Skip to content

How to Install OpenClaw on a Hostinger VPS (2026 Step-by-Step Guide)

A tested guide to install OpenClaw on a Hostinger VPS: every SSH command, running it 24/7 as a daemon, connecting WhatsApp or Telegram, and securing it.

Sunny Kumar
Sunny Kumar7 min read
TL;DR

To install OpenClaw on a Hostinger VPS, buy a KVM VPS with Ubuntu, connect over SSH, install Node.js and build tools, then run npm install -g openclaw@latest. Use openclaw onboard to add your LLM key (Claude, ChatGPT, Gemini or a local Ollama model), install it as a daemon so it runs 24/7, and connect a channel like WhatsApp or Telegram. A 2 vCPU / 8GB plan is the comfortable choice; a 1 vCPU / 4GB plan works with a swap file.

I run OpenClaw on a Hostinger VPS so a self-hosted AI assistant answers me on WhatsApp and Telegram around the clock, on my own server, with my own keys. No third party in the middle.

Setting it up is not hard, but the guides I found were either vague or skipped the parts that actually break. So here is the full process, every command included, exactly as I did it.

By the end you will have OpenClaw running 24/7 on a VPS, connected to a messaging app, with skills installed and the box properly secured.

What is OpenClaw?

OpenClaw is an open-source, self-hosted AI personal assistant. You run it on your own machine, connect it to your messaging apps, and plug in whichever LLM you want to power it.

Diagram showing you chatting from WhatsApp, Telegram, Slack or Discord to the OpenClaw daemon running on your Hostinger VPS, which connects to your chosen LLM (Claude, ChatGPT, Gemini or a local Ollama model)
The whole stack is yours: you message it from any app, it runs on your VPS, and it thinks with the LLM you choose.

It is genuinely popular, with over 380,000 GitHub stars, and the appeal is ownership: your assistant, your data, your model choice. You can point it at Claude, ChatGPT or Gemini, or run a local model with Ollama so nothing leaves your server at all.

The openclaw/openclaw GitHub repository showing 381k stars, 79.9k forks, an MIT license, and the description: your own personal AI assistant, any OS, any platform, the lobster way
OpenClaw is MIT-licensed and open-source, so the whole thing is auditable. This is the live repo.

Because it is self-hosted, it needs somewhere to live that is always on. That is where the VPS comes in.

What do you need before starting?

Five things: a Hostinger VPS running Ubuntu, an LLM API key (or Ollama for a local model), a messaging account, basic terminal comfort, and about fifteen minutes. Have these ready and the install goes through in one sitting.

  • A Hostinger VPS with Ubuntu (22.04 or 24.04). Any KVM plan works; more on sizing below.
  • An LLM API key from Anthropic, OpenAI or Google, or Ollama installed for a local model.
  • A messaging account for the channel you want (WhatsApp, Telegram, Slack or Discord).
  • Basic terminal comfort — you will paste commands over SSH, nothing more.
  • About 15 minutes, most of it waiting on installs.

Why a Hostinger VPS for OpenClaw?

Because OpenClaw has to run as a background service that never sleeps, and shared hosting cannot do that. You need root access, SSH and the ability to run a long-lived process, which means a VPS.

The Hostinger VPS hosting page showing KVM plans with AMD EPYC processors, NVMe SSD storage, free weekly backups, an AI web terminal and a 30-day money-back guarantee
Hostinger's KVM VPS gives you full root access, NVMe storage and a 30-day money-back guarantee to test on.

I have run OpenClaw on DigitalOcean and Hetzner too, and Hostinger is the one I keep pointing people to for this: full root on a KVM VPS with NVMe storage, weekly backups included, and a 30-day money-back window so you can try it risk-free. The AMD EPYC cores handle the Node process comfortably. If you are weighing hosts more widely than this one job, I compared the main web hosting providers separately.

Which Hostinger VPS plan should you pick?

The 2 vCPU / 8GB plan (KVM 2) is the comfortable pick, especially if you plan to run a local model. The smallest plan works too, with one tweak.

PlanSpecsGood for
KVM 11 vCPU, 4GB RAM, 50GB NVMeA basic OpenClaw setup (add a swap file)
KVM 2 ⭐2 vCPU, 8GB RAM, 100GB NVMeThe recommended, comfortable choice
KVM 44 vCPU, 16GB RAM, 200GB NVMeRunning larger local Ollama models

Pricing shifts with promotions and term length (usually somewhere around $5-9 a month for KVM 1-2), so check the current rate before you buy.

Get a Hostinger VPS

How to install OpenClaw on Hostinger, step by step

Here is the whole install at a glance, then each step in full with the exact commands.

Seven-step install flow: buy a VPS, SSH in, install Node and dependencies, npm install openclaw, run onboarding, start as a daemon, then connect a messaging channel
Seven steps from a blank server to an assistant running around the clock.

Buy a Hostinger VPS and choose Ubuntu

Pick a KVM plan and, when prompted for an operating system, choose Ubuntu 24.04. Set a strong root password and note the server's IP address from your Hostinger panel.

Connect to your VPS over SSH

From your own terminal, connect as root using the IP from the panel:

bash
ssh root@YOUR_VPS_IP

Accept the fingerprint on first connect and enter your password.

Update the system and install dependencies

Bring the server up to date, then install Node.js 24, Git and the build tools OpenClaw needs:

bash
sudo apt update && sudo apt upgrade -y
curl -fsSL https://deb.nodesource.com/setup_24.x | sudo -E bash -
sudo apt install -y nodejs git build-essential
node -v
npm -v

The last two lines should print version numbers, confirming Node and npm are ready.

