OpenClaw Skills Guide

Word of advice: The more popular OpenClaw becomes, the more you should study Skills — don't miss the point!

Why Skills Matter

OpenClaw (affectionately known as "lobster") has seen explosive growth recently. Major cloud providers have added one-click deployment, and LLM companies launched Coding Plans to integrate with OpenClaw. But the real power of OpenClaw lies in Skills.

Core Value of Skills

• Capability Extension: Skills give OpenClaw new tools and abilities
• Automation: Encapsulate repetitive tasks into reusable skills
• Specialization: Create domain-specific skills for different scenarios

OpenClaw Skills Architecture

Skill Loading Locations

OpenClaw loads Skills from three locations:

Location	Priority	Description
<workspace>/skills	Highest	Workspace-specific skills
~/.openclaw/skills	Medium	Shared skills, visible to all agents
bundled skills	Lowest	Built-in skills

Skill Format

Each Skill is a directory containing SKILL.md:

---
name: skill-name
description: Skill description
metadata: {"openclaw": {"requires": {"bins": ["tool"]}}}
---

# Skill instructions

ClawHub - Skills Marketplace

Visit ClawHub to discover, install, and sync Skills:

# Install skill to workspace
clawhub install <skill-slug>

# Update all skills
clawhub update --all

# Sync and publish
clawhub sync --all

Recommended Skills

1. coding-agent

• Purpose: Delegate coding tasks to Codex, Claude Code
• Scenarios: Feature building, code review, refactoring

2. skill-creator

• Purpose: Create and optimize AgentSkills
• Scenarios: Custom skill development

3. github

• Purpose: GitHub operations (issues, PRs, CI)
• Scenarios: Project management, code review

4. weather

• Purpose: Get weather and forecasts
• Scenarios: Daily queries, integrate into other skills

5. 1password

• Purpose: Secret management
• Scenarios: Secure storage and injection

6. gog

• Purpose: Gmail/Calendar/Drive one-click management
• Scenarios: Email automation, calendar scheduling, file management

7. youtube-full

• Purpose: Video summary tool
• Scenarios: Extract video insights, generate summaries, content analysis

8. gemini-cli

• Purpose: AI research assistant
• Scenarios: Research queries, data analysis, AI-powered insights

9. coding-agent

• Purpose: Code without asking for help
• Scenarios: Autonomous code generation, debugging, refactoring

10. weather

• Purpose: Daily weather push notifications
• Scenarios: Weather alerts, forecast integration

11. apple-reminders

• Purpose: To-do reminders
• Scenarios: Task management, reminder automation

12. spotify/apple-music

• Purpose: Voice-controlled music
• Scenarios: Music playback control, playlist management

13. twitter

• Purpose: Auto-tweet management
• Scenarios: Social media automation, content scheduling

14. clickup

• Purpose: Project management
• Scenarios: Task tracking, team collaboration

15. todoist

• Purpose: Task list management
• Scenarios: Personal productivity, task organization

16. jira

• Purpose: Workflow management
• Scenarios: Issue tracking, sprint planning

17. obsidian

• Purpose: Note organization
• Scenarios: Knowledge management, note linking

18. slack

• Purpose: Message management
• Scenarios: Team communication, notification handling

19. fal-ai

• Purpose: AI image generation
• Scenarios: Create images, visual content generation

20. elevenlabs

• Purpose: Text-to-speech
• Scenarios: Voice synthesis, audio content creation

21. polymarket

• Purpose: Prediction markets
• Scenarios: Market analysis, prediction tracking

22. n8n

• Purpose: Workflow automation
• Scenarios: Complex automation, integration workflows

23. linear

• Purpose: Team collaboration
• Scenarios: Issue tracking, project planning

24. context7

• Purpose: Document query
• Scenarios: Search documentation, knowledge retrieval

25. cron-natural

• Purpose: Scheduled tasks
• Scenarios: Cron job management, task scheduling

How to Choose the Right Skills?

Match by Need

1. Development Efficiency → coding-agent, github
2. Content Creation → writing-skills, knowledge-site-creator
3. DevOps → healthcheck, system-maintenance
4. Research & Learning → research, web-search

Check Compatibility

Check requirements before using a skill:

metadata: {"openclaw": {"requires": {"bins": ["python"], "env": ["API_KEY"]}}}

• bins: Command-line tools that must be on PATH
• env: Required environment variables

Best Practices

1. Start with Official Skills

Built-in Skills are thoroughly tested for stability.

2. Vet Third-Party Skills Carefully

⚠️ Security Note: Treat third-party Skills as untrusted code. Always read the source before enabling.

3. Combine Skills Synergistically

Multiple Skills can work together for greater capabilities.

4. Build Custom Skills

When existing skills don't meet your needs:

# Create skill directory structure
mkdir -p my-skill/{scripts,references,assets}

FAQ

Q: What's the difference between Skills and Plugins?

• Skills: Teach the agent how to use tools
• Plugins: Extend OpenClaw core functionality

Q: How to debug Skills?

Use --verbose to see skill loading logs.

Q: Are Skills free?

Most official Skills are free; some third-party ones may be paid.

Conclusion

The more popular OpenClaw becomes, the more important it is to focus on studying Skills. That's the real key to unlocking OpenClaw's full potential.

---

Related Links:

• OpenClaw Official Docs
• ClawHub Skills Marketplace
• Skill Development Tutorial

Related reading

• OpenClaw Agent Architecture Explained | AI Agent Runtime System: Deep dive into OpenClaw Agent execution architecture. Learn the complete message flow from channel adapter to response output, session scheduling, and the Agentic Loop. Production-ready AI agent system design guide.
• Start Here | Guide Tracks for AI Agents, Engineering, Frontend, and LLM Work: Follow CoworkAI's structured guide tracks for AI agents, backend engineering, frontend implementation, and LLM foundations instead of jumping randomly between isolated posts.
• LLM, RAG & AI Agents: Complete Engineering Guide from Concept to Production: Comprehensive guide to LLM, RAG, and AI Agents three-layer architecture - thinking, memory, and execution layers with implementation patterns, engineering practices, and production deployment strategies

What to open next

• Continue with the guide tracks: place this page back inside a larger collection or reading path instead of ending the session here.
• OpenClaw Agent Architecture Explained | AI Agent Runtime System: Deep dive into OpenClaw Agent execution architecture. Learn the complete message flow from channel adapter to response output, session scheduling, and the Agentic Loop. Production-ready AI agent system design guide.
• Start Here | Guide Tracks for AI Agents, Engineering, Frontend, and LLM Work: Follow CoworkAI's structured guide tracks for AI agents, backend engineering, frontend implementation, and LLM foundations instead of jumping randomly between isolated posts.
• LLM, RAG & AI Agents: Complete Engineering Guide from Concept to Production: Comprehensive guide to LLM, RAG, and AI Agents three-layer architecture - thinking, memory, and execution layers with implementation patterns, engineering practices, and production deployment strategies
• Bookmark the homepage: keep the workspace one click away so new additions are easy to revisit.
• Subscribe by RSS: RSS is the cleanest return channel here if you want updates without email capture.
• Suggest a tool or topic: send the next gap you want this site to cover.