Browser automations with open law (help needed)

I bought a VPS that runs Linux without a GUI to try OpenAI out. But I ran into problems even after installing the skills I thought I needed. Is it because my VPS doesn’t have a GUI?
I also ran into issues where the agent told me it couldn’t perform certain automations because they were against the terms of service of certain sites. I didn’t have these same issues a week ago before the updates. Is there any way to fix this?
Is there a decent guide for deploying sub-agents? I’ve messed up my setup several times while trying to configure them.
I’ve also never tried connecting Chat for my account instead of using API keys. Is GPT Plus enough, or should I get Pro?
Using another Windows VPS isn’t a problem, and I also have some decent Windows devices I could run this on—I’d just need to wipe them clean first.
Thanks for any links or input on this.

If you’re doing browser automation for legit use cases, make sure you stay within the platform’s terms and focus on stable frameworks like Playwright or Selenium with proper rate limits and human-like timing. Most issues come from over-aggressive actions or bad fingerprint setups, not the automation itself.

Yeah, but it straight up told me, it can't even log in to the account, since it is due to that website tos. I did not load my framework to openclaw..
But I know that week ago, when I was trying this for the first time, it had no issues or what so ever with performing such tasks and much more.. but it wasn't able to make my automation work..