What Is SkillNet?

skillnet-ai is the installable Python package for SkillNet. It gives applications, scripts, and agent runtimes one interface for the skill lifecycle:

• Find skills with keyword or semantic search.
• Install skills from GitHub skill directories.
• Create structured skill packages from prompts, execution traces, repositories, or documents.
• Evaluate skill quality across safety, completeness, executability, maintainability, and cost awareness.
• Analyze how local skills compose into relationship graphs or scenario-level workflows.
• Orchestrate a preset scene by selecting the right skills and generating a downstream agent prompt.

Search and public skill download do not require an API key. Create, evaluate, and analyze use OpenAI-compatible endpoints. Orchestration runs through Claude Agent SDK and requires a compatible gateway configured through API_KEY, BASE_URL, and SKILLNET_MODEL.

For the full project overview, research context, integrations, and roadmap, see the main SkillNet repository.

Installation

pip install skillnet-ai

Optional extras:

pip install "skillnet-ai[graph]"        # scenario graph analysis
pip install "skillnet-ai[orchestrate]"  # scene orchestration via Claude Agent SDK

Quick Start

from skillnet_ai import SkillNetClient

client = SkillNetClient()

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

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

Equivalent CLI:

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

Feature Overview

Python SDK

Initialize

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:

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

Search

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)

Download

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

Create

# From a conversation or execution trace
client.create(
    trajectory_content="User: rename .jpg files 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
client.create(
    office_file="./guide.pdf",
    output_dir="./skills",
)

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

Created skills follow the standard layout:

skill-name/
├── SKILL.md
├── scripts/
├── references/
└── assets/

Evaluate

report = client.evaluate("./my_skills/table-extractor")

print(report["overall_score"])
print(report["summary"])

SkillNet evaluates five dimensions:

• Safety
• Completeness
• Executability
• Maintainability
• Cost awareness

Analyze

analyze reads a local directory of skill folders. The default basic mode infers lightweight relationships from skill names, descriptions, and metadata:

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

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

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

For workflow composition, scenario mode extracts each skill's required and produced scenarios, retrieves candidate handoffs with embeddings, verifies them with the configured LLM, and builds a directed skill graph. It analyzes local skill folders only.

Install the graph extra before using scenario mode:

pip install "skillnet-ai[graph]"

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",
    max_workers=4,
    top_k=30,
    timeout=120,
)

print(graph["scenario_skill_graph"]["edges"])
print(graph["relationships"])  # compatibility view

basic mode returns list[dict]. scenario mode returns a graph result dictionary with a relationships compatibility view.

Scenario mode writes graph artifacts under SKILLS_DIR/skillnet_graph by default:

skillnet_graph/
├── skill_scenarios.json
├── scenario_dedup.json
├── scenario_alignment.json
├── scenario_alignment_keep.json
├── skill_edge_redundancy_reviews.json
├── scenario_alignment_nonredundant_keep.json
├── scenario_skill_graph.json
└── relationships.json

Scenario embedding configuration is intentionally separate from chat LLM configuration. API_KEY, BASE_URL, and SKILLNET_MODEL configure extraction and verification; EMBEDDING_API_KEY, EMBEDDING_BASE_URL, and EMBEDDING_MODEL configure the embedding API.

Orchestrate

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)

Returned object:

result.package_url  # GitHub URL for the full skill collection
result.skills       # selected skills with skill_id and name
result.prompt       # downstream execution-agent prompt

CLI

The CLI is installed with the package:

skillnet --help
skillnet <command> --help

Commands

Search

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

Download

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 ./skills
skillnet create --github https://github.com/owner/repo
skillnet create --office ./docs/guide.pdf
skillnet create --prompt "A skill for table extraction"
skillnet create --office report.pdf --model gpt-4o

Evaluate

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

Analyze

Basic relationship analysis:

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

Scenario graph analysis requires skillnet-ai[graph] and an OpenAI-compatible embedding endpoint:

pip install "skillnet-ai[graph]"

export EMBEDDING_API_KEY="your-embedding-api-key"
export EMBEDDING_BASE_URL="https://embedding.example/v1"
export EMBEDDING_MODEL="your-embedding-model"

skillnet analyze ./my_skills --mode scenario \
  --output-dir ./my_skills/skillnet_graph \
  --max-workers 4 \
  --top-k 30 \
  --timeout 120

You can also pass --embedding-api-key, --embedding-base-url, and --embedding-model directly. --max-workers controls scenario extraction, candidate verification, and redundancy review concurrency. --top-k controls how many candidate scenario handoffs are retrieved for each produced scenario before LLM verification. Use --force to recompute existing artifacts.

Orchestrate

pip install "skillnet-ai[orchestrate]"

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

The terminal output includes the collection URL, selected skills, and the downstream agent prompt. Use --json when calling from scripts.

Configuration

search and public download require no credentials.

Linux / 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"

Skill Package Layout

Every created or downloaded skill uses the same basic structure:

skill-name/
├── SKILL.md          # required metadata and instructions
├── scripts/          # optional executable helpers
├── references/       # optional reference material
└── assets/           # optional templates, examples, or media

This layout keeps routing instructions, deterministic helper code, and heavier reference material separate so agents can load only what they need.

Contributing

Contributions are welcome. Open an issue for bugs, feature requests, or documentation improvements, and submit pull requests for focused changes.

License

MIT