good catch. yes, local installs are still the best option if you want max security and control. the hostinger setup is mainly for people who want the fastest way to get openclaw running without dealing with install friction. once people understand the system, many move to a hardened local or hybrid setup.
After reading this post I worked with GPT to compare different VM hosts for the purposes of having an OpenClaw agent, and Hostinger actually came up as having a very good offer for the money. So I started a VM with Hostinger, and I used their Docker template to install OpenClaw. When I used GPT in Codex to check the security and setup of the installation, it found a couple of bad security settings baked into the Hostinger Docker template. It then basically cloned the agent into a separate container, minus the security flaws. The copy that was generated from the Hostinger template is essentially trash now. So much for one-click install :)
Getting the agent set up is actually pretty complex if you are not a super-techie. The good news is that in your Visual Studio Code, you can SSH into the server, then open Claude or GPT inside that connection. It can then basically do all the setup tasks and security hardening, etc., for you.
I am already extremely impressed with my agent, although I am still in the middle of getting him completely set up foundationally ,so that I can then move forward into building and running projects with him.
Isn't a local install recommend for security reasons using your hardening settings in another post?
good catch. yes, local installs are still the best option if you want max security and control. the hostinger setup is mainly for people who want the fastest way to get openclaw running without dealing with install friction. once people understand the system, many move to a hardened local or hybrid setup.
Awesome, thanks for the feedback
of course!
After reading this post I worked with GPT to compare different VM hosts for the purposes of having an OpenClaw agent, and Hostinger actually came up as having a very good offer for the money. So I started a VM with Hostinger, and I used their Docker template to install OpenClaw. When I used GPT in Codex to check the security and setup of the installation, it found a couple of bad security settings baked into the Hostinger Docker template. It then basically cloned the agent into a separate container, minus the security flaws. The copy that was generated from the Hostinger template is essentially trash now. So much for one-click install :)
Getting the agent set up is actually pretty complex if you are not a super-techie. The good news is that in your Visual Studio Code, you can SSH into the server, then open Claude or GPT inside that connection. It can then basically do all the setup tasks and security hardening, etc., for you.
I am already extremely impressed with my agent, although I am still in the middle of getting him completely set up foundationally ,so that I can then move forward into building and running projects with him.