← Back to Knowledge Base

Getting Started with OpenClaw

A beginner's guide to setting up your first OpenClaw agent, understanding its core concepts, and making it work for you.

OpenClaw is an open-source platform for running persistent AI agents that can help you with daily tasks, automate workflows, and serve as a personal assistant. This guide will walk you through the basics of getting started.

What is OpenClaw?

OpenClaw is infrastructure for AI agents. Unlike chatbot interfaces that forget everything after each session, OpenClaw agents maintain persistent memory, can run scheduled tasks, access tools and APIs, and evolve over time based on your interactions.

Think of it as a framework that lets you deploy a personalized AI assistant that actually remembers who you are and what you need.

Core Concepts

  • Agents: Persistent AI assistants that can be configured with different models, skills, and behaviors.
  • Skills: Reusable modules that give your agent new capabilities (like reading emails, checking calendars, controlling smart devices).
  • Memory: Your agent stores context, preferences, and learnings in plain markdown files you can read and edit.
  • Tools: Built-in functions that let agents interact with the world (read/write files, execute commands, search the web).

Installation Options

Option 1: Managed Hosting (Molten.bot)

The fastest way to get started is with Molten.bot, a managed OpenClaw service. You get:

  • Zero setup – sign up and have an agent running in under a minute
  • Automatic updates and maintenance
  • Isolated, secure containers
  • Choice of 400+ AI models via OpenRouter

Option 2: Self‑Hosted

For full control, you can deploy OpenClaw on your own infrastructure:

# Using Docker (recommended)
docker run -d \
  --name openclaw \
  -v ./workspace:/workspace \
  -p 8080:8080 \
  moltenbot/openclaw:latest

Or install via npm:

npm install -g openclaw
openclaw gateway start

Your First Agent

Once installed, you'll need to:

  1. Choose a model: Select from OpenAI, Anthropic, Google, or OpenRouter models.
  2. Configure skills: Add capabilities like email reading, calendar management, or custom workflows.
  3. Set up memory: Define where your agent stores its learnings (local files, database, or cloud storage).
  4. Connect channels: Link your agent to communication platforms (WhatsApp, Telegram, Signal, Discord, etc.).

Common First Tasks

Here are some practical things to try with your new agent:

  • Daily check-ins: Ask your agent to summarize your emails, calendar events, and notifications each morning.
  • Project assistance: Have it help organize files, draft documentation, or track todo lists.
  • Automated monitoring: Set up periodic checks for website uptime, security alerts, or social mentions.
  • Learning companion: Use it to explain technical concepts, summarize articles, or practice languages.

Next Steps

After you've got the basics down, explore:

  • Building custom skills: Create your own tools and workflows.
  • Integrating with APIs: Connect to your favorite services.
  • Multi‑agent systems: Deploy specialized agents that work together.
  • Community resources: Join the OpenClaw Discord and explore skills on ClawHub.

Need help? Check out the OpenClaw documentation or ask in the community forums. The OpenClaw ecosystem is growing fast, and there's always someone ready to help newcomers.