OpenClaw is an AI agent platform that gives your AI assistant real-world capabilities. Unlike chatbots that just respond to messages, OpenClaw agents can control apps, manage services, automate workflows, and interact with hardware — running as a persistent background service.

OpenClaw is open source and free to self-host. You'll need an AI provider API key (like Claude or OpenAI) for the AI backend, which has its own pricing.

What can OpenClaw Skills do?

OpenClaw Skills cover developer tools (coding, GitHub), media (image generation, TTS, video), messaging (Discord, Slack, WhatsApp), productivity (Notion, Obsidian, Trello), smart home (lights, music), and more — 55 skills and growing.

Tutorial February 22, 2026 · MCP Hub Team · 14 min read

OpenClaw Skills Setup: Installation to Automation

OpenClaw turns your AI into a true autonomous agent. This guide walks you through installation, skill setup, and building automated workflows that run 24/7.

Why OpenClaw?

Most AI tools are reactive — you ask, they answer. OpenClaw is different. It creates a persistent AI agent that:

Runs in the background 24/7
Controls real-world apps and services
Responds to messages across platforms (Discord, Slack, Telegram, WhatsApp)
Automates workflows proactively
Maintains memory and context between conversations

With 55 OpenClaw Skills in our directory, the possibilities are extensive.

Prerequisites

Node.js 18+ — Download here
An AI API key — Claude (Anthropic) or OpenAI
A computer that stays on — OpenClaw runs as a background service
Basic terminal comfort — You'll run a few commands

Step 1: Install OpenClaw

# Install OpenClaw globally
npm install -g openclaw

# Initialize a new agent
openclaw init my-agent
cd my-agent

# Configure your AI provider
openclaw config set provider anthropic
openclaw config set apiKey YOUR_API_KEY

Step 2: Start Your Agent

# Start the agent
openclaw start

# Check status
openclaw status

Your agent is now running! But it's a blank slate — time to add skills.

Step 3: Install Your First Skills

Skills are installed with a single command. Here are the essentials to start with:

Essential Developer Skills

# Coding agent — autonomous code writing
openclaw skill install coding-agent

# GitHub integration
openclaw skill install github-cli

# Terminal management
openclaw skill install tmux-manager

Essential Productivity Skills

# Note-taking with Obsidian
openclaw skill install obsidian

# Task management
openclaw skill install trello

# Email management
openclaw skill install himalaya-email

Essential Communication Skills

# Discord messaging
openclaw skill install discord

# Slack integration
openclaw skill install slack

Popular Skills Walkthrough

‍ Coding Agent

The Coding Agent is OpenClaw's most powerful skill. It can autonomously write, edit, and debug code across your projects.

What it does:

Reads and writes files in your codebase
Runs shell commands to test changes
Creates new files and directories
Debugs errors by reading logs and stack traces

Example: "Add a dark mode toggle to the settings page" → The agent reads your current code, creates the toggle component, updates styles, and tests it.

Spotify Player

The Spotify Player skill gives your agent control over Spotify playback.

Example: "Play some focus music" → Agent starts a lo-fi playlist. "Skip this track" → Done.

Nano Banana Pro

The Nano Banana Pro skill generates images using Google's Gemini image model.

Example: "Generate a logo concept for my coffee shop" → Agent creates and displays the image.

ElevenLabs Voices

The ElevenLabs Voices skill adds text-to-speech with natural-sounding voices.

Example: "Read me the summary of today's news" → Agent speaks the content aloud.

Discord

The Discord skill lets your agent participate in Discord servers — reading and sending messages, managing channels, and responding to mentions.

Building Automated Workflows

The real power of OpenClaw comes from combining skills into workflows. Here are practical examples:

Morning Briefing Workflow

Skills used: Weather, Himalaya Email, ElevenLabs Voices

Every morning, your agent checks the weather, scans your email for urgent items, and reads you a summary using natural text-to-speech. All automated — no prompting required.

Code Review & PR Workflow

Skills used: GitHub CLI, GitHub PR Manager, Coding Agent

When a new PR is opened, your agent reviews the code changes, checks for common issues, and posts review comments. You can configure thresholds for auto-approval or flagging for your review.

Content Creation Pipeline

Skills used: Nano Banana Pro, YouTube API CLI, Canvas

Generate images for blog posts, create video thumbnails, and publish content — all orchestrated by your agent.

Smart Home Integration

Skills used: OpenHue, Weather, Spotify Player

"Set the mood for a dinner party" → Lights dim to warm, dinner jazz plays on Spotify, and the agent adjusts if the playlist gets too upbeat.

Advanced Configuration

Skill Permissions

Control what each skill can access:

# List skill permissions
openclaw skill permissions coding-agent

# Restrict file access to specific directories 
openclaw skill config coding-agent --allowed-paths ~/projects

Heartbeats and Proactive Actions

Configure your agent to check on things periodically:

# Set heartbeat interval (in minutes)
openclaw config set heartbeatInterval 30

# Enable proactive notifications
openclaw config set proactiveNotifications true

Multi-Channel Communication

Connect your agent to multiple platforms simultaneously:

# Add Telegram as a channel
openclaw channel add telegram --token YOUR_BOT_TOKEN

# Add Discord
openclaw channel add discord --token YOUR_BOT_TOKEN

Troubleshooting

Agent won't start

Check your API key: openclaw config get apiKey. Verify Node.js version: node --version (need 18+).

Skill not responding

Reinstall the skill: openclaw skill reinstall skill-name. Check logs: openclaw logs --tail 50.

High API costs

Adjust the model for routine tasks: openclaw config set defaultModel claude-haiku. Use smarter models only for complex tasks.