEY is set in your environment variables.
export API_KEY=sk-xxxxx
export BASE_URL=xxxxxx  # Optional custom LLM base URL

Windows PowerShell:

# Requirement: Ensure API_KEY is set in your environment variables.
$env:API_KEY = "sk-xxxxx"
$env:BASE_URL = "xxxxxx"  # Optional custom LLM base URL

Usage Examples:

# Analyze a directory containing multiple skill folders
skillnet analyze ./my_agent_skills

# Analyze without saving the result file (just print to console)
skillnet analyze ./my_agent_skills --no-save

# Specify a model for the analysis
skillnet analyze ./my_agent_skills --model gpt-4o

To use Creation, Evaluation, or Analyze features, set your environment variables:

Linux/macOS:

export API_KEY="your_api_key"
export BASE_URL="https://xxxxx"  # Optional

Windows PowerShell:

$env:API_KEY = "your_api_key"
$env:BASE_URL = "https://xxxxx"  # Optional

Standardized structure for all SkillNet packages:

skill-name/
├── SKILL.md          # [Required] Metadata (YAML) + Instructions
├── scripts/          # [Optional] Executable Python/Bash scripts
├── references/       # [Optional] Static docs or API specs
└── assets/           # [Optional] Icons, templates, examples

• ✅ Keyword Match & Semantic Search
• ✅ Skill Installer
• ✅ Skill Creator (Local File & GitHub Repository)
• ✅ Skill Evaluation & Scoring
• ✅ Skill Relationship Analysis

Contributions are welcome! Please submit a Pull Request or open an Issue on GitHub.

This project is licensed under the MIT License.