OpenClaw Skills: Extend Your AI Assistant's Capabilities
OpenClaw Skills: Extend Your AI Assistant's Capabilities
Last Updated: February 2026
Skills are what transform OpenClaw from a chatbot into a true autonomous agent. This guide covers what skills are, how to use them, and how to create your own.
What Are OpenClaw Skills?
Skills are modular capability packages that teach OpenClaw how to do specific things. They contain:
- Instructions — What the skill does and when to use it
- Tools — Scripts and commands the skill can execute
- Configuration — Settings and API credentials
Think of skills like apps for your AI assistant. Just as you install apps on your phone to add functionality, you install skills on OpenClaw to expand what it can do.
Built-In Skills
OpenClaw comes with 50+ skills out of the box:
Productivity
- Gmail — Read, write, and organize emails
- Google Calendar — Schedule events, check availability
- Apple Reminders — Create and manage reminders
- Things 3 — Task management for Mac users
Development
- GitHub — Issues, PRs, repo management
- Coding Agent — Run Claude Code, Codex, or other AI coders
- Oracle — Query multiple AI models
Communication
- iMessage — Send messages on Mac
- WhatsApp — Message via wacli integration
- Himalaya — Email via IMAP/SMTP
Utilities
- Weather — Current conditions and forecasts
- Web Search — Brave Search integration
- Firecrawl — Web scraping and research
- 1Password — Secure credential management
How to Use Skills
List Available Skills
openclaw skills list
Or ask your assistant:
"What skills do you have?"
Install a Skill
Most skills are bundled with OpenClaw. For community skills:
# From ClawHub
clawdhub install weather
# From GitHub
git clone https://github.com/user/skill-name ~/.openclaw/skills/skill-name
Enable/Disable Skills
Skills are enabled by placing them in your workspace:
~/clawd/skills/ # Your custom skills
~/.openclaw/skills/ # Shared skills
OpenClaw automatically discovers and loads skills from these directories.
Skill Structure
Every skill has a SKILL.md file with this structure:
---
name: weather
description: Get weather forecasts and current conditions
---
# Weather Skill
## Usage
Ask about weather in any location:
- "What's the weather in Tokyo?"
- "Will it rain tomorrow?"
## Commands
\`\`\`bash
curl wttr.in/location
\`\`\`
The description field is crucial—OpenClaw uses it to decide when to activate the skill.
Creating Your Own Skills
Method 1: Ask OpenClaw
The easiest way:
"Create a skill to check my Todoist tasks"
OpenClaw will generate the skill files for you.
Method 2: Manual Creation
Create a directory with the required files:
my-skill/
├── SKILL.md # Instructions (required)
├── tools/ # Scripts (optional)
│ └── check.sh
└── config.json # Configuration (optional)
SKILL.md Example:
---
name: todoist
description: Manage Todoist tasks - list, create, complete, and delete tasks
---
# Todoist Integration
When the user asks about tasks or todos, use the Todoist CLI.
## Available Commands
- `todoist list` — Show all tasks
- `todoist add "task name"` — Create a task
- `todoist complete <id>` — Mark task as done
## Requirements
- Todoist CLI installed: `pip install todoist-cli`
- API token in `~/.todoist.cfg`
Method 3: Use the Skill Creator
OpenClaw includes a skill-creator skill:
"Use the skill-creator to make a Notion integration"
This provides a structured wizard for building skills.
Advanced Skill Features
Environment Variables
Skills can define required environment variables:
---
name: openai-whisper
env:
- WHISPER_MODEL: base
---
Tool Scripts
Add executable scripts in the tools/ directory:
#!/bin/bash
# tools/transcribe.sh
whisper "$1" --model $WHISPER_MODEL
Dependencies
Specify required CLI tools:
## Requirements
- ffmpeg (video processing)
- whisper (speech-to-text)
Community Skills
ClawHub Marketplace
Browse and install skills from clawhub.com:
clawdhub search "crypto"
clawdhub install polymarket-skill
Popular Community Skills
- DeFi Operations — Swap tokens, check balances
- Polymarket — Prediction market trading
- MCP Servers — Model Context Protocol integrations
- Browser Automation — Web scraping and testing
- Home Assistant — Smart home control
Best Practices
1. Clear Descriptions
Write descriptions that help OpenClaw know when to use your skill:
✅ Good: "Send messages via WhatsApp to any contact" ❌ Bad: "WhatsApp integration"
2. Comprehensive Instructions
Include examples of:
- What the user might ask
- How to respond
- Edge cases and errors
3. Secure Credentials
Never hardcode API keys. Use:
- Environment variables
- 1Password integration
- Config files (gitignored)
4. Error Handling
Tell OpenClaw how to handle failures:
## Error Handling
If the API returns 401, ask the user to refresh their token.
If the API is down, inform the user and try again later.
Security Considerations
Skills can execute arbitrary code, so:
- Review before installing — Check what commands run
- Use trusted sources — Prefer official and verified skills
- Limit permissions — Don't give skills more access than needed
- Monitor activity — Check logs for unexpected behavior
Recent security research found malicious skills in the wild. Always verify skill sources.
Troubleshooting
Skill Not Loading
- Check the skill is in the correct directory
- Verify SKILL.md has valid frontmatter
- Look for syntax errors:
openclaw doctor
Skill Not Activating
- Make the description more specific
- Add trigger phrases in the instructions
- Ask directly: "Use the X skill to..."
Commands Failing
- Check required tools are installed
- Verify environment variables are set
- Test commands manually in terminal
Skills are what make OpenClaw truly powerful. Start with built-in skills, explore community creations, then build your own to automate your unique workflows.
Ready to try it yourself?
Get the free guide and set up your AI assistant in 30 minutes.
📚 Explore More
Tools, Skills & MCP — Extending OpenClaw
Complete guide to extending OpenClaw with custom skills, configuring search providers, troubleshooting tool errors, and understanding the skill loading system.
How to Build Your Own Personal AI Assistant
Architecture guide for creating a customized AI assistant. Configure personality, add integrations, build custom skills, and create an AI that truly works for you.
Raycast
Supercharge Raycast with AI assistant capabilities. Quick actions, clipboard history, snippets, and window management — all enhanced with conversational AI.
Running OpenClaw in Docker — Setup & Troubleshooting
Complete guide to running OpenClaw in Docker containers, including setup on Synology NAS, fixing permission errors, resolving pairing/1008 errors, and configuring volumes for persistent skills.