OpenClaw
What Is OpenClaw
OpenClaw is a local-first personal AI assistant that runs on your own devices. It was started by Austrian developer Peter Steinberger (the founder of PSPDFKit). Originally named Clawdbot, it was renamed to OpenClaw in early 2026 and quickly became one of the most-watched open-source AI agent projects on GitHub.
The core idea behind OpenClaw is simple: run a long-lived “gateway” daemon on your own machine so the assistant can talk to you through your everyday messaging channels, call tools, and run automated tasks.
Key features include:
- Multi-channel messaging — Connect WhatsApp, Telegram, Slack, Discord, and other platforms and chat with the assistant directly inside the apps you already use
- Voice and Canvas — Voice interaction on macOS/iOS/Android, plus a visual “live Canvas”
- Tool execution — Built-in tools for browser, session management, and Canvas, executed in sandboxed environments
- Automation — Trigger workflows via cron jobs and webhooks
- Custom skills — Extend the assistant’s capabilities with custom skills
OpenClaw is a community-driven open-source project licensed under MIT. The project home and source code live at github.com/openclaw/openclaw .
System Requirements
OpenClaw runs on Node.js. Before installing, make sure you have:
| Item | Requirement |
|---|---|
| Runtime | Node.js 24 (recommended) or Node.js 22.19+ |
| OS | Ubuntu 22.04+, Ubuntu 26.04 LTS recommended |
| Network | Required to connect your chosen messaging channels and AI model APIs |
If Node.js is not installed yet, see the Node.js Environment section to install it via nvm:
# Install Node.js 24 with nvm
nvm install 24
nvm use 24
node --versionInstallation
Install OpenClaw globally via npm:
npm install -g openclaw@latest
# Verify the installation
openclaw --versionOnboard and Install the Daemon
After the first install, run the guided onboarding flow and install the long-running daemon:
openclaw onboard --install-daemonThe onboarding flow walks you through connecting messaging channels and configuring models.
Running the Gateway
The heart of OpenClaw is a long-lived gateway daemon. Once the daemon is installed, you can check its status:
# Check the gateway daemon status
openclaw gateway statusOr run it in the foreground for debugging:
openclaw gateway --port 18789 --verboseBasic Usage
Once installed and connected to a channel, you can chat with the assistant directly inside the connected app. Common command-line operations:
# Send a message to the assistant
openclaw agent --message "Summarize my schedule for today"
# Send a message through a channel
openclaw message send
# Manage DM pairing approvals
openclaw pairing approveIn a conversation, OpenClaw also supports a set of chat commands, for example:
/status— show the current status/new— start a new session/reset— reset the current session
OpenClaw runs on your own devices, and your messages and data stay local by default. It still calls the AI model APIs you configure (such as cloud LLMs), so review the relevant privacy and data policies.
Related Links
- Source code and docs: github.com/openclaw/openclaw
If you want AI coding assistants in your terminal, see AI Coding Assistants (CLI); to run local LLMs that tools like OpenClaw can call, see Local LLMs.