Skills for OpenClaw Agents
Skills are like cheat sheets for your AI agent. They teach it specific tasks so it can handle stuff without you re-explaining every time.
How Skills Work
Each skill is a folder with a SKILL.md file at the root. This file tells the agent what to do in clear steps.
~/.openclaw/workspace/skills/
└── my-skill/
├── SKILL.md
└── scripts/ (optional)
A Sample Skill
# SKILL.md — Email Outreach
## Description
Send professional outreach emails via SMTP. Use when: user asks to send
emails to leads, prospects, or contacts.
## Prerequisites
- `SMTP_HOST`, `SMTP_PORT`, `SMTP_USER`, `SMTP_PASS` in env
- `himalaya` CLI installed (`brew install himalaya`)
## Steps
1. Load recipient context from $LEADS_PATH.
2. Generate personalized message using template.
3. Send via himalaya.
4. Log the send to leads/sent-log.csv.
Agent follows these steps exactly, no guessing.
Skill Lifecycle
- Install — copy skills to
~/.openclaw/workspace/skills/ - Active — agent picks them up next session
- Update — pull new version from ClawHub
- Disable — rename dir with
_prefix - Publish — package and push to ClawHub
Installing Skills
Use the clawhub CLI:
# Install clawhub CLI
npm install -g clawhub
# Search for skills
clawhub search email
# Install a skill
clawhub install himalaya
# Install specific version
clawhub install himalaya@2.1.0
# List installed skills
clawhub list
# Update all skills to latest
clawhub update
# Update specific skill
clawhub update gh-issues
Skills install automatically on next message.
Writing a Skill
Basic structure:
# SKILL.md — Camera Snapshot
## Description
Capture frames or clips from RTSP/ONVIF cameras. Use when user asks for
camera snapshot, doorbell image, or security feed capture.
## Prerequisites
- `ffmpeg` installed
- Camera URLs in TOOLS.md under "Cameras" section
## Steps
1. Parse camera URLs.
2. Capture frame with ffmpeg.
3. Return the image path.
Agent will follow these steps exactly.
Skill Best Practices
- Be specific in descriptions.
- List prerequisites upfront.
- Use absolute paths.
- Put ref docs in
references/.
Skills vs. Hardcoded Tools
Skills are instructions; tools are code. They work together:
| Skills | Tools | |
|---|---|---|
| What they are | Instructions (Markdown) | Executable functions |
| How they're added | Copy a dir | Register in tool config |
| Who writes them | Anyone | Developers |
| When to use | Workflows with human-like decisions | Deterministic ops |
| Example | "Send an outreach email" | web_search(query) |
Publishing Your Skill
To share your skill:
# Initialize a clawhub package
cd ~/.openclaw/workspace/skills/my-skill
clawhub init
# Publish
clawhub publish
# Update published skill
clawhub publish --bump patch
Published skills are versioned, searchable, and installable.
Ecosystem Effect
Skills make your agent smarter. A workspace with 10-15 well-written skills means the agent handles tasks without you explaining context.
For a media-focused fork, see Creative Claw's OpenClaw extension and the source-based overview of its local creative AI workflow. Creative Claw's capabilities and setup requirements are documented in its own repository.
