SkillNet: The Operating System for AI Skills

Search, install, evaluate, and connect reusable AI skills.

SkillNet treats AI agent skills as first-class software assets.
It provides a public skill library, Python SDK, CLI, quality evaluation, and a skill graph.

Get started · Explorer · Python SDK · CLI · REST API · Citation

---

What is SkillNet?

SkillNet is open infrastructure for building and reusing AI agent skills. A skill can package instructions, metadata, references, scripts, and evaluation results so that an agent can install it, inspect it, and use it again later.

Use SkillNet when you want to:

• Find existing skills before an agent rebuilds the same capability.
• Install skills from public GitHub repositories into a local agent workspace.
• Create new skills from repositories, office documents, prompts, or execution traces.
• Evaluate skills across five quality dimensions.
• Build a local graph that shows how skills relate to each other.

Get started

Install the Python package:

pip install skillnet-ai

Search and download a skill:

from skillnet_ai import SkillNetClient

client = SkillNetClient()

results = client.search("pdf understanding", limit=5)
print(results[0].skill_name)
print(results[0].skill_url)

client.download(url=results[0].skill_url, target_dir="./my_skills")

Or use the CLI:

skillnet search "pdf understanding" --limit 5
skillnet download <skill_url> -d ./my_skills

search and public GitHub downloads do not require credentials. Set API_KEY when you use create, evaluate, or analyze with an LLM.

News

• [2026-03-26] JiuwenClaw integration released. JiuwenClaw now includes SkillNet as a built-in skill marketplace. View guide
• [2026-03-12] SkillNet MCP server released. MCP support is maintained by CycleChain.
• [2026-03-04] Technical report released. Read the SkillNet report on arXiv.
• [2026-02-23] OpenClaw integration released. SkillNet is available as a built-in skill for OpenClaw.

Table of contents

• SkillNet Explorer
• Core capabilities
• Use SkillNet in Code Agents
• Python SDK
• CLI Reference
• Configuration
• REST API
• Examples and experiments
• More integrations
• Roadmap
• Contributing
• Citation

---

SkillNet Explorer

SkillNet Explorer is the visual entry point for the public skill library. It is designed for browsing skills the way developers browse packages or model hubs.

Use the Explorer to:

• Search skills by keyword or semantic meaning.
• Browse quality-ranked skills and curated collections.
• Inspect skill graph visualizations.
• Find the install path for a skill.

The website also includes interactive scenarios for web scraping, paper summarization, and experiment planning.

skillnet.mp4

---

Core capabilities

Capability	What it does
Search	Find skills through keyword search or semantic vector search.
Install	Download a skill folder from GitHub into a local workspace.
Create	Convert repositories, documents, prompts, or trajectories into structured skill packages.
Evaluate	Score skills for Safety, Completeness, Executability, Maintainability, and Cost-Awareness.
Connect	Infer similar_to, belong_to, compose_with, and depend_on relationships between skills.

SkillNet treats a skill as a software primitive, not just a prompt. The package format keeps the skill portable; the evaluation layer makes quality visible; the graph layer makes skills easier to compose.

---

Use SkillNet in Code Agents

SkillNet is also packaged as a portable agent skill at skills/skillnet/. Install this folder into your code agent's local skills directory, then the agent can search, download, create, evaluate, and organize skills during coding tasks.

claudecode-skillnet.mp4

Claude Code

Claude Code discovers user skills from ~/.claude/skills/ and project skills from .claude/skills/.

Install as a user skill:

git clone https://github.com/zjunlp/SkillNet.git
cd SkillNet

mkdir -p ~/.claude/skills
cp -R skills/skillnet ~/.claude/skills/skillnet

Or install as a project-local skill:

mkdir -p .claude/skills
cp -R /path/to/SkillNet/skills/skillnet .claude/skills/skillnet

Restart Claude Code or start a new session, then try:

Use SkillNet to search for a docker skill and summarize the top result.

Codex

Codex discovers user skills from $CODEX_HOME/skills. If CODEX_HOME is not set, use ~/.codex/skills.

git clone https://github.com/zjunlp/SkillNet.git
cd SkillNet

CODEX_HOME="${CODEX_HOME:-$HOME/.codex}"
mkdir -p "$CODEX_HOME/skills"
cp -R skills/skillnet "$CODEX_HOME/skills/skillnet"

Restart Codex or start a new session, then try:

Use $skillnet to search for a LangGraph skill before planning this task.

---

Python SDK

Initialize

from skillnet_ai import SkillNetClient

client = SkillNetClient(
    api_key="YOUR_API_KEY",          # Required for create, evaluate, and analyze
    # base_url="https://api.openai.com/v1",
    # github_token="YOUR_GITHUB_TOKEN"
)

Credentials can also be set through environment variables: API_KEY, BASE_URL, and GITHUB_TOKEN.

