Open source · Local-first · Multi-channel

OpenClaw: Your personal AI assistant, running on your devices

More than chat. Stay always-on inside the channels you already use, powered by a local Gateway and a tool system to take actions. Your data, your keys, your rules.

Read the install guide Official GitHub

Gateway

Local control plane

Channels

WhatsApp/Telegram/Slack…

Runtime

Node 22+

Nov 2025: Clawd begins (a weekend hack)

In the official blog, the author describes it as a weekend prototype that started with a simple goal: let an AI reply inside WhatsApp. “Clawd” was a playful pun (Claude + claw).

Late 2025: Moltbot (molting = growth)

The community then moved to “Moltbot”, where molting symbolizes growth. During this phase, the project spread rapidly and evolved from a chat relay into a broader agent platform.

Jan 29, 2026: Rebranded to OpenClaw (Open + Claw)

The official announcement introduced the final name: OpenClaw—Open for open source and community; Claw for the lobster heritage. They also completed trademark checks, domain purchases, and migration code.

2026: The Gateway becomes the core

In the GitHub README, the Gateway is defined as the control plane: sessions, channels, tools, events, and the Web UI. Think of it as the local hub for your agent.

Ecosystem: channels, nodes, skills

Over time, it added the pieces that connect to the real world: multi-channel messaging, iOS/Android nodes, browser tooling, cron automation, and skills.

Why it took off

The blog notes the growth from a weekend hack to 100k+ stars and millions of visitors. A key reason: it is not SaaS—it’s your assistant, on your machine.

Note

OpenClaw is a personal AI assistant platform you run on your own machine. The Gateway is the control plane (sessions, channels, tools, UI), and the magic is staying always-on inside the chat apps you already use.

⚡

Multi-channel messaging (meet you where you are)

The README lists many channels: WhatsApp, Telegram, Slack, Discord, Google Chat, Signal, iMessage/BlueBubbles, IRC, Teams, Matrix, Feishu, LINE, Mattermost, Nextcloud Talk, Twitch, WebChat, and more.

🧭

Gateway: the control plane

The Gateway centralizes sessions, channels, tools, events, and the web UI. You can chat in the Control UI or operate via the CLI.

🧠

Models + auth + failover

Docs emphasize multi-provider setups with OAuth/API keys and failover—critical for always-on assistants.

🧰

Tools: browser / canvas / nodes / cron

OpenClaw can act via tools: browser control (managed Chrome/Chromium), Canvas workspace, device nodes (camera/screen/notifications), cron automation and webhooks.

🧩

Skills: modular capabilities

Skills package tools + prompts + workflows. Use bundled/managed/workspace skills to adapt the assistant to your own workflows.

🛡️

Security-by-default: treat inbound as untrusted

For Telegram/WhatsApp/Signal/iMessage/Discord/Slack etc., the default is DM pairing (code + allowlist). Start single-user, then open up cautiously.

🛰️

Remote setups: VPS / homelab friendly

The README notes the Gateway can run on Linux and be exposed via Tailscale Serve/Funnel or SSH tunnels (with proper security).

📦

Multiple install modes: script / npm / Docker / Nix

Docs cover the recommended script install, npm/pnpm, Docker/Podman, Nix, Ansible and more—local or cloud.

System requirements (official)

Node 22+(The installation script will detect and install it if it’s missing)
macOS / Linux / Windows (Windows: WSL2 recommended)
pnpm only if you build from source

Recommended install (script)

macOS / Linux / WSL2

curl -fsSL https://openclaw.ai/install.sh | bash

Windows (PowerShell)

iwr -useb https://openclaw.ai/install.ps1 | iex

Windows tip

OpenClaw strongly recommends running on Windows via WSL2.

Alternative install (if Node is ready)

npm

npm install -g openclaw@latest
openclaw onboard --install-daemon

pnpm

pnpm add -g openclaw@latest
pnpm approve-builds -g
openclaw onboard --install-daemon

From source (developers)

git clone https://github.com/openclaw/openclaw.git
cd openclaw
pnpm install
pnpm ui:build
pnpm build
openclaw onboard --install-daemon

Note: pnpm may require explicit approval for packages with build scripts (e.g., sharp).

Linux VPS + Docker quick deploy (recommended Compose script)

When to use this

If you want a containerized Gateway on a VPS with config/workspace persisted on the host, the official docs provide a docker-compose based setup script.

1) Prepare the environment (example using UBUNTU/DEBIAN)

sudo apt-get update
sudo apt-get install -y ca-certificates curl git

# Install Docker Engine (It is recommended to use the official installation method; skip this step if already installed.)# https://docs.docker.com/engine/install/

# Install Docker Compose v2 (Most new versions already come with the docker compose subcommand)docker compose version

2) Clone the repository and run the official script

git clone https://github.com/openclaw/openclaw.git
cd openclaw

# Official recommendation: Run from the repository root directory./docker-setup.sh

3) Common commands after startup (execute in the repository root directory)

# Check gateway container statusdocker compose ps

# Open (or print) the Dashboard URL (browser does not open automatically)docker compose run --rm openclaw-cli dashboard --no-open

# Health check (authentication not required)curl -fsS http://127.0.0.1:18789/healthz
curl -fsS http://127.0.0.1:18789/readyz

Port exposure & security notes (VPS)

The dashboard usually runs on port 18789. If you expose it remotely, read the official network exposure hardening guidance and restrict access via firewall/allowlists. Do NOT expose an un-hardened control plane to the public internet.

docs.openclaw.ai/gateway/security#04-network-exposure-bind--port--firewall

4) Optional: Use a pre-built image (skip local BUILD)

# Use the official GHCR mirror (do not use unofficial mirrors with the same name)export OPENCLAW_IMAGE="ghcr.io/openclaw/openclaw:latest"
./docker-setup.sh

These steps follow the official Docker docs: docker-setup.sh builds/pulls the image, runs onboarding, starts the compose stack, and generates a Control UI token (written into .env). docs.openclaw.ai/install/docker

1) Run the onboarding wizard

The recommended path is the wizard: it configures auth, gateway, optional channels, and can install a background daemon.

openclaw onboard --install-daemon

2) Verify everything works

openclaw doctor
openclaw status
openclaw dashboard

dashboard opens the browser UI (or visit http://127.0.0.1:18789/ on the gateway host).

3) Run in foreground (for debugging)

openclaw gateway --port 18789 --verbose

You can also send messages via CLI to configured channels.

4) DM safety & pairing (important)

OpenClaw connects to real messaging surfaces. By default it pairs unknown DMs with a code; only approved senders can talk. Avoid overly-open allowlists.

Official security guide: docs.openclaw.ai/gateway/security

Config paths & env vars (official)