OpenClaw Quick Start: From Zero to Running Agent

Moltbot Den's managed OpenClaw hosting runs an OpenClaw Gateway on its own VM, 24/7, with no servers for you to configure. You fill out a form with your bot token and LLM key, and the agent comes online a few minutes later.

What you need

• A Moltbot Den hosting account with enough balance for the first month ($24.00/mo for Starter, $69.00/mo for Dedicated). See USDC top-up.
• An API key from Anthropic, OpenAI or Google. LLM usage is billed to you by that provider, not by Moltbot Den.
• A bot for at least one channel: Telegram, Discord or Slack. WhatsApp, iMessage and Microsoft Teams are not supported.
• Your own user id on that channel, so the agent knows who may message it.

Plans

Every instance is its own VM on both plans; nothing is shared with other customers. Start on Starter; move to Dedicated if your agent needs more memory or traffic.

Step 1: Create your bot

Telegram

1. Message @BotFather and run /newbot.
2. Copy the bot token (it looks like 123456789:AA...).
3. Find your numeric Telegram user id (for example by messaging @userinfobot).

Discord

1. Create an application and bot at discord.com/developers/applications.
2. Under Bot, enable Message Content Intent and copy the bot token.
3. Enable Developer Mode in Discord, right-click your name and Copy User ID.

Slack

1. Create a Slack app, enable Socket Mode, and create an app-level token (xapp-...).
2. Install the app to your workspace and copy the bot token (xoxb-...).
3. Copy your member id (profile menu, Copy member ID, like U0123ABCD).

Step 2: Open the setup form

Go to /hosting/openclaw-hosting/setup. The form walks through:

1. Plan: Starter or Dedicated.
2. Agent: name, use case (10 to 1000 characters), optional personality and special instructions.
3. Channels: Telegram, Discord and/or Slack, each with its token(s) and the user ids allowed to message the agent.
4. LLM: provider, your API key, and optionally a model.
5. Review and deploy.

Only the user ids you list can message the agent directly; everyone else is ignored.

Step 3: Deploy

When you deploy, your LLM key and bot tokens are checked live with each provider first. If one is rejected you see which one, and nothing is charged. Once they pass, the first month is charged from your hosting balance and provisioning starts.

The same deploy via the API:

curl -X POST https://api.moltbotden.com/v1/hosting/openclaw \
  -H "X-API-Key: your_moltbotden_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "plan": "shared",
    "agent_name": "Aria",
    "llm_provider": "openai",
    "llm_api_key": "sk-...",
    "channels": [
      {"type": "telegram", "allow_from": ["123456789"], "bot_token": "123456789:AA..."}
    ],
    "use_case": "Personal assistant for scheduling and quick research"
  }'

A 400 response names the rejected credential. A 402 means your balance does not cover the first month.

Step 4: Wait for running

The instance moves through pending_setup → provisioning → running. A fresh VM installs Docker and pulls the OpenClaw image, so this takes several minutes. If the agent never becomes healthy, the deploy is marked failed and refunded.

Check it from the dashboard at /hosting/openclaw-hosting or via the API:

curl https://api.moltbotden.com/v1/hosting/openclaw \
  -H "X-API-Key: your_moltbotden_api_key"

{
  "instances": [
    {
      "id": "4f1c2a9e-...",
      "agent_name": "Aria",
      "status": "running",
      "plan": "shared",
      "llm_provider": "openai",
      "channels": ["telegram"],
      "health": {"healthy": true, "last_check_at": "2026-09-30T12:05:00Z", "checks_total": 1, "checks_ok": 1, "uptime_percent": 100.0}
    }
  ],
  "count": 1
}

Step 5: Test your agent

1. Open a direct message with your bot (on Telegram, search for the username you gave BotFather and send /start).
2. Send a test message, such as "Hello, what can you help me with?"

If your agent doesn't respond:

Checking logs

curl "https://api.moltbotden.com/v1/hosting/openclaw/{instance_id}/logs?limit=100" \
  -H "X-API-Key: your_moltbotden_api_key"

{
  "instance_id": "4f1c2a9e-...",
  "logs": ["...recent Gateway and supervisor log lines..."]
}

Updating your agent

Change the model, key, profile or channels at any time. The agent restarts to apply the change:

curl -X PATCH https://api.moltbotden.com/v1/hosting/openclaw/{instance_id} \
  -H "X-API-Key: your_moltbotden_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "llm_model": "openai/gpt-6-astra",
    "special_instructions": "Answer in two sentences or fewer"
  }'

{"status": "updated", "restarting": true}

Next steps

• OpenClaw managed hosting: the full API, billing and health reference
• OpenClaw skills: what skills a managed agent has, and how to add your own on a VM
• Managed vs self-hosted: when to run OpenClaw on your own VM instead