How to Self-Host OpenClaw and Access It Remotely

What Is OpenClaw?

OpenClaw (originally released as Clawdbot in November 2025, briefly renamed Moltbot, then finally OpenClaw) is a free, open-source, self-hosted AI agent built by Peter Steinberger, founder of PSPDFKit. It became one of the fastest-growing GitHub repositories ever, reaching over 140,000 stars and 20,000 forks by early February 2026.

The key difference between OpenClaw and tools like ChatGPT or Claude.ai is that OpenClaw does not just answer questions. It executes tasks. It can browse the web, read and write files on your machine, run shell commands, manage your calendar, clear your inbox, control your browser, and proactively reach out to you on WhatsApp or Telegram without being asked. It remembers everything across all conversations indefinitely, stored locally on your machine.

The project is model-agnostic. You bring your own API key for Claude, GPT-4, Gemini, DeepSeek, or route inference through a local Ollama instance. All memory, history, and configuration stay on your hardware. Nothing is sent to third-party servers beyond the LLM API calls you authorize.

Why Do You Need a Public URL for OpenClaw?

OpenClaw's Gateway runs on http://localhost:18789 by default. This means it is only reachable from the machine it runs on. There are several real scenarios where you need it to be accessible from the outside world:

How Localtonet Works with OpenClaw

Localtonet creates an encrypted outbound connection from your machine to Localtonet's servers. Any request to your public HTTPS URL is forwarded through that tunnel to the OpenClaw Gateway running on localhost:18789. No inbound ports need to be open. No static IP needed. Works behind NAT, CGNAT, and corporate firewalls.

Prerequisites

Step 1: Install OpenClaw

npm install -g openclaw@latest

openclaw --version

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

docker pull ghcr.io/openclaw/openclaw:latest
docker run -it -p 18789:18789 \
  -v $HOME/.openclaw:/home/node/.openclaw \
  ghcr.io/openclaw/openclaw:latest

Step 2: Run the Onboarding Wizard

OpenClaw includes an interactive wizard that configures your LLM provider, messaging channels, and system service. Running it once handles everything you would otherwise need to configure manually.

openclaw onboard --install-daemon

openclaw doctor

Step 3: Start the Gateway

openclaw gateway --port 18789

# macOS:
launchctl list | grep openclaw

# Linux:
systemctl --user status openclaw

Step 4: Create the Localtonet Tunnel

# macOS (Homebrew):
brew tap localtonet/tap && brew install localtonet

# Linux (install script):
curl -fsSL https://localtonet.com/install.sh | sh

# Windows: download from localtonet.com/download or install via Microsoft Store

localtonet --authtoken YOUR_TOKEN_HERE

sudo localtonet --install-service --authtoken YOUR_TOKEN_HERE
sudo localtonet --start-service --authtoken YOUR_TOKEN_HERE

Channel Setup: Telegram (Easiest)

# Via CLI (recommended):
openclaw config set channels.telegram.enabled true
openclaw config set channels.telegram.botToken "YOUR_BOT_TOKEN_HERE"
openclaw config set channels.telegram.dmPolicy "pairing"
openclaw config set channels.telegram.allowFrom '["YOUR_NUMERIC_USER_ID"]'

{
  "channels": {
    "telegram": {
      "enabled": true,
      "botToken": "YOUR_BOT_TOKEN_HERE",
      "dmPolicy": "pairing",
      "allowFrom": ["YOUR_NUMERIC_USER_ID"]
    }
  }
}

# If running as daemon (macOS):
launchctl stop openclaw && launchctl start openclaw

# If running as daemon (Linux):
systemctl --user restart openclaw

# If running manually:
openclaw gateway --port 18789

openclaw pairing approve telegram <CODE>

Channel Setup: Discord

{
  "channels": {
    "discord": {
      "enabled": true,
      "token": "YOUR_DISCORD_BOT_TOKEN",
      "dm": {
        "enabled": true,
        "policy": "pairing"
      },
      "guilds": {
        "YOUR_GUILD_ID": {
          "users": ["YOUR_DISCORD_USER_ID"],
          "requireMention": true,
          "channels": {
            "YOUR_CHANNEL_ID": {
              "allow": true,
              "requireMention": true
            }
          }
        }
      }
    }
  }
}

openclaw pairing list discord
openclaw pairing approve discord <CODE>

openclaw doctor
openclaw channels status --probe

Other Channel Integrations

Advanced: Multi-Agent Configuration

OpenClaw supports routing different channels or accounts to isolated agents, each with their own workspace, session history, and system prompt. This is useful for keeping work and personal contexts completely separate, or for running multiple assistant personas.

{
  "agents": {
    "list": [
      { "id": "home", "default": true, "workspace": "~/.openclaw/workspace-home" },
      { "id": "work", "workspace": "~/.openclaw/workspace-work" }
    ]
  },
  "bindings": [
    { "agentId": "home", "match": { "channel": "telegram", "accountId": "personal" } },
    { "agentId": "work", "match": { "channel": "telegram", "accountId": "biz" } }
  ]
}

With this configuration, messages from your personal Telegram account go to the home agent with its own memory, and messages from a business Telegram account go to the work agent. The two agents never share context.

Security Best Practices

Troubleshooting

What You Can Do with a Running OpenClaw

Once your self-hosted OpenClaw is running and connected to a messaging channel, you have an agent that operates in natural language. Here are practical examples organized by category.

Frequently Asked Questions