Awesome OpenClaw Skills Tutorial: Install 5,400+ Tools for Your Local AI Assistant

Prerequisites
Before diving in, ensure you have:
- Node.js version 22 or higher (run
node --versionto check). - A terminal with
npmorpnpminstalled. - An LLM API key (Anthropic Claude, OpenAI, or compatible model) – store it securely in your environment.
- Basic command-line knowledge (no prior OpenClaw experience required).
- macOS, Linux, or Windows with WSL2 recommended for best performance.
OpenClaw runs locally on your machine, so no cloud costs beyond your LLM usage.
Step 1: Install OpenClaw
OpenClaw is the self-hosted gateway that powers your AI assistant across chat apps like WhatsApp, Telegram, Slack, and Discord.
Run these commands:
npm install -g openclaw@latest
openclaw onboard --install-daemon
The wizard will guide you through:
- Setting your workspace folder.
- Connecting messaging channels (e.g., Telegram bot token).
- Choosing your LLM model.
After onboarding, start the gateway:
openclaw gateway --port 18789 --verbose
Expected output: Gateway running on http://localhost:18789 and a daemon service (launchd on macOS, systemd on Linux) keeps it alive 24/7.
Step 2: Discover Skills in the Awesome OpenClaw Collection
Skills are self-contained folders with a SKILL.md file that teaches the agent new tools – from GitHub integration to calendar management.
Visit the curated collection (filtered from the official registry, removing spam and low-quality entries):
Browse categories like:
- Coding Agents & IDEs (1,200+ skills)
- Git & GitHub (170+ skills)
- Browser & Automation (335+ skills)
- DevOps & Cloud (400+ skills)
- Productivity & Tasks (200+ skills)
Search for skills by name or use case (e.g., github, linear, google-calendar). Each listing links to the exact skill repo for review.
Pro tip: Paste any skill's GitHub URL directly into your OpenClaw chat – the agent can auto-install it.
Step 3: Install Skills with ClawHub CLI
ClawHub is the official skill registry CLI (auto-available after OpenClaw setup).
Install a skill into your workspace:
clawhub install github
This downloads the skill to ./skills (workspace) or ~/.openclaw/skills (global shared).
Install multiple high-impact ones:
clawhub install linear
clawhub install agentmail
clawhub install google-calendar
clawhub install browser-automation
Update all skills:
clawhub update --all
OpenClaw auto-detects new skills on the next agent turn. Workspace skills override global ones.
Step 4: Configure and Enable Skills
Edit your config file for API keys and toggles:
nano ~/.openclaw/openclaw.json
Example snippet:
{
"skills": {
"entries": {
"github": {
"enabled": true,
"apiKey": { "source": "env", "id": "GITHUB_TOKEN" }
},
"google-calendar": {
"env": { "GOOGLE_API_KEY": "your-key-here" }
}
}
}
}
Restart the gateway (openclaw gateway restart) for changes to apply.
Skills with gating (e.g., requires git binary) auto-enable only if prerequisites exist.
Step 5: Test Your New Skills
Send a test message to your assistant:
openclaw agent --message "Create a new GitHub issue in my repo about adding dark mode" --thinking high
Or chat via WhatsApp/Telegram: "Use Linear skill to update my sprint tasks."
Expected behavior: The agent invokes the skill, executes the tool (e.g., creates PR, sends email), and replies with results plus proof (screenshots or logs).
Verify loaded skills:
openclaw skills list
Common Issues & Troubleshooting
- Skill not loading: Check
~/.openclaw/openclaw.jsonforenabled: false. Runopenclaw skills listto see eligibility. Ensure binaries (e.g.,git) are in$PATH. - Permission errors: On macOS, grant Full Disk Access and Screen Recording in System Settings → Privacy & Security.
- API key missing: Skills fail silently – add via config or
export ENV_VAR=valuebefore running. - Too many skills slow prompt: Limit to 7-10 active skills per agent (token overhead ~24 per skill). Use
skills.entries.<name>.enabledto disable. - ClawHub command not found: Re-run
openclaw onboardor install vianpm install -g clawhub-cli. - Daemon not running: Check status with
openclaw daemon statusand restart.
For security, always review SKILL.md before install and use VirusTotal scan via ClawHub.
Next Steps
- Create your own skill: Copy any
SKILL.md, add custom YAML frontmatter and tools, then drop into./skills. - Contribute: Submit vetted skills to the awesome collection via PR.
- Scale up: Connect multiple agents (per-workspace skills) or add macOS nodes for GUI automation.
- Advanced: Enable skills watcher in config for live editing and explore Agent-to-Agent protocols for multi-agent teams.
You're now ready to turn OpenClaw into a true productivity powerhouse. Start with 3-5 skills today and watch your automation grow.