A collection of high-quality LLM Skills I've built, following skills.sh protocol best practices. Compatible with Claude, and other LLM platforms that support the open skill standard.
| Skill Name | Description | Installation |
|---|---|---|
| skill-reviewer | Review Skill quality against 10 best practices, identify design flaws, and provide actionable improvement suggestions. Supports automated checks and auto-fix. | npx skills add liwuhou/skills/skill-reviewer |
| multi-agent | Coordinate multiple agents to execute tasks in parallel. Automatically assesses task parallelizability, splits subtasks, and coordinates execution with isolated environments and flexible merge strategies. | npx skills add liwuhou/skills/multi-agent |
| codex-verify | Use OpenAI Codex CLI to independently verify code written by Claude. Two-phase verification (defect detection + quality assessment), safe Bash-array commands, automatic JSONL fallback, and user-approval-only fixes. | npx skills add liwuhou/skills/codex-verify |
- Node.js 16+
- Claude Code v0.10+
npx skills add <github-username>/<repo-name>/<skill-folder>
# Example: Install skill-reviewer
npx skills add liwuhou/skills/skill-reviewernpx skills add liwuhou/skillsAfter installation, skills are available in supported LLM clients (Claude Code, etc.) via slash commands:
/skill-reviewer <path-to-skill> # Review a skill
/skill-reviewer --quick # Quick check for critical issues
/skill-reviewer --fix # Auto-fix automatable issuesTo add a new skill:
- Create a new directory for your skill
- Follow the Skill Writing Guide
- Add an entry to the table in this README
- Add a README.md in your skill's directory
MIT License - feel free to use and modify these skills for your own projects.
Contributions are welcome! Feel free to submit PRs with improvements or new skills.