Search

# Keyword search
results = client.search(q="pdf", limit=10, min_stars=5, sort_by="stars")

# Semantic search
results = client.search(
    q="analyze financial PDF reports",
    mode="vector",
    threshold=0.85,
)

if results:
    print(results[0].skill_name)
    print(results[0].skill_url)

Install

local_path = client.download(
    url="https://github.com/anthropics/skills/tree/main/skills/skill-creator",
    target_dir="./my_skills",
)
print(local_path)

Create

create requires API_KEY.

# From a conversation log or execution trace
client.create(
    trajectory_content="User: rename .jpg to .png\nAgent: Done.",
    output_dir="./skills",
)

# From a GitHub repository
client.create(github_url="https://github.com/zjunlp/DeepKE", output_dir="./skills")

# From an office document: PDF, PPT, or Word
client.create(office_file="./guide.pdf", output_dir="./skills")

# From a natural language prompt
client.create(prompt="A skill for web scraping article titles", output_dir="./skills")

Evaluate

evaluate requires API_KEY. It accepts local paths and GitHub URLs.

result = client.evaluate(
    target="https://github.com/anthropics/skills/tree/main/skills/algorithmic-art"
)

print(result["safety"]["level"])
print(result["executability"]["reason"])

Analyze

analyze requires API_KEY. It maps relationships between skills in a local directory.

relationships = client.analyze(skills_dir="./my_skills")

for rel in relationships:
    print(f"{rel['source']} --[{rel['type']}]--> {rel['target']}")

Relationship types: similar_to, belong_to, compose_with, and depend_on.

---

CLI Reference

The CLI ships with pip install skillnet-ai.

Command	Description	Example
search	Find skills	skillnet search "pdf" --mode vector
download	Install a skill	skillnet download <url> -d ./skills
create	Create from repos, docs, logs, or prompts	skillnet create log.txt --model gpt-4o
evaluate	Quality report	skillnet evaluate ./my_skill
analyze	Relationship graph	skillnet analyze ./my_skills

Use skillnet <command> --help for full options.

Search

skillnet search "pdf"
skillnet search "analyze financial reports" --mode vector --threshold 0.85
skillnet search "visualization" --category "Development" --sort-by stars --limit 10

Install

skillnet download https://github.com/anthropics/skills/tree/main/skills/algorithmic-art
skillnet download <url> -d ./my_agent/skills
skillnet download <private_url> --token <your_github_token>
skillnet download <url> --mirror https://ghfast.top/

Create

skillnet create ./logs/trajectory.txt -d ./generated_skills
skillnet create --github https://github.com/owner/repo
skillnet create --office ./docs/guide.pdf
skillnet create --prompt "A skill for extracting tables from images"

Evaluate

skillnet evaluate https://github.com/anthropics/skills/tree/main/skills/algorithmic-art
skillnet evaluate ./my_skills/web_search
skillnet evaluate ./my_skills/tool --category "Development" --model gpt-4o

Analyze

skillnet analyze ./my_agent_skills
skillnet analyze ./my_agent_skills --no-save
skillnet analyze ./my_agent_skills --model gpt-4o

---

Configuration

Variable	Required for	Default
API_KEY	create, evaluate, analyze	-
BASE_URL	Custom OpenAI-compatible endpoint	https://api.openai.com/v1
GITHUB_TOKEN	Private repositories or higher rate limits	-
SKILLNET_MODEL	Default LLM model for all commands	gpt-4o
GITHUB_MIRROR	Faster GitHub downloads in restricted networks	-

search and public GitHub downloads work without credentials.

Linux and macOS:

export API_KEY="YOUR_API_KEY"
export BASE_URL="https://..."

Windows PowerShell:

$env:API_KEY = "YOUR_API_KEY"
$env:BASE_URL = "https://..."

---

REST API

The SkillNet search API is public and requires no authentication.

# Keyword search
curl "http://api-skillnet.openkg.cn/v1/search?q=pdf&sort_by=stars&limit=5"

# Semantic search
curl "http://api-skillnet.openkg.cn/v1/search?q=reading%20charts&mode=vector&threshold=0.8"

Full parameter reference

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

Parameter	Type	Default	Description
q	string	required	Search query, keywords or natural language
mode	string	keyword	keyword or vector
category	string	-	Filter: Development, AIGC, Research, Science, etc.
limit	int	10	Results per page, max 50
page	int	1	Page number, keyword mode only
min_stars	int	0	Minimum star count, keyword mode only
sort_by	string	stars	stars or recent, keyword mode only
threshold	float	0.8	Similarity threshold from 0.0 to 1.0, vector mode only

Response:

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

---

Examples and experiments

Scientific discovery

SkillNet can help an agent plan and execute a scientific workflow, from raw scRNA-seq data to a cancer target validation report.

