Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
Version History
Skillnet 2.0.0 — Major Structural and Security Updates - Reorganized scripts: added skillnet_create.py and skillnet_validate.py, removed older script and setup files for improved maintainability. - Introduced references/security-privacy.md to document security and privacy practices. - Clarified usage: updated guidance on when to search, create, and apply skills, and emphasized security for downloaded scripts. - Refined metadata and streamlined task triggers in SKILL.md for easier integration and clearer logic. - Enforced GitHub-only download restriction and outlined strict user consent before executing any downloaded code.
What This Skill Does
SkillNet is a CLI for searching a global library of reusable agent skills, downloading them from GitHub, and creating new ones from repos, documents, or execution logs. Search requires no API key; creating, evaluating, and analyzing skills requires an OpenAI-compatible key. The core workflow is search first, build only if nothing fits, then preserve non-obvious solutions.
Instead of solving the same problem repeatedly, SkillNet lets agents search a shared skill library in seconds and reuse existing solutions, with a structured create-evaluate loop to contribute new knowledge back.
When to use it
- Search for an existing LangGraph skill before starting a multi-agent project
- Convert a PDF report into a reusable skill for future reference
- Package a complex debugging solution as a skill after completing it
- Evaluate the quality of a downloaded third-party skill before using it
- Analyze a local skill library to find duplicates and prune outdated entries
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Help me set up a multi-agent LangGraph system with a searcher, coder, and reviewer agent
- 1Run `skillnet search "langgraph" --limit 5` to check for existing skills
- 2Confirm with user, then download a relevant result and preview its SKILL.md before loading
- 3Extract the supervisor routing pattern and state schema; build the three specialized agents from scratch
- 4After completing the task, run `skillnet create --prompt "..."` to capture the solution
- 5Run `skillnet evaluate` on the new skill to verify Safety, Completeness, and Executability scores
Working multi-agent LangGraph pipeline with a reusable skill saved to ~/.openclaw/workspace/skills
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.