SkillNet

Open infrastructure for discovering, evaluating, composing, and orchestrating reusable AI agent skills.

SkillNet treats agent skills as software assets: searchable, installable, inspectable, evaluable, and composable.

Website · Python SDK · Examples · Experiments · Paper

Agents should not rebuild the same capability from scratch every time. SkillNet provides the infrastructure layer for skill reuse:

• Discovery: search a public skill library by keyword or semantic intent.
• Installation: download skill folders from GitHub into local agent workspaces.
• Creation: generate structured skills from repositories, documents, prompts, or execution traces.
• Evaluation: score skills for safety, completeness, executability, maintainability, and cost awareness.
• Composition: infer relationships and scenario handoffs between local skills.
• Orchestration: select scene-specific skills and generate a prompt for a downstream execution agent.

Search and public skill installation are credential-free. Create, evaluate, and analyze work with OpenAI-compatible endpoints. Orchestration runs through Claude Agent SDK and requires a compatible gateway configured with the same API_KEY, BASE_URL, and SKILLNET_MODEL variables.

Install the SDK and CLI:

pip install skillnet-ai

Search and install 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(results[0].skill_url, target_dir="./my_skills")

CLI equivalent:

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

No API key is required for search or public GitHub downloads.

• [2026-07-11] SkillNet update. The library now indexes 500K+ GitHub skills with improved deduplication, expands scientific-research and data-analysis skill coverage, and adds local scenario graphs plus orchestration.
• [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.

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

Use it to:

• search skills by keyword or semantic meaning
• inspect quality-ranked skills and curated collections
• explore skill graph visualizations
• copy installable GitHub skill URLs

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

pip install skillnet-ai

Optional extras:

pip install "skillnet-ai[graph]"        # scenario-level graph analysis
pip install "skillnet-ai[orchestrate]"  # scene orchestration

from skillnet_ai import SkillNetClient

client = SkillNetClient(
    api_key="your-api-key",       # required for create, evaluate, analyze, orchestrate
    base_url="https://api.openai.com/v1",
    github_token=None,            # optional, for private repos or higher GitHub rate limits
)

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

results = client.search(
    q="analyze financial PDF reports",
    mode="vector",
    threshold=0.85,
    limit=10,
)

for skill in results:
    print(skill.skill_name, skill.stars, skill.skill_url)

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

client.create(
    prompt="A skill for extracting tables from academic PDFs",
    output_dir="./skills",
)

client.create(
    github_url="https://github.com/zjunlp/DeepKE",
    output_dir="./skills",
)

client.create(
    office_file="./guide.pdf",
    output_dir="./skills",
)

report = client.evaluate("./my_skills/table-extractor")
print(report["overall_score"])
print(report["summary"])

Basic relationship analysis:

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

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

Scenario graph analysis:

graph = client.analyze(
    "./my_skills",
    mode="scenario",
    embedding_api_key="your-embedding-api-key",
    embedding_base_url="https://embedding.example/v1",
    embedding_model="your-embedding-model",
    output_dir="./my_skills/skillnet_graph",
    timeout=120,
)

print(graph["scenario_skill_graph"]["edges"])

orchestrate requires API_KEY. It selects skills for a preset scene and returns a skill collection URL, selected skill names, and a downstream agent prompt. The first release supports scene="sciatlas". Its BASE_URL must support Claude Agent SDK requests; an OpenAI-only endpoint is not sufficient.

pip install "skillnet-ai[orchestrate]"

result = client.orchestrate(
    "Find recent papers on retrieval-augmented generation and propose three follow-up ideas.",
    scene="sciatlas",
    timeout=240,
)

print(result.package_url)
print([skill.name for skill in result.skills])
print(result.prompt)

The CLI ships with skillnet-ai.

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

skillnet search "pdf"
skillnet search "analyze financial reports" --mode vector --threshold 0.85

skillnet download <url> -d ./my_agent/skills
skillnet download <url> --mirror https://ghfast.top/

skillnet create --prompt "A skill for extracting tables from images"
skillnet evaluate ./my_skills/table_extractor
skillnet analyze ./my_skills

Scenario graph analysis:

pip install "skillnet-ai[graph]"

skillnet analyze ./my_skills --mode scenario \
  --embedding-api-key "$EMBEDDING_API_KEY" \
  --embedding-base-url "$EMBEDDING_BASE_URL" \
  --embedding-model "$EMBEDDING_MODEL" \
  --output-dir ./my_skills/skillnet_graph \
  --timeout 120

pip install "skillnet-ai[orchestrate]"

skillnet orchestrate "Find recent RAG papers and propose three follow-up ideas" \
  --scene sciatlas \
  --timeout 240

The command returns the SciAtlas skill collection URL, selected skills, and a downstream agent prompt. Use --json for machine-readable output.

Linux and macOS:

export API_KEY="your-api-key"
export BASE_URL="https://api.openai.com/v1"
export SKILLNET_MODEL="gpt-4o"

Windows PowerShell:

$env:API_KEY = "your-api-key"
$env:BASE_URL = "https://api.openai.com/v1"
$env:SKILLNET_MODEL = "gpt-4o"

search and public GitHub downloads require no credentials.

The SkillNet search API is public and requires no authentication.

curl "http://api-skillnet.openkg.cn/v1/search?q=pdf&sort_by=stars&limit=5"
curl "http://api-skillnet.openkg.cn/v1/search?q=reading%20charts&mode=vector&threshold=0.8"

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

SkillNet is packaged as a portable agent skill at skills/skillnet/. Install it into an agent runtime and the agent can search, download, create, evaluate, and analyze skills during coding or research tasks.

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

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

Try:

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

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"

Try:

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

The SkillNet MCP server is maintained by CycleChain.

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

Docker:

docker pull fmdogancan/skillnet-mcp:latest

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

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

• examples/: SDK demos and notebook workflows.
• experiments/: reproduction scripts for ALFWorld, WebShop, and ScienceWorld.
• Scientific workflow demo: using skills in a multi-step scientific discovery workflow.

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

• Broader scene orchestration beyond SciAtlas.
• More curated skill collections and routing wikis.
• Stronger skill evaluation and regression testing.
• SkillFabric workflow substrates for routing across skill collections.
• SkillGym lifecycle evaluation and training environments.

Contributions are welcome: bug fixes, documentation, examples, integrations, and new skills all help. Please keep pull requests focused and include reproduction steps or examples when possible.

If SkillNet is useful in your research or agent system, please cite:

@article{liang2026skillnet,
  title={Skillnet: Create, evaluate, and connect ai skills},
  author={Liang, Yuan and Zhong, Ruobin and Xu, Haoming and Jiang, Chen and Zhong, Yi and Fang, Runnan and Gu, Jia-Chen and Deng, Shumin and Yao, Yunzhi and Wang, Mengru and others},
  journal={arXiv preprint arXiv:2603.04448},
  year={2026}
}

MIT