Install OpenClaw

Install it globally from npm, then check the version:

bash
npm install -g openclaw@latest
openclaw --version

Run the onboarding wizard

This is where you choose your LLM and paste your API key (or point it at a local Ollama model):

bash
openclaw onboard

Follow the prompts. It will ask for your provider, your key, and a few preferences, then write the config to ~/.openclaw/.

Start OpenClaw as a 24/7 daemon

So it survives reboots and keeps running after you disconnect, install it as a service and start the gateway:

bash
openclaw onboard --install-daemon
openclaw gateway start

Confirm it is healthy with the built-in check:

bash
openclaw doctor
openclaw gateway status

Connect your first messaging channel

Link the app you want to chat from:

bash
openclaw channels

Pick WhatsApp, Telegram, Slack or Discord and follow the prompt (usually scanning a QR code or pasting a bot token). Send it a message, and OpenClaw replies. You are live.

Tip

On the 1 vCPU / 4GB plan, add a swap file

The smallest VPS can run short of memory during installs or with a local model. Give it a 2GB swap file first:

bash
sudo fallocate -l 2G /swapfile
sudo chmod 600 /swapfile
sudo mkswap /swapfile
sudo swapon /swapfile

How do you install OpenClaw skills?

Install skills from ClawHub with openclaw plugins install. Skills are plugins that add abilities like web search, reminders or calendar access:

bash
openclaw plugins install @clawhub/web-search
openclaw plugins list
openclaw plugins uninstall @clawhub/web-search

For a curated set worth starting with, see the roundup of the best OpenClaw skills.

Useful commands for managing OpenClaw

These are the ones I actually use day to day. Keep them handy.

CommandWhat it does
openclaw doctorFull health check
openclaw doctor --fixAuto-repair common problems
openclaw gateway start/stop/restartControl the running service
openclaw gateway statusSee if it is running
journalctl -u openclaw -fWatch live logs
npm update -g openclawUpdate to the latest version
openclaw models set <model>Switch the LLM

To edit config by hand, the files live at ~/.openclaw/openclaw.json and ~/openclaw/.env.

Why is OpenClaw not working?

Nine times out of ten it is one of four things: a crash on startup, memory running out on a small plan, a dropped messaging channel, or the daemon step being skipped. These are the problems I see most, and the fix for each.

  • Crashes on startup: run openclaw doctor --fix, then check the logs with journalctl -u openclaw -f for the real error.
  • Out of memory: you are on a small plan without swap. Add the swap file from the callout above.
  • Messages not received: the channel dropped. Re-link it with openclaw channels and confirm the gateway is running.
  • OpenClaw stops after you close SSH: you skipped the daemon step. Run openclaw onboard --install-daemon so it runs as a service, not just in your session.

How do you secure an OpenClaw install?

Create a non-root user, turn on the firewall, log in with SSH keys instead of a password, and keep OpenClaw updated. You are exposing a server to the internet, so none of this is optional.

Start with the firewall and the config file's permissions:

bash
sudo ufw allow OpenSSH
sudo ufw enable
chmod 600 ~/openclaw/.env

Then keep it patched. Run npm update -g openclaw on a schedule, since updates carry security fixes as well as features (a remote-code-execution issue, CVE-2026-25253, was patched in an earlier release). Because OpenClaw is MIT-licensed and open-source, you can audit exactly what it does with your accounts.

OpenClaw alternatives worth knowing

OpenClaw is my default, but it is not the only self-hosted assistant. If you want to compare before committing, I put together the best OpenClaw alternatives, including options like Grip AI for heavier agent workflows.

Summing up

Installing OpenClaw on a Hostinger VPS is a fifteen-minute job once you know the order: spin up the VPS, install Node and OpenClaw, onboard your LLM, run it as a daemon, and connect a channel. The commands above are the whole recipe.

The payoff is a private AI assistant that is entirely yours, running on hardware you control, answering on the apps you already use. Get it running, secure it, then add the skills that make it genuinely useful.

Common questions

How much does it cost to run OpenClaw on Hostinger?

Just the VPS. A suitable Hostinger KVM plan runs from roughly $5-9 a month depending on the term, and OpenClaw itself is free and open-source. Your only other cost is your LLM usage, which is zero if you run a local model through Ollama.

Can I install OpenClaw on shared hosting?

No. OpenClaw needs to run as a background process 24/7 and requires SSH and root access, which shared hosting does not give you. A VPS is the minimum; that is why this guide uses a Hostinger KVM VPS rather than a shared plan.

Which LLM works best with OpenClaw?

Claude and GPT models give the best reasoning for a personal assistant. If privacy or cost matters more, run a local model with Ollama on the same VPS, though you will want more RAM (8GB or more) for that. You set this in the onboarding wizard.

How do I update OpenClaw to the latest version?

Run npm update -g openclaw, then restart the service with openclaw gateway restart. Run openclaw doctor afterwards to confirm everything is healthy. Updating regularly matters because it also pulls in security fixes.

Can I connect OpenClaw to WhatsApp?

Yes. WhatsApp is one of the 20-plus channels OpenClaw supports, alongside Telegram, Slack and Discord. You link it during or after onboarding by running openclaw channels and following the prompts to scan a QR code or add a token.

Is OpenClaw safe to run with my messaging accounts?

It is open-source and MIT-licensed, so the code is auditable, and self-hosting means your data stays on your server. The usual server hygiene applies: keep it updated, run it as a non-root user, lock down the firewall, and use SSH keys rather than passwords.

Written by
Sunny Kumar
Sunny KumarSEO Specialist & product builder

SEO Specialist and product builder with 10+ years in search. The notes come from the work, not the theory.