Step	What happens
Task	User asks: "Analyze scRNA-seq data to find cancer targets"
Plan	Agent decomposes the job into data, mechanism, validation, and report steps
Discover	client.search() finds useful skills such as cellxgene-census and kegg-database
Evaluate	Skills are quality-gated before use
Execute	Skills run in sequence and produce the final report

Try the interactive demo or open the scientific workflow notebook.

Benchmark scripts

Reproduction scripts for ALFWorld, WebShop, and ScienceWorld are available under experiments/.

cd experiments

python alfworld_run.py --model o4-mini --split dev --max_workers 10 --exp_name alf_test --use_skill
python scienceworld_run.py --model o4-mini --split test --max_workers 5 --exp_name sci_test --use_skill
python webshop_run.py --model o4-mini --max_workers 3 --exp_name web_test --use_skill

---

More integrations

OpenClaw

SkillNet integrates with OpenClaw as a built-in, lazy-loaded skill. The agent can search, download, create, evaluate, and analyze skills from inside OpenClaw.

openclaw-skillnet.mp4

Install with the CLI:

npm i -g clawhub
clawhub install skillnet --workdir ~/.openclaw/workspace
openclaw gateway restart

Or ask OpenClaw:

Install the skillnet skill from ClawHub.

Optional OpenClaw configuration:

{
  "skills": {
    "entries": {
      "skillnet": {
        "enabled": true,
        "apiKey": "YOUR_API_KEY",
        "env": {
          "BASE_URL": "https://api.openai.com/v1",
          "GITHUB_TOKEN": "YOUR_GITHUB_TOKEN"
        }
      }
    }
  }
}

Model Context Protocol (MCP)

The SkillNet MCP server is maintained by CycleChain. It lets MCP-compatible clients such as Claude Desktop, Cursor, Antigravity, and Windsurf call SkillNet tools directly.

Source build:

git clone https://github.com/CycleChain/skillnet-mcp
cd skillnet-mcp
npm install && npm run build

Docker:

docker pull fmdogancan/skillnet-mcp:latest

Claude Desktop configuration with Docker:

{
  "mcpServers": {
    "skillnet": {
      "command": "docker",
      "args": ["run", "-i", "--rm", "fmdogancan/skillnet-mcp:latest"],
      "env": {
        "API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

search_skills and download_skill do not require an API key. create, evaluate, and analyze do.

JiuwenClaw

JiuwenClaw integrates SkillNet as a built-in skill marketplace. See the JiuwenClaw guide.

---

Roadmap

SkillNet is growing beyond search and installation.

• SkillFabric: workflow-level skill substrates and routing across skill collections.
• SkillGym: lifecycle evaluation and training environments for skills.

Start from skillnet.openkg.cn and open SkillFabric or SkillGym from the website navigation.

---

Contributing

Contributions are welcome: bug fixes, docs, examples, and new skills all help.

1. Fork the repository.
2. Create a feature branch: git checkout -b feat/amazing-feature.
3. Commit your changes: git commit -m 'feat: add amazing feature'.
4. Push to the branch: git push origin feat/amazing-feature.
5. Open a pull request.

You can also open an issue or contribute skills through the SkillNet website.

---

Citation

If SkillNet helps your work, please cite the paper:

@misc{liang2026skillnetcreateevaluateconnect,
      title={SkillNet: Create, Evaluate, and Connect AI Skills},
      author={Yuan Liang and Ruobin Zhong and Haoming Xu and Chen Jiang and Yi Zhong and Runnan Fang and Jia-Chen Gu and Shumin Deng and Yunzhi Yao and Mengru Wang and Shuofei Qiao and Xin Xu and Tongtong Wu and Kun Wang and Yang Liu and Zhen Bi and Jungang Lou and Yuchen Eleanor Jiang and Hangcheng Zhu and Gang Yu and Haiwen Hong and Longtao Huang and Hui Xue and Chenxi Wang and Yijun Wang and Zifei Shan and Xi Chen and Zhaopeng Tu and Feiyu Xiong and Xin Xie and Peng Zhang and Zhengke Gui and Lei Liang and Jun Zhou and Chiyu Wu and Jin Shang and Yu Gong and Junyu Lin and Changliang Xu and Hongjie Deng and Wen Zhang and Keyan Ding and Qiang Zhang and Fei Huang and Ningyu Zhang and Jeff Z. Pan and Guilin Qi and Haofen Wang and Huajun Chen},
      year={2026},
      eprint={2603.04448},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2603.04448